Bulletin 69 Revised 2011

STATE OF FLORIDA
DEPARTMENT OF ENVIRONMENTAL PROTECTION
Herschel T. Vinyard Jr., Secretary

LAND AND RECREATION
Erma Slager, Acting Deputy Secretary

FLORIDA GEOLOGICAL SURVEY
Jonathan D. Arthur, State Geologist and Director

Bulletin No. 69 (Revised)

REGIONAL AND STATEWIDE TRENDS IN FLORIDA'S SPRING
AND WELL GROUNDWATER QUALITY (1991-2003)

By

Rick Copeland, Neal A. Doran, Aaron J. White, and Sam B. Upchurch

Published for the

FLORIDA GEOLOGICAL SURVEY
Tallahassee, Florida
2011

STATE OF FLORIDA
DEPARTMENT OF ENVIRONMENTAL PROTECTION
Herschel T. Vinyard Jr., Secretary

LAND AND RECREATION
Erma Slager, Acting Deputy Secretary

FLORIDA GEOLOGICAL SURVEY
Jonathan D. Arthur, State Geologist and Director

Bulletin No. 69 (Revised)

REGIONAL AND STATEWIDE TRENDS IN FLORIDA'S SPRING
AND WELL GROUNDWATER QUALITY (1991-2003)

By

Rick Copeland, Neal A. Doran, Aaron J. White, and Sam B. Upchurch

Published for the

FLORIDA GEOLOGICAL SURVEY
Tallahassee, Florida
2011

Printed for the
Florida Geological Survey

Tallahassee
2011

ISSN 0271-7832

ii

PREFACE

FLORIDA GEOLOGICAL SURVEY

Tallahassee, Florida
2011

The Florida Geological Survey (FGS), is publishing as its Bulletin No. 69 (Revised),
Regional and Statewide Trends in Florida's Spring and Well Groundwater Quality (1991-2003)
(Revised), authored by Rick Copeland, Neal A. Doran, Aaron J. White, and Sam B. Upchurch.

After the publication of the original publication, several minor errors were discovered.
Most errors were found in the statistical tables. This bulletin summarizes the results of a multi-
year cooperative investigation on spring and groundwater quality between the Florida
Department of Environmental Protection's FGS and the Bureau of Watershed Management,
Division of Environmental Assessment and Restoration. The data presented will be useful to
scientists, planners, and citizens in understanding the quality of Florida's groundwater resources.

Jonathan D. Arthur, Ph.D., P.G.
State Geologist and Director
Florida Geological Survey

TABLE OF CONTENTS

E x ecu tiv e su m m ary ................................................................................................................................................... x i
B background ........................................................................ .......... ..... .. xi
A approach .............................................................................................................. ...................... xii
Results and Conclusions .................. ................................................................................... xiii
Springs .................................. ......... .. .. ......... .... ... ................. xiii
W ells ............................................................................. ....................... xv
C o n cern s ................... ................................ ....... .................................................. ... .................. .... ....... ..... xv i
Rock-matrix and saline indicators: Saltwater encroachment............ ............................................ xvi
Nutrients ........................... ......... ..... ... .. ................. xviii
M monitoring ................................................... .... ....... .................... xviii
R ecom m endations .............................................................. .......................................................... ............ xix
Recommendation Synopsis........................................................... xxi
Research ........................... ..... ............ .. ................. xxi
M monitoring ........................................................ ... ... ................. xxi
Introduction ........................ .................. ..................... ..................... .. 1
A know ledgem ents ............................................................... ............................................................... . ............... 2
F lorida 's sp ring s ............................................................................................................ ............. ................ . 2
C classification of springs ......................................................... .......................................................... ........... 3
O offshore Springs ............................................. ................................................................... ............... . 5
Spring recharge basins........................... ................. .................. ........................ 6
Overview of the hydrogeology of Florida's groundwater ................................................................... ..................7...
Q quality of ground after and spring w after ................................................................... ........................................9...
Natural factors affecting groundwater and spring-water quality.......................................................................9...
D differences in spring- and w ell-w ater quality ................................................................................... ................. 10
Indicators of groundwater and spring-water quality problems........................................................................ 10
Spring selection process ..................... ...................................................................................... 12
W ell selection process ........................ ........................................................................................ 13
Methods ............................................................................... ....................... 16
D definition of trends.............................................................................................................................. . ........ .. 17
Problems with trends ................................................................................. 18
"Rem aining the sam e" Possibility of m issed trends............................................................... ................. 18
Outliers ............................................................................ ...................... 19
D etectio n lev els ............................................. ........................................................................... .......... .. 19
Sp arse d ata ...................................................................................................................................... . ......... .. 2 1
Analytes and indicators .................................................... ..........................................22
Sam ple collection and laboratory analyses...................................................... ............................................... 23
A nalytes used in this study ..................................................... ........................................... .................. 25
G rouping of analytes .......................................................... ............................................... ................ . 25
D description of analyte groups.................................. .......................................................... .................. 26
F field analytes ........................................... .................................................. ............. ............... . 26
R ock-m atrix analytes ....................................... .............................................................. ..................26
Saline or saltw after analytes.................................. .......................................................... ..................27
N utrient analytes .............................................................................................................. . 27
O th er an aly tes ................................................................................................................................. . ......... .. 2 7
Data ................................................................................... .................. 27
D ata sou rces................................................ ..................................................................... ............... . 2 8
D ata v verification ........................................... ................................................................... ............... . 28
D ata p rep aratio n ...................................................................................................................... ............... . 2 8
Tim e sequences................................... ................ .... ............................... ........................ 29
Data used for analyses and explanation of appendices................................................................. ................... 29
Inform action goals and data analysis protocols...................................................... ............................................... 30
Inform action introduction... ..............................................................................................30
O overview of statistical analyses procedures ..................................................................... .............................. 30

iv

D descriptive statistics .........................................................................................30
Kruskal-Wallis, Mann-Whitney, and Wilcoxon rank sum tests ...............................................................31
D eseasonalized data....................................... .............................................................. ............... . 33
M ann-K endall test ......................................................... .......................................................... . ...............34
Seasonal K endall test...................................... ............................................................. ................ . 34
S en slop e ............................................... ................................................................................. . ......... .. 34
Sign test ........................................................................................................................35
C aveats and assume options ................................... ........................................................... ..................35
R e su lts ....................................................................................................................................................................... 3 6
Springs ..................................................... .......... ............................. ........................ 36
N orthw est Florida W ater M anagem ent D istrict....................................................................... ................... 36
Rock and saline analytes, nutrients, and flow ..................................................................... ................... 36
Suw annee River W ater M anagem ent D district ............................................... ...................... .....................40
R ock-m atrix and saline analytes........................................................ .................................................. 41
F low ............................................................................................................................... 46
N utrient analytes......................................................... ...................................... 51
St. Johns River W ater M anagem ent D district .............................................. .......................... ................... 54
R ock-m atrix and saline analytes........................................................ .................................................. 55
N utrient analytes.... ............................................................................................59
South est Florida W after M anagem ent ..................................................... ................................................ 59
R ock-m atrix and saline analytes........................................................ .................................................. 59
F low ............................................................................................................................... 66
N utrient analytes....................................... .............................................................. ............... . 66
F field analy tes.......................................................................................................................... . ........ .. 67
W ells ................................................................................................................. ........................ 67
N orthw est Florida W ater M anagem ent D istrict....................................................................... ................... 71
W after levels and pH .................................................... ......................................71
Suw annee River W ater M anagem ent D district ............................................. ......................... ................... 74
W after lev els and pH ................................................................................... ....................................... 74
St. Johns R iver W ater M anagem ent D district .............................................. .......................... ................... 76
South est Florida W ater M anagem ent D istrict...................................................................... ................... 79
South Florida W after M anagem ent D istrict............................................. ........................... ................... 81
D istrictw ide sp ring trend s....................................................................................................................... ........... 8 1
Districtwide spring trends in the Suwannee River Water Management District ........................................86
Districtwide spring trends in the St. Johns River Water Management District ..........................................86
Districtwide spring trends in the Southwest Florida Water management District .......................................86
Statewide spring trends....... ........................................................................................ 100
Constrained version of state ide trends..................................................... ................................................ 101
Independence of springs and w ells.......................................................... ................................................... 102
D istrictw ide w ell-w after trends......................................................................................................................... 103
Evidence of districtwide well water-quality trends Northwest Florida Water Management District.......... 104
Evidence of districtwide well water-quality trends Suwannee River Water Management District.......... 108
Evidence of districtwide well water-quality trends St. Johns River Water Management District...........110
Evidence of districtwide well water-quality trends Southwest Florida Water Management District .........113
Evidence of districtwide well water-quality trends South Florida Water Management District...............117
State ide w ell w ater-quality trends.................................................................... .......................................... 120
Comparison of strong statewide trends for groundwater and spring water............................ .................... 120
C onstrained version of state ide trends ....................................................... ................................................ 121
D iscu ssion .. . . . . .............................................. ................ .............................. ........................................ 122
Major cause of statewide trends: Drought and consequential saltwater encroachment................................... 123
D rought.............................. ............ ................... ................ ............... ............... . 126
Rock-matrix and saline indicator evidence in spring-water quality................................................... 128
Rock-matrix and saline indicator evidence in well-water quality............ ......................................... 128
Regional to sub-regional evidence of saltwater encroachment............ .......................................... 130
Groundwater withdrawals............. ................................................................................. 133
G round after sum m ary ...................................................... .................................................................. 134

V

M miscellaneous and im portant issues ............................................................................. ................................ 134
Falling well-water levels a districtwide and statewide problem in wells....................... .................... 134
Nitrogen and phosphorous nutrients Regional and local problem in springs....................................... 137
Nitrogen in the Northwest Florida Water Management District............ .......................................... 137
Nitrogen in the Suwannee River Water Management District..................................................................139
Nitrogen in the St. Johns River and Southwest Florida Water Management Districts ............................... 148
M arion County.................... ..................................................................................... 148
Rainbow Springs Group....................................................................................... 148
Citrus County ................... .................................................................................... .. 149
K ing's B ay Springs G group ................................................................................................... 149
H om osassa Springs G group ................................................................................................... 149
C hassahow itzka Springs G roup...................................................... ................................................ 149
H ern an d o C ou nty ...................................................................................................................................... 14 9
W eeki W achee Springs G group ............................................................................................................. 150
Boat Springs, Bobhill, and Magnolia Springs............................................................ 150
H illsborough C county ..................................................................................... .............150
Lithia Spring and Buckhorn Spring............................................. ............................ ................. 150
Summ ary of the nitrate problem in spring water ............................................................... ................. 150
Phosphorus in spring water by water management district..................................................................... 150
Suw annee R iver W ater M anagem ent .................................................................................... ................. 151
St. Johns River and Southwest Florida Water Management Districts .......................... .....................161
Comparison of coastal to inland and tidal to non-tidal springs............ ......................................... 162
Global factors influencing Florida's groundwater................................................................ ................. 164
Global long-term cycles: Atlantic mutidecadal oscillation...................................................................... 164
Global short-term cycles: El Nifio and La Nifia ................................................................. ................. 166
A cid rain ................................................................................ ..................... ................. 168
Implications of future low rainfall and increasing state water demands............................... .................... 168
Implications regarding long-term sustainability ................... .................................................................. 169
R references .............................. ......................................................... ... .. ........................ 171
Appendices (Note: Appendices B2,C,E,H,I,J,K,L,and M may also be found at:
http://publicfiles.dep.statefl. us/FGS/FGS Publications/B/B69Appendices/ )
Appendix A Relationships among spring flows, rock, and salinity indicator concentration for
selected springs ........................................... ... ................................................ 179
Appendix B Glossary of terms and possible causes of trends ................. ............................183
A pp en d ix B l G lo ssary ............................................................................................................................... 18 3
Appendix B2 Interpretations of the origins of temporal trends in Florida's groundwater .......................191
Appendix C Well construction and location data ............................................................... ................. 196
Appendix D Spring locations.......................................................................................... 197
A appendix E Statistics....................................................................................................... 199
A ppendix E l Statistical m ethodologies............................................................................... ................. 199
Appendix E2 Macro codes for the Mann-Kendall tests and Sen slope......................... ......................206
Appendix F A nalytes ....... ........................................................................................209
A appendix F 1 A nalyte descriptions ....................................................... ................................................ 209
Appendix F2 A nalyte list w ith STORET codes...................................................................... .................. 216
Appendix G Data quality assurance (QA) officer contact information ....... ..................... ......................219
Appendix H D ata from springs and w ells.............................................. ............................ ................... 220
A appendix I D descriptive statistics ............................................................................ ................................... 221
A appendix J Seasonality results ................................ ......................................................................326
Appendix K Mann-Kendall and Sen slope results..................................................................... 346
A appendix L D istrictw ide m aps ........................................................................................................................ 347
Appendix LI Northwest Florida Water Management District springs ......................... ......................349
Appendix L2 Northwest Florida Water Management District wells ............................ ......................350
Appendix L3 Suwannee River Water Management District springs............................ ......................354
Appendix L4 Suwannee River Water Management District wells ....... ............................................359
Appendix L5 St. Johns River Water Management District springs....... ...................... ......................364
Appendix L6 St. Johns River Water Management District wells....... .............................................368
Appendix L7 Southwest Florida Water Management District springs ......................... ......................372

vi

Appendix L8 Southwest Florida Water Management District wells ...... ..................... ..................... 377
Appendix L9 South Florida Water Management District Wells..................................................... 382
A ppendix M Rainfall and tem perature data............................................................................. ................. 386
Appendix N Atmospheric deposition station information....................................................................... 393

Figures

Figure 1. Schematics of freshwater/saltwater transition zone and possible mechanism for saltwater
in tru sio n ...................................................................................................... . ................ x v ii
Figure 2. Inverse relationship of flow to rock and saline indicator concentrations .................................. xx
Figure 3. Locations of Florida's springs .......................................... ............................................. 4
F figure 4 O offshore springs ............................................................................................... 6
Figure 5. Median nitrate concentrations in 13 selected first-magnitude springs ..................................7...
Figure 6. Location of springs analyzed in this report ..................... ............... 14
Figure 7. Location of Temporal Variability Network (TVN) wells................................................... 15
Figure 8. Illustration of three options for water-quality trends........................................... ............... 18
Figure 9. Exam ple of a spurious trend .......................... ......... ................ .... ............... 20
Figure 10. Example of sporadic, unsystematic, and incomplete sampling ..........................................22
Figure 11. Monthly water temperatures plotted over the 1992 calendar year for an imaginary well ........31
Figure 12. Example illustration of seasonality with a six-year cycle.................................................. 32
Figure 13. Location of Wakulla Spring within the NWFWMD ......................................................... 37
Figure 14. Increasing rock analytes at W akulla Spring ....................................................... ............... 38
Figure 15. Decreasing nitrates and water levels at Wakulla Spring ................................................... 39
Figure 16. Location of springs within the SRWMD .................................................. 40
Figure 17. Increasing rock analytes at Fanning and Gilchrist Blue Springs....................................... 42
Figure 18. Increasing rock analytes at Suwannee Blue and Troy Springs.......................................... 43
Figure 19. Increasing rock analytes at Manatee and Hart Springs...................................................... 44
Figure 20. Increasing rock analytes at Poe and Lafayette Blue Springs............................................. 45
Figure 21. Decreasing flow at Fanning and Hart Springs.................................................... ............... 47
Figure 22. Decreasing flow at Rock Bluff and Hornsby Springs ........................................................48
Figure 23. Decreasing flow at Poe and Ruth/Little Sulfur Springs..................................................... 49
Figure 24. Decreasing flow at Troy and Telford Springs .................................................... ............... 50
Figure 25. D ecreasing nitrates at Poe Spring .................................................................... .................. 52
Figure 26. Increasing nutrient analytes at Poe and Lafayette Blue Springs........................................ 53
Figure 27. Location of springs within the SJRW M D .......................................................... ............... 54
Figure 28. Increasing rock analytes at Palm Springs ........................................................... ............... 56
Figure 29. Increasing rock analytes at Sanlando Springs .................................................... ............... 57
Figure 30. Increasing rock analytes at Wekiwa Spring.................................................. 58
Figure 31. Decreasing phosphate concentrations at Palm and Starbuck Springs................................. 60
Figure 32. Location of springs within the SW FW M D ......................................................... ............... 61
Figure 33. Increasing saline analytes at Rainbow and Bubbling Springs ........................................... 63
Figure 34. Increasing saline analytes for Hunter and Trotter Main Springs....................................... 64
Figure 35. Increasing saline analytes for Weeki Wachee and Bobhill Springs .................................. 65
Figure 36. Decreasing flow at Hom osassa No. 1 Spring ..................................................... ............... 67
Figure 37. Long-term flow trends at two SWFWMD springs ............................................................ 68
Figure 38. Increasing nitrates at Hunter and Magnolia Springs.......................................................... 69
Figure 39. Increasing nutrient analytes at Weeki Wachee and Boyette Springs................................. 70
Figure 40. Location of wells within the NW FW M D ........................................................... ............... 71
Figure 41. Decreasing pH and water levels in NWFWMD wells (#91 and #129) ................................ 72
Figure 42. Decreasing pH and water levels in NWFWMD wells (#131 and #312)................................ 73

Figure 43. Location of w ells w within the SRW M D .............................................................. ................ 74
Figure 44. Decreasing pH and water levels in SRWMD wells (#1943 and #2465) ............................... 75
Figure 45. Decreasing pH and water levels in SRWMD wells (#2585 and #2675) ............................... 76
Figure 46. Location of w ells w within the SJRW M D ................................................................ ................ 77
Figure 47. Temperature and specific conductance in SJRWMD wells (#1417 and #1763).................. 78
Figure 48. Temperature and specific conductance in SJRWMD well (#1762) .................................. 79
Figure 49. Location of wells within the SW FW M D ........................................................... ................ 80
Figure 50. Decreasing pH and water levels in SWFWMD wells (#996 and #1087)............................. 82
Figure 51. Decreasing pH and water levels in SWFWMD well (#707) ............................................. 83
Figure 52. Location of w ells w within the SFW M D ............................................................... ................ 84
Figure 53. Decreasing pH and water levels in SFWMD wells (#6490 and #3398).................................. 85
Figure 54. Weighted mean temperature and rainfall data in Florida (1991-2003)............................... 125
Figure 55. Relative position of rock-matrix and saline analytes in the Upper Floridan aquifer system. 126
Figure 56. Fresh groundwater wedge before and during a drought.................................................. 127
Figure 57. Water level and pH relationships for a well with a falling water table .............................. 129
Figure 58. Weighted mean annual rainfall during Sequence A .......... ....................................... 131
Figure 59. Exam ple of aliasing ................................................... ...... ................... ............... 136
Figure 60. Nitrate concentrations in Wakulla Spring between 1972 and 2005................................. 137
Figure 61. Flow adjustment for nitrate in Troy Spring ....... ........ ...... ..................... 140
Figure 62. Flow adjustment for TKN in Troy Spring ....... ........ ...... ..................... 141
Figure 63. Flow adjustment for nitrate in Homsby Spring .......................................................... 142
Figure 64. TKN versus time and log of TKN versus log of flow in Homsby Spring .......................... 143
Figure 65. Flow adjustment for nitrate in Fanning Spring....... .......................... 145
Figure 66. Flow adjustment for TKN in Fanning Spring............ ........................ 146
Figure 67. Flow adjustment for phosphorus in Troy Spring ....... ............ ....................................... 152
Figure 68. Flow adjustment for phosphate in Troy Spring ................................................ ............... 153
Figure 69. Flow adjustment for phosphorus in Ruth/Little Sulfur Springs.................. .................. 154
Figure 70. Flow adjustment for phosphate in Ruth/Little Sulfur Springs.................... .................. 155
Figure 71. Flow adjustment for phosphorus in Fanning Spring.......... ...................................... 156
Figure 72. Flow adjustment for phosphate in Fanning Spring ....... .......... ...................................... 157
Figure 73. Flow adjustment for phosphorus in Little River Spring ....... .................... .................. 158
Figure 74. Flow adjustment for phosphate in Little River Spring .............................................. 159
Figure 75. Flow adjustment for a phosphate in Homsby Spring .......... ..................................... 160
Figure 76. Atlantic Multidecadal Oscillation and Florida spring flow ........................ .................. 165
Figure 77. Sea surface temperature, La Nifia, and El Nifio.................................... .................. 167
Figure 78. Average monthly pH from seven atmospheric rain stations (1991-2003).......................... 169
Figure Al. Juniper Springs alkalinity versus flow.......................... ..................... 179
Figure A2. Wakulla Spring sodium versus stage....... ........... .......... ..................... 180
Figure A3. Alapaha River Rise Fanning sodium versus flow.......... ...................................... 180
Figure A4. Fanning Spring sodium versus flow ....... ........................ ..................... 181
Figure A 5. Poe Spring sodium versus flow ..................................... ......................... ............... 181
Figure A6. Homosassa No. 3 Spring sodium versus flow ............... ........................ 182
Figure A7. Chassahowitzka No. 1 Spring sodium versus flow.......... ..................................... 182

Tables
T able 1. Spring M magnitude ............................ ........................................................... ................ 3
Table 2. Florida's spring classification system ......................................... ........................................... 5
T able 3. A nalyte and indicator list...................................................... ............................................... 23
Table 4. Analytes and indicators displaying trends ........................................................... ................ 24
T able 5. A nalyte G roups.................................................................... .. ................ 26

Table 6. Exam ple of descriptive statistics table................................................. ............. ................ 31
Table 7. Suwannee River Water Management District spring names and abbreviations ..................... 41
Table 8. St. Johns River Water Management District spring names and abbreviations ....................... 55
Table 9. Southwest Florida Water Management District spring names and abbreviations .................. 62
Table 10. Spring trends in the SRWMD (plus Wakulla Spring), Sequence A (1991-2003).................. 87
Table 11. SRWMD (plus Wakulla Spring) districtwide trends based on sign tests, Sequence A ............ 87
Table 12. Spring trends in the SRWMD (plus Wakulla Spring), Sequence B (1991-1997).................. 88
Table 13. Spring trends in the SRWMD, (plus Wakulla Spring) Sequence C (1998-2003).................. 89
Table 14. SRWMD (plus Wakulla Spring) districtwide trends based on sign tests, Sequence C............. 89
Table 15. Spring trends in the SJRWMD, Sequence A (1991-2003) ..................................................90
Table 16. SJRWMD districtwide trends based on sign test, Sequence A........................................... 90
Table 17. Spring Trends in the SJRWMD, Sequence B (1991-1997) .................................................91
Table 18. Spring Trends in the SJRWMD, Sequence C (1998-2003) .................................................92
Table 19. SJRWMD districtwide trends based on sign tests, Sequence C.......................................... 92
Table 20. Spring Trends in the SWFWMD, Sequence A (1991-2003) .............................................. 93
Table 20. Spring Trends in the SWFWMD, Sequence A (1991-2003) (continued)................................ 94
Table 21. SWFWMD districtwide trends based on sign tests, Sequence A........................................ 94
Table 22. Spring Trends in the SWFWMD, Sequence B (1991-1997)............................................... 95
Table 22. Spring Trends in the SWFWMD, Sequence B (1991-1997) (continued) ................................. 96
Table 23. SWFWMD districtwide trends based on sign tests, Sequence B........................................ 96
Table 24. Spring Trends in the SWFWMD, Sequence C (1998-2003)............................................... 97
Table 24. Spring Trends in the SWFWMD, Sequence C (1998-2003) (continued) ................................ 98
Table 25. SWFWMD districtwide trends based on sign tests, Sequence C........................................ 98
Table 26. Spring flow from three stations in the SWFWMD ............................................................. 99
Table 27. Statewide spring trend summary by WMD and time sequence ........................ ................ 99
Table 28. Statewide trends based on sign tests for 57 springs, Sequence A (1991-2003)................... 100
Table 29. Statewide trends based on sign tests for 57 springs, Sequence B (1991-1997) ................... 100
Table 30. Statewide trends based on sign tests for 57 springs, Sequence C (1998-2003) ................... 101
Table 31. Statewide trends in at least two WMDs, Sequence A (1991-2003).................................. 101
Table 32. Statewide trends in at least two WMDs, Sequence C (1998-2003) .................................. 102
Table 33. Selected statewide analyte results, Sequence A (1991-2003)...................... .................. 103
Table 34. Well trends in the NWFWMD, Sequence A (1991-2003)........................... .................. 105
Table 35. Potential NWFWMD districtwide trends, Sequence A ............... ................... 105
Table 36. Well trends in the NWFWMD, Sequence B (1991-1997)........................... .................. 106
Table 37. Potential NWFWMD districtwide trends, Sequence B....... ........................................ 106
Table 38. Well trends in the NWFWMD, Sequence C (1998-2003)........................... .................. 107
Table 39. Potential NWFWMD districtwide trends, Sequence C....... ........................................ 107
Table 40. Well trends in the SRWMD, Sequence A (1991-2003)....... ........................................ 108
Table 41. Potential SRWMD districtwide trends, Sequence A .......... ....................................... 108
Table 42. Well trends in the SRWMD, Sequence B (1991-1997) ....... ....................................... 109
Table 43. Potential SRWMD districtwide trends, Sequence B....... .... ..................................... 109
Table 44. Well trends in the SRWMD, Sequence C (1998-2003) ....... ....................................... 110
Table 45. Potential SRWMD districtwide trends, Sequence C....... .... ..................................... 110
Table 46. Well trends in the SJRWMD, Sequence A (1991-2003) ..................................................... 111
Table 47. Potential SJRWMD districtwide trends, Sequence A.......................................................... 111
Table 48. Well trends in the SJRWMD, Sequence B (1991-1997) .............................. 112
Table 49. Potential SJRWMD districtwide trends, Sequence B ....... ......................................... 112
Table 50. Well trends in the SJRWMD, Sequence C (1998-2003) .............................. 113
Table 51. Potential SJRWMD districtwide trends, Sequence C ....... ......................................... 113
Table 52. Well trends in the SWFWMD, Sequence A (1991-2003)........................... .................. 114
Table 53. Potential SWFWMD districtwide trends, Sequence A ....... ........................................ 114

Table 54. Well trends in the SWFWMD, Sequence B (1991-1997)....... .................... .................. 115
Table 55. Potential SWFWMD districtwide trends, Sequence B .......... ..................................... 115
Table 56. Well trends in the SWFWMD, Sequence C (1998-2003)....... .................... .................. 116
Table 57. Potential SWFWMD districtwide trends, Sequence C .......... ..................................... 116
Table 58. Well trends in the SFWMD, Sequence A (1991-2003) .............................................. 117
Table 59. Well trends in the SFWMD, Sequence B (1991-1997) .............................................. 118
Table 60. Well trends in the SFWMD, Sequence C (1998-2003) .............................................. 119
Table 61. Potential SFWMD districtwide trends, Sequence C ....... .... ...................................... 119
Table 62. Statewide trends based on sign tests: Sequences A, B, and C ..................... .................. 120
Table 63. Statewide spring-water quality summary for rock and saline indicators ............................. 121
Table 64. Statewide trends for at least three WMDs: Sequences A, B, and C for combined
ground ater resources ............................................................. ....... ................ ............... 122
Table 65. Constrained statewide spring and groundwater-quality summary ................................... 123
Table 66. Summarized annual weather data in Florida (1991-2003)............................ ............... 124
Table 67. Annual weather data, four WMDs compared to the SWFWD (1991-2003)........................ 131
Table 68. Summarized rainfall, Sequence C minus Sequence B ...................................................... 132
Table 69. Relationships among concentration and loading of nitrate and TKN versus
time and flow in selected springs in the SRWMD............................................. 147
Table 70. Inland and Coastal Springs within the SRWMD and the SWFWMD .................................. 162
Table 71. Comparison of coastal and inland springs for upward trends during Sequence C,
excluding springs w within the SJRW M D ............................................................. ............... 163
Table 72. Comparison of tidal and non-tidal springs for upward trends during Sequence C
in the SW FW M D ............... ................... .. .......... .................................... ... 164

EXECUTIVE SUMMARY

Background

Over the past several decades, it has been observed that the flows in Florida's springs are
declining and water quality is degrading. The primary chemical concern is considered to be
increased nutrients, including soluble forms of nitrogen and phosphorus. The sources are
predominantly from animal waste, human waste, and from the synthetic fertilizers used on lawns,
golf courses, or for agricultural activities.

In recognition of these issues, the Secretary of the Florida Department of Environmental
Protection (FDEP) directed the formation of the Florida Springs Task Force in 1999. The multi-
agency task force consisted of 16 scientists, planners, and citizens who were concerned about the
"environmental health" of Florida's springs. By 2000 the task force made a series of
recommendations to protect and restore Florida's springs. They are outlined in detail in Florida's
Springs: Strategies for Protection and Restoration (Florida Springs Task Force, 2000). Two of
the recommendations were to:

(1) Implement springs monitoring programs to detect and document long-term
trends in water quantity and quality

(2) Conduct research that will allow cause-and-effect relationships to be
established between land use and water management activities.

The purpose of monitoring is to both support research efforts and to confirm the effectiveness of
spring protection efforts. As a direct result of the first recommendation, the Florida Geological
Survey (FGS) took the lead in implementing a spring monitoring program. By 2004 it published
the latest Springs of Florida bulletin-a descriptive overview of Florida's springs.

The main purposes of this document are to: (1) determine trends in groundwater where
sufficient data are available; (2) establish prototype methods for evaluating and reporting trends
for future applications; and (3) enhance the efforts of determining cause-and-effect relationships
between anthropogenic activities and the resulting spring-water quality and quantity on regional
(water management district-wide) and statewide scales. The reason for the latter is that many
other publications have addressed the causes of trends on an individual spring basis. If we
attempted to develop an exhaustive list of possible causes of trends for each spring, it could take
many years to accomplish. We decided to emphasize regional and statewide scales. An endeavor
of this nature has never been attempted. If regional or statewide trends were found, the causes
and possible solutions to those causes may become the highest priority water management issues.

In order to fully comprehend the implications of trends in springs, a thorough
understanding of the behavior of groundwater in wells is also necessary. In 1983, the FDEP
began a statewide groundwater quality monitoring network (Florida Statutes 403.063). Scott et
al. (1991) stated that the purpose of the network was to detect or predict contamination of
Florida's groundwater resources. Currently, several thousand wells are included in the network.
However, a subset of the wells are conducive to trend analysis (the Temporal Variability

Network, or simply the TV Network). Since the FGS was asked to evaluate data for trends from
springs, it followed to simultaneously do the same for groundwater from TV Network wells.

Approach

The FGS spring monitoring program commenced operations in 2001. For many of the
springs, previous samples had never been collected, so long-term trend analyses were not
possible. However, the FGS contacted each of the four northern water management districts
(WMDs), the U.S. Geological Survey (USGS), and programs within the FDEP to request copies
of their historical spring-water quality and quantity data. The entities each graciously delivered
data to the FGS for analyses. It should be noted that the South Florida Water Management
District (SFWMD) had insufficient data for trend analyses.

The FGS obtained sufficient data, meeting preset criteria, from 58 springs and 46 wells
for the period January 1991 through December 2003. For reference, the study was divided into
three time sequences. Sequence A represents the entire length of the study (1991-2003).
Sequence B represents the January 1, 1991 December 31, 1997 time frame, while Sequence C
represents January 1, 1998 December 31, 2003. The two shorter sequences were used to assist
in identifying and evaluating shorter-term trends. As it turns out, Sequence B coincided with
relatively normal rainfall, whereas Sequence C covered a time that Florida experienced an
extended drought.

The analytes (constituents of interest) for this report can be broken down into five
groups: (1) nutrients, (2) saline (or salt-water), (3) rock-matrix (or rock), (4) field, and (5) other.
Of these, the three major groups are nutrients, saline, and rock-matrix. Nutrients are compounds
that are essential for the growth of living organisms. Unfortunately, high concentrations in
spring-water can adversely affect the biota in spring runs. Saline analytes are related to salts.
The most significant sources of salt are from the ocean or deep groundwater in Florida's
aquifers. High concentrations of saline compounds (e.g. sodium or chloride) can restrict the
usage of water. Rock-matrix analytes have their sources in the aquifer material (e.g. limestones
and dolostones). They occur naturally and, unless they occur in extremely high concentrations,
are generally not harmful to our environment. The field and other analyte groups consist of
miscellaneous constituents that are useful in explaining trends in other analytes.

Think of a trend as a direction of movement (Berube and Boyer, 1985). Although there
are many secondary questions that pertain to trends, trend analysis can be broken down to one
fundamental, primary question, "over time, are conditions changing (getting better or getting
worse), or are they remaining the same?" One can think in terms of concentrations of water
quality (measured by analytes) or water quantity (measured by flow or water levels). Subjective
descriptions-such as better, worse, and remaining the same-are based on objective changes
over time, or trends.

Throughout this report the term significant refers to statistical significance. Some of the
trends reflect very important changes in water quality, whereas some only represent relatively
minor changes in water quality that are not indicative of impending problems. If, during our
analyses, a trend was discovered, it was based on statistical significance. That is, within a

predefined probability, we do not expect the trend to occur randomly. Since not all of the
audience of this report is familiar with the statistical procedures employed, we decided to
simplify the procedures to the extent practical. For this reason, most analyses were restricted to
linear trends, using nonparametric techniques. Data were checked for seasonality, and if found,
were deseasonalized prior to trend analyses using a method recommended by the U.S
Environmental Protection Agency (EPA, 1989). Trend analyses were conducted using the
Mann-Kendall test, while the rates of change over time were determined using the Sen slope
(Gilbert, 1987).

Results and Conclusions

The most important conclusion to be derived from this report is that Florida springs truly
represent the "canary in the coal mine" with respect to assessing regional groundwater quality in
Florida. As will be summarized, springs are apparently much better at indicating over-all change
in a groundwater flow system than wells.

Monitoring wells only allow for the sampling of a discrete portion of the water in an
aquifer. They limit detection to a particular depth interval and a relatively limited spatial extent.
Karstic aquifers are especially limited in this respect as fractures and cavernous conduits may
direct water flow around or below the location of a monitoring well. On the other hand, the
quality of water discharging from a spring is an integral of the water quality of the total flow
system within a springshed. This water is derived from deep and shallow flow systems and
conduit and diffuse flow. Furthermore, the water quality, and flow, data are weighted according
to the relative importance of the flow systems and chemical sources within the springshed. As a
result, springs appear to be much better at detecting regional changes in a springshed water
quality than do wells. This conclusion is supported by the fact that water-quality trends were
much more obvious in spring data than in well data.

Springs

Of the analyte groups, rock-matrix and saline analytes had the greatest frequency of
trends. Both analyte groups showed strong negative correlations with spring flow. For example,
as spring flow decreased saline and rock-matrix analyte concentrations increased. The
relationship was observed throughout the state. The greatest increases in the concentrations of
rock-matrix and saline analytes occurred during a drought that occurred between late 1998 and
mid 2002.

There are several probable explanations and all can be a result of the drought. First,
during the drought there was less rainfall, and consequently there was less surface-water flow.
In karst terrains, much surface-water flows directly to groundwater through sinking streams
(swallets). Typically, this rapidly recharged groundwater is transmitted in well-developed
subsurface conduits. Thus, there is very little contact time with the aquifer matrix before it
discharges from springs, and it tends to have lower concentrations of rock-matrix analytes.
During a drought, there is a decrease in the proportion of freshly recharged "surface water." This,
at least partially explains the correlation between decreased spring discharge and increased
concentrations of rock analytes.

A second probable explanation is related to the removal of older, saline-rich and usually
more mineralized water from storage, often in the deeper portions of the Florida's aquifer
systems. Beneath the state of Florida lies a "lens" of fresh water, which is replenished by
rainfall. Freshly recharged water is flushed through Florida's karstic (sinkholes, caves, springs
etc.) aquifers relatively quickly to springs. In contrast, the deeper water is older (Upchurch,
1992; and Katz, 2004). Because it has been in contact with the aquifer matrix for a relatively
long period of time, the aquifer water has had a longer time to "pick up" dissolved matrix
material constituents such as calcium and magnesium, especially in the Floridan aquifer system.
With longer residence times, the older water tends to have higher concentrations of rock-matrix
material.

A third explanation is similar to the second. Older, mineralized residual saltwater, was
never fully flushed from the rock interstices in some portions of Florida (Johnson and Bush,
1986). With less rainfall during the drought, the water levels in the aquifers were lowered, and
the size of the freshwater "lens" decreased. With decreasing freshwater potentials (e.g., water
levels) the deeper and older connate water can find its way upward toward aquifer discharge
points, such as springs. Thus, during the drought, increased concentrations in rock-matrix and
saline analytes were observed, along with decreases in spring discharges. The trends were
statewide in scale. The magnitude of scale was the most surprising and most significant finding
of the study.

After the driest portion of the drought (2002), Florida's hydrologic conditions began to
recover and the concentrations of both types of analytes began to decrease, as rainfall, recharge,
and spring flow began to increase. The inverse relationship between spring-water discharge, and
both rock-matrix and saline analyte concentrations, was also observed in a study by Katz (2004).
In addition, Katz also found a positive correlation between concentrations of rock and saline
analytes and spring-water age.

Nutrients in groundwater discharging from springs were one of the most important
concerns of the Springs Task Force. Evaluation of trends in this report revealed that nutrient
trends in springs had an uneven, or patchwork, distribution across the state. That is, both
increasing and decreasing nutrient trends were common and were observed throughout Florida.
This suggests that the trends were often related to local land-use and water-use activities. As
such, most nutrient concentrations observed in springs are localized and should be analyzed in
relation to the corresponding springshed.

Nitrogen and phosphorus comprised the most frequent nutrient exhibiting trends.
Nitrogen in the form of nitrate (nitrate plus nitrite as N) had the greatest frequency of increasing
(degrading) trends. However, some springs actually had decreasing nitrate trends. Phosphorus,
as total phosphorus and orthophosphate, had both increasing and decreasing trends, depending on
the springshed.

Note that decreasing nutrient trends are not necessarily good news. During the drought,
an important observation was that some nitrate concentrations had positive correlations with
spring flow. One possible explanation is that nitrogen can be stored in the soils of Florida's
springsheds (Bruland et al., 2008). During the drought, soils may have stored the nitrogen

originating from fertilizer applications and the nitrogen did not find its way to the groundwater
regime. When rainfall conditions return to normal, the soils will release the nitrogen and
concentrations in spring water will eventually increase. On a similar note, decreases in
phosphorus in some areas may likewise not be a reflection of improved management. It is
possible that the upward migration of older water, with different chemistry, reduced the
phosphorus concentrations in many springs. If so, reduction of phosphorus could simply be a by-
product of mixing with deeper, higher pH water-not an improvement in water quality. This
mechanism is discussed by Hem (1985) and by Odum (1953). They indicated that the solubility
of phosphorous can be controlled by pH. Dissolved phosphorous is generally more abundant in
lower pH (more acidic) water. Conversely, higher pH (more basic) water contributes to the
precipitation of phosphate and lowers the concentration of dissolved phosphorous in
groundwater.

Wells

Within Florida's aquifers, the flow paths of spring-water can potentially be from both
deep and shallow sources. Conversely, wells typically are drilled to a specific depth in an
aquifer. Consequently, flow paths of well water are from a much narrower thickness of the
aquifer, relative to spring water flow paths. Although there are exceptions, most of the 46 wells
used in this study generally tap only the shallower portions of the aquifers. The wells tend to be
less than 30 m (100 feet) deep. Because of the shallower depth, the older, deeper, and more
mineralized deeper aquifer water had a lower probability of being observed in the shallow wells.
Thus, rock-matrix and saline trends were not seen as frequently in wells as in springs.
Nevertheless, decreasing trends in water levels within wells were common. In addition, pH-a
field analyte-had a positive correlation with water levels; as water levels in wells decreased, so
did pH.

A possible explanation for this positive correlation is as follows. Well intake zones for
most wells in Florida are generally set at specified depths below the lowest predicted aquifer
water levels. This is done in order to guarantee water to the well during drought conditions.
During dry times the upper surface of the saturated zone is lowered downward toward the
uppermost point of the intake zone. For the aquifers tapped by the 46 shallow wells used in this
study, most recharge is from water, typically rainfall, penetrating the land surface and moving
downward through the soil to the groundwater regime. Rainfall has a lower pH than most
aquifer water. The pH is lowered further as rainwater picks up carbonic acid as it moves
downward through Florida's soils (Freeze and Cherry, 1979; and Upchurch, 1992). Therefore,
as the water table (or the potentiometric surface in confined aquifers) drops, generally the
younger, freshly recharged water with lower pH has an increasing probability of entering well
intake zones. As such, the lowering of the water table is a potential cause for decreasing trends
in pH values across the state during the drought. A detailed description of this hypothesis, along
with other related hypotheses, is discussed in the body of this report.

Another field analyte that displayed a trend was well water temperature. Between 1991
and 2003, its temperature typically increased; the reason is believed to be an increase in air
temperature. Air temperature increased across Florida (Southeast Regional Climate Center,
2006). Since the wells used in this report tend to be shallow, it is believed that well water readily

responded to air temperature changes. On the other hand, the sources of spring water are from
shallow and deep portions of our aquifers. Deeper water tends not to respond to changes in air
temperature. Thus, spring water displayed fewer temperature trends than did well water.

Concerns

Rock-Matrix and Saline Indicators: Saltwater Encroachment

Saltwater encroachment is the displacement of fresh groundwater by the advance of
saltwater due to its greater density (Neuendorf et al., 2005). It can occur during a drought when
recharge declines and the freshwater "lens" shrinks in size. Over geologic time, it can occur with
sea-level rise. It can also occur when excessive groundwater pumping causes the advancement
of saltwater. Freeze and Cherry (1979) use the term saltwater intrusion as the migration of
saltwater into freshwater aquifers under the influence of groundwater development (pumping).
For this paper, we use the term intrusion to indicate a man-induced process and use the term
encroachment to make no distinction between natural and man-made causes.

Figure 1 (top) displays the unconfined, surficial aquifer system. The saltwater/freshwater
interface is represented by a transition zone. During a drought, the water table lowers, the
transition zone migrates inland and the thickness of the freshwater zone ("lens") decreases in
size.

In his work in northeastern Florida, Spechler (2001) mentioned several possible
mechanisms that can drive encroachment and intrusion. During the drought, they included: (1)
the movement of "un-flushed" pockets of relict seawater within the Floridan aquifer system, (2)
the landward movement of the freshwater/saltwater interface, (3) regional upcoming of saltwater
below pumped wells, and (4) the upward leakage of saltwater from deeper, saline water-bearing
zones through confining units. The latter can occur where the units are thin or are breached by
joints, fractures, collapse features, or other structural anomalies. Examples are displayed in
Figure 1 (bottom).

During the 1999-2002 drought, the flows in many springs decreased, and one spring
(Hornsby Spring) stopped flowing altogether for a period of time. In addition to the decreased
rainfall, there was an increased demand for groundwater (Verdi et al., 2006). The drought and
the subsequent lowering of aquifer water levels resulted in decreasing spring flows throughout
the state. The increased demand for groundwater during the drought exacerbated the problem in
some of the springs. The increasing trends in rock-matrix analytes during the drought is an
indication of a reduction in size of the fresh water "lens" underlying the state and an indication
of saltwater encroachment. Because the concentrations of saline analytes increased almost
everywhere in the state during the drought, it is an indication that encroachment occurred on a
statewide scale.

surficial aquifer system

Freshwater

&W--;- v. * I?
... o ",'....;'.' '.;y .;.i. '+ ...^-.'* w :- A-?3-; .:...' ::! ..; .:
.... .... ...... ...

_ .. .. ",. ;' .;" '
..".:.* ..'...........'

:. -... ".'.'-.. "'... "intermediate confining unit ." : ": *

Not to scale

Modified from Cooper (1964)

UNDIFFERENTIATED CONFINING UNrIT

Not to scale Modified from Spechler (2001)

EXPLANATION

Brackish water
- Saltwater
I Freshwater

S Direction of ground-water flow

Figure 1. Schematics of freshwater/saltwater transition zone and possible mechanisms
for saltwater/freshwater intrusion. Note Cooper (top) represents the saltwater/freshwater in-
terface in the surficial aquifer system as a transition zone, whereas Spechler (bottom) depicts it
as a sharp boundary.

xvii

0 .. .
..*.*.:.';..& *

The 1998-2002 drought was one of the worst historical droughts to affect Florida (Verdi
et al, 2006). Except for south Florida, during the drought the deficit rainfall ranged from about
10 inches in southwest Florida to almost 40 inches in northwest Florida. In order to make up for
the drought, groundwater pumping increased, largely for irrigation (Verdi et al., 2006). Because
an increase in groundwater pumping occurred during one of worst droughts, it is likely that
human-induced saline intrusion took place and contributed to the increase in saline and rock-
matrix analyte trends. On a statewide scale, the extent and severity of the intrusion is difficult to
quantify. However, within the northern portion of the SWFWMD, a water budget and a regional
groundwater flow model indicated that the increase [0.3 cm/yr (+0.1 in/yr)] in groundwater
withdrawals was less than 2.0% of the decline in recharge due to the decrease [18.3 cm/yr (7.2
in/yr)] in rainfall (Ron Basso, Southwest Florida Water Management District, personal
communications). Nevertheless, intrusion should be a concern. If another drought of this
magnitude occurs, depending on the amount of increased pumping, it could potentially have
adverse effects on the long-term sustainability of Florida's groundwater resources.

Nutrients

The Florida Springs Task Force (2000) indicated that Florida's springs face serious
threats due to rapid and continuing population growth. The state's increasing population has
resulted in extensive land-use changes, increased demand for freshwater, and an increased use of
fertilizers. As rainfall seeps through the soils, and moves the nutrients into Florida's underlying
aquifers, it creates localized degradation in Florida's groundwater resources. A report regarding
FDEP's Springs Initiative Program efforts (Florida Department of Environmental Protection, and
Florida Department of Community Affairs, 2002) noted that nitrates have increased since the
1970s. It also noted that over the past 30 years many of Florida's springs experienced an
increase in nuisance algae and invasive exotic aquatic plants. These plants tend to thrive on
excess nutrients and decrease dissolved oxygen levels in spring runs.

Analyses for the 1991-2003 time frame indicated that trends in nutrient concentrations in
Florida's spring-water increased in some springs, while they decreased in others. It is
encouraging to note that there are some decreasing trends. The fact that nutrients (especially
nitrate) tended to increase is an indication that some land-use management practices warrants
reevaluation. But as noted previously, the relationship of these apparent decreasing trends may
be related to diminishing spring flow.

Monitoring

The current study revealed an inverse relationship between rock and saline indicators and
spring flow. The relationship was observed across the state (Figure 2). Note that changes in
spring-water quality often lag behind changes in spring flow. For detail, the smaller charts
depicted in Figure 2 have been enlarged and can be found in Appendix A.

Historically, the WMDs and the USGS have monitored spring-water quality and
discharge. With the commencement of the Springs Initiative, FDEP joined in the monitoring
efforts. Considerable efforts were made to eliminate inconsistencies in monitoring activities.
Unfortunately, at the beginning of the study, the efforts were not always successful. Specifically,
the WMDs, USGS, and FDEP did not always monitor the same analytes, use the same laboratory

xviii

analytical methods, or collect flow data on the same date as chemical and biological data were
collected. In addition, they often sampled at different frequencies. All these inconsistencies
made statewide comparisons very difficult. The results of this investigation demonstrate that
statewide monitoring must continue. For this reason, it is hoped that, in the future, the state can
find ways to minimize monitoring inconsistencies.

Recommendations

One of the most surprising and most significant observations of this study was that rock-
matrix and saline analytes were increasing almost everywhere in Florida's springs, especially
during the drought of Sequence C (1998-2003). Saltwater encroachment is a hugely significant
issue. Saltwater can restrict water use and negatively affect freshwater ecology, and can
adversely affect the long-term term sustainability of Florida's water resources. The relationships
among rainfall, recharge, groundwater withdrawals, groundwater quality and levels, plus spring-
water flows warrant further research, as does the effects of global climate change.

The concentrations of at least one nutrient (nitrate) in numerous springs have been
excessively increasing since the 1970s (Florida Department of Environmental Protection and
Florida Department of Community Affairs, 2002). One of the most visible changes in spring-
water quality has been the increase in nuisance algae and invasive exotic aquatic plants. What is
the relationship between the increases in nutrients and the nuisance plants? Further research is
needed. In addition, land-use management practice modifications are needed in order to reverse
the increasing trends. It is beyond the scope of this study to elaborate on the management
strategies. For a detailed discussion of many of the available strategies, an excellent reference is:
Protecting Florida's Springs Land Use Planning Strategies and Best Management Practices
(Florida Department of Environmental Protection and Florida Department of Community
Affairs, 2002).

Spring-water quality is sensitive to changes in spring flow and to aquifer water levels.
Springs represent excellent natural sampling locations for monitoring saline encroachment. It is
recommended that, to the extent practical, springs should be incorporated into a statewide
saltwater encroachment monitoring network. The results of the spring monitoring could then
potentially be used to supplement well monitoring networks that are often used for saltwater
encroachment purposes.

Although the monitoring of springs and wells is critical for the sustainability of Florida's
water resources, not all analytes of concern are sampled. Synthetic organic, other supplementary
analytes supplementalss), as well as biological indicators, should be included on the monitoring
lists. It should be understood that supplementals are expensive to collect and analyze, and for
these reasons, they can only be sampled on a low frequency basis. It should also be noted that
supplemental monitoring is often determined by site-specific issues. For example, pesticides
may only be detected at certain times of the year or in certain locales, determined by land use
conditions. Supplementals such as pesticides, synthetic organic compounds, and trace metals
should occasionally be sampled.

Figure 2. Inverse relationship of flow to rock and salinity indicator concentrations. Darker
lines represent water levels (whether by stage or spring flow); lighter lines represent saline or
rock-matrix indicators (sodium or alkalinity). Time axes vary. The graphs indicate reciprocity
between decreases in water levels and increases of salinity, regardless of location in the state.
Florida's spring-water chemistry shows a high sensitivity to changes in flow (See Appendix A
for enlarged versions of inset charts).

XX

It is critical that evaluations of spring water and groundwater be clearly disseminated to
the public as efficiently as practical. One efficient method is the use of indices. Stock exchange
indices have been used in the financial community for many years. Groundwater quantity
indices are used by the Edwards Aquifer Authority in Texas. As an example, the authority use
real-time water levels in the Bexar County Index well as an index (indicator) for the entire
county. During dry times, as water levels fall, water restriction measures may be invoked by the
authority. When water levels rise, the restrictions are lifted (Edwards Aquifer Authority, 2006).
There are several potential indices that could be developed for use in Florida. If one or more
indices were developed, they have the potential to become very useful in informing the public
about the status of our springs. However, in order to be viable, buy-in by both the public and
scientific communities are essential. Hopefully, indices will be adopted in the future.

It is essential that technical reports regarding the results of analyses be generated
frequently and in a relatively short time frame. It is acknowledged that it takes a considerable
amount of time for an initial report to be generated. However, after the initial report, the lag time
between sample collection and report generation should reduce considerably. In addition,
subsequent reports using similar interpretative methods could employ computer programs to
create "boiler plate" reports as quickly as analytical data are received from a laboratory.

Standardized spring and well sampling throughout the state is a critical need. If
standardization is achieved, analyses of trends in the future will be much easier to conduct. This
in turn will make the resulting interpretations more comprehensive, and the dissemination of the
interpretive results will be more meaningful to the public. Specific aspects of the standardization
effort include: core and supplemental water-quality analytes and indicators, data reporting,
sampling and laboratory quality assurance, data management, data analysis, and assessment
reporting,

Recommendation Synopsis

Research

Determine the relationships between increases in nutrients and nuisance plants/algae

Determine the best land-use management practice needed in order to reverse increasing
nutrient trends

Improve our understanding of the relationships among: (1) rainfall, (2) recharge, (3)
groundwater withdrawals, (4) groundwater quality and levels, and (5) spring-water
quality and discharge

Develop a "spring environmental health" report card.

Monitoring

Recognize the importance of springs in saltwater encroachment monitoring and
incorporate spring monitoring into that effort

* Add supplemental analytes to spring monitoring lists on a periodic basis

* Develop spring water-quality and quantity interpretative reports on a regular basis

* Adopt area-wide randomized spring sampling on a periodic basis in order to produce a
synoptic report of all springs in Florida

* Continue to use the Florida Water Resource Monitoring Council to increase monitoring
efficiency. Topics for discussion should include:

core and supplemental water quality analytes and indicators
possible development of a "spring environmental health" index
possible implementation of the random sampling of springs
sampling and laboratory quality assurance
data management
data analysis
assessment reporting

xxii

BULLETIN NO. 69

REGIONAL AND STATEWIDE TRENDS IN FLORIDA'S SPRING
AND WELL GROUNDWATER QUALITY (1991-2003)
by
Rick Copeland (PG #126), Neal A. Doran,
Aaron J. White, Sam B. Upchurch

INTRODUCTION

Florida is blessed with some of the most spectacular springs in the world. There are
estimated to be over 700 springs in the state. People have been attracted to our springs since
before Florida became a state. From a scientific perspective, some of Florida's springs have
been sampled for over a century. The FGS published its first Springs of Florida bulletin
(Ferguson et al., 1947), which documented the chemical and flow data of the major springs. The
bulletin was revised in 1977 (Rosenau et al., 1977) and a new bulletin was generated in 2004
(Scott et al., 2004). In each revision, additional chemical data were presented. Unfortunately, as
beautiful as the springs are, not all is well. As Florida's population continues to grow, water-use
and land-use changes are reflected in our spring water. The quantity and quality of spring water
are both changing, and at least some of the changes are directly related to human activities.

Since the 1940s Florida's population has grown from about two million to about 18
million in 2000. This means that Florida has increased its population by a rate of about 600
people per day for those 60 years. In fact, between the years 2000 and 2005, the net rate of
increase has been over 700 people per day (U.S. Census Bureau, 2006). In the year 2000,
Floridians withdrew 3.14 billion gallons of groundwater daily (Marella and Berndt, 2005).
Marella and Berndt (2005) indicated that agriculture and public supply accounted for over 82
percent oft he groundwater use. Based on these data, each person used over 150 gallons per day
of groundwater. It is not surprising that an extensive increase in water use has followed
Florida's population growth. Neither is it surprising that there has been a noticeable decline in
the discharge of many of Florida's springs and that the intensive land-use changes have been
followed by a noted deteriorations in spring-water quality. Scott et al. (2004) mentioned that one
of the most notable deteriorations has been the increase in nutrient concentrations in spring
water. While nutrients such as nitrogen and phosphorous are required by aquatic organisms for
growth and reproduction, when the concentrations are found to be higher than natural levels,
problems can arise. Since the 1970s, concentrations of nitrate, a soluble form of nitrogen, have
been found to be increasing in a number of Florida springs (Florida Springs Task Force, 2000).

Over the past several decades, flows in Florida's springs are declining and water quality
is degrading. The primary chemicals of concern are nutrients, including soluble forms of
nitrogen and phosphorus.

In order to improve and protect our springs, the Florida Springs Task Force (2000) made
a series of recommendations to the Governor of Florida. One was that Florida should implement
spring monitoring programs in order to detect and document long-term trends in water quality.
In addition, it was recommended that the state should conduct research in order to determine the

FLORIDA GEOLOGICAL SURVEY

cause-and-effect relationship between land-use and water-management activities, and the
resulting changes in spring-water quality and quantity.

As a result of the Florida Spring Task Force's first recommendation, the FGS was asked
to evaluate historical spring data in order to detect and document trends in spring-water quality
and quantity. This document reports the findings of analyses for trends in springs, using data
from the Springs Initiative of FDEP, the WMDs, and the USGS spring sampling programs.

ACKNOWLEDGEMENTS

The authors wish to acknowledge a number of individuals and to thank them for their
assistance. From the Florida Department of Environmental Protection, Division of
Environmental Assessment and Restoration, Bureau of Watershed Management, we would like
to thank Gail Sloane and Jay Silvanima for supplying the authors with data from the TV Network
and their miscellaneous assistance on numerous occasions. Laura Morse assisted in supplying
quality assurance information. Debra Harrington, Rick Hicks, Gary Maddox, Jay Silvanima,
Chris Sedlacek and Paul Hansard (now with the Colorado School of Mines) supplied numerous
editorial comments during the course of the project. From the FGS, we would like to thank
Doug Calman, Rick Green, Tom Greenhalgh, Harley Means, Frank Rupert, Tom Scott, and
especially Ellen McCarron, for their many helpful editorial comments.

We would also like to acknowledge the efforts of numerous people from the water
management districts who supplied us with spring data and constructive comments regarding the
document. In particular the authors would like to thank Kris Barrios, Angela Chelette, Tony
Countryman, Kevin De Fosset, Tom Pratt, and Nick Wooten, from the Northwest Florida Water
Management District (NWFWMD); Ron Ceryak and David Hornsby of the Suwannee River
Water Management District (SRWMD); and Ron Basso, Eric DeHaven, David DeWitt, Joe
Haber, Robert Peterson, and Roberta Starks from the SWFWMD.

We would like to thank Brian Katz and Stuart Tomlinson of the USGS. Both individuals
supplied data and other information that was invaluable to the project. We would like to thank
Dr. Xu-Feng Niu of the Florida State University, Department of Statistics, for contributing to the
section regarding statistical methodologies and to Rich Smith, a graphic designer, who assisted
with making many of the figures.

FLORIDA'S SPRINGS

Scott et al. (2004) presented an excellent overview of Florida's springs. Although they
did not specifically evaluate trends, the authors described hundreds of Florida's springs,
including a description of their water quality. In doing so they described many aspects that
control the water quality and quantity of groundwater. With this in mind, their work can be
considered a precursor to the present trend analysis document. With the authors' permission,
much of the following introduction from the sections labeled "Florida's Springs" to "Differences
in Spring and Well Water Quality" are paraphrased from their work, "Springs of Florida."

BULLETIN NO. 69

Many terms relating to hydrogeology and springs may be unfamiliar to the reader. For this
reason a glossary of terms is found in Appendix B 1. In addition, Appendix B2 elaborates on the
sources of the analytes discussed in this report, along with the probable causes for the trends
observed.

Spring-water discharge comes primarily from the Floridan aquifer system, which is also
the state's principle source of groundwater. The springs provide a "window" into the aquifer,
allowing for a measure of the health of the aquifer. Chemical and biological constituents that
enter the aquifer through recharge processes may negatively affect the water quality in aquifers,
as well as the flora and fauna of springs and spring runs. The declines in water quality can be
directly attributed to Florida's increased population and changing land-use patterns (Florida
Springs Task Force, 2000).

Classification of Springs

Springs are most often classified on the amount of flow or discharge of water. The flow-
based classification listed in Table 1 is taken from Meinzer (1927) (Table 1). One discharge
measurement is all that is required to place a spring into one of eight magnitude categories.
However, it should be understood that each spring exhibits a variable discharge, depending upon
rainfall, recharge and groundwater withdrawals within their recharge areas. This can result in a
spring being classified as a first magnitude spring at one point in time and a second magnitude at
another. In the past, a spring assigned a magnitude when it was first described and continued
with that magnitude designation even though the discharge may have changed considerably over
time. To alleviate this confusion, the FGS (Copeland, 2003) adopted a system using the
historical median of the flow measurements to classify a spring's magnitude. Using the new
system along with the Meinzer system, a spring's magnitude is now based on the median value
of all annual median discharge measurements for the period of record. Of the over 700 springs
inventoried by the FGS, there are 33 first-magnitude springs, 191 second-magnitude, and 151
third-magnitude springs. Most are located in the northern portion of the state (Figure 3).

Table 1. Spring Magnitude.
Discharge
Magnitude Metric Units English Units
1 > 2.832 cms > 100 cfs (> 64.6 mgd)
2 > 0.283 to 2.832 cms > 10 to 100 cfs (> 6.46 to 64.6 mgd)
3 > 0.028 to 0.283 cms > 1 to 10 cfs (> 0.646 to 6.46 mgd)
4 > 0.0063 to 0.028 cms > 100 gpm to 1 cfs (> 100 to 448gpm)
5 > 0.631 to6.308 1ps > 10 to 100 gpm
6 >0.063 to 0.631 1ps > 1 to 10 gpm
7 > 0.473 to 3.785 1pm 1 pint/min to 1 gpm
8 < 0.473 1pm < 1 pint/min
cms = cubic meters per second ips = liters per second
cfs = cubic feet per second pint/min = pints per minute
mgd = million gallons per day 1pm = liters per minute
gpm = gallons per minute

FLORIDA GEOLOGICAL SURVEY

* 462 FGS Visited Springs
O Springs from all Databases
Water

0 25 50 Miles N

0 50 100 Kilometers

Figure 3. Locations of Florida's springs (From Scott et al., 2004).

A second spring classification system is also in use. The Florida Spring Classification
System (Copeland, 2003) (Table 2) is based on an assumption that karst activities have
influenced almost all springs in Florida. Under this system, all springs in Florida can be
classified into one of four categories, based on the spring's point of discharge. Is the point of
discharge a vent or is it a seep and is the point of discharge located onshore or offshore? Since all
springs are either vents or seeps, the classification can be simplified into the following
categories.

A spring vent is defined as an opening that concentrates groundwater discharge to the
Earth's surface, including the bottom of the ocean. The opening is significantly larger than the
average pore space of the surrounding aquifer matrix. A vent is occasionally considered to be a
cave, and groundwater flow from this type of vent is typically turbulent. On the other hand, a
spring seep is composed of one or more small openings in which water discharges diffusely (or
"oozes") from the groundwater environment. The diffuse discharge originates from the
intergranular pore spaces in the aquifer matrix. Flow from seeps is typically laminar.

BULLETIN NO. 69

Offshore Springs

Springs occur both onshore and offshore in Florida. Currently, little is known about the
offshore, or submarines springs, with the exception of the Spring Creek Group-the largest spring
group in Florida, averaging more than one billion gallons of water discharged per day (maximum
flow estimated at more than two billion gallons of water per day [Rosenau et al., 1977; Lane,
2001]). Offshore or submarine springs (Figure 4) are known to exist off Florida's Atlantic and
Gulf of Mexico coastlines. These springs are most common in the offshore portion of Florida
from Crystal Beach Spring (Figure 4, Spring No. 7) to Bear Creek Spring (Figure 4, Spring No.
1). Offshore springs have also been identified off the northeastern and southwestern parts of the
Florida and the western panhandle (Rosenau et al., 1977) (Figure 4). Water-quality data from
some of these springs indicate that, at best, the water is brackish. There are anecdotal reports of
"fresh water" flowing from Florida's offshore springs.

Table 2. Florida's Spring Classification System. (From Copeland, 2003)
SPRINGS
Onshore Offshore
Vent Onshore Vent Offshore Vent
Examples Examples
Karst spring Offshore karst spring
Resurgence (River Rise) Unnamed offshore vent
Estavelle (intermittent resurgence or Offshore estavelle vent
exsurgence)
Subaqueous riverine vent
Subaqueous lacustrine vent
Sand boil
Seep Onshore Seep Offshore Seep
Examples Examples
Subareal riverine seep Unnamed offshore seep
Subaqueous lacustrine seep Offshore estavelle seep

FLORIDA GEOLOGICAL SURVEY

SUBMARINE SPRINGS 7!
1. Bear Creek Spring
2. Cedar Island Spring
3. Cedar Island Springs
4. Choctawhatchee Springs
5. Crays Rise
6. Crescent Beach
7. Crystal Beach Spring
8. Freshwater Cave
9. Mud Hole 9
10. Ocean Hole Spring SCALE
11. Ray Hole Spring
12. Red Snapper Sink 0 80 kms
13. Spring Creek Springs Group I I-
14. Tarpon Springs 0 50 Miles
15. Jewfish Hole
16. Unnamed Spring No. 4

Figure 4. Offshore Springs (From Rosenau et al., 1977).

Spring Recharge Basins

In addition to the awareness of increasing trends in contaminants such as nitrate over the
past several years (Figure 5), there has also been an increased awareness on the drainage basins
that supply water to Florida's groundwater and springs. The amount of water and the nature and
concentrations of chemical constituents that discharge from springs are functions of the geology,
hydrology, weather conditions and land uses within the spring recharge basin. This type of basin,
often referred to as a springshed, consists of those areas within groundwater and surfacewater
basins that contribute to the discharge of the spring (Dehan, 2002; Copeland, 2003). The
springshed consists of all areas where water can be shown to contribute to the groundwater flow
system that discharges from the spring of interest. Karst systems frequently include sinking
streams that transmit surface water directly to the aquifer; the recharge basin may include surface
water drainage basins that bring water into the spring drainage from outside of the groundwater
basin.

BULLETIN NO. 69

Median Nitrate Concentrations in 13
Selected Springs in Florida

1.0 -
0.9 -
0.8 -/
L 0.7 -
S0.6 -/-

S0.4- /-
0 //

z 0.1 -
0

0.0 --
1970 1980 1990 2000
Year

Figure 5. Median nitrate concentrations in 13 selected first-mag-
nitude springs. Springs are Alexander, Chassahowitzka Main,
Fanning, Ichetucknee, Jackson Blue, Madison Blue, Manatee,
Rainbow, Silver, Silver Glen, Volusia Blue, Wakulla, and Wacissa #2
(From Scott et al., 2004).

OVERVIEW OF THE HYDROGEOLOGY OF FLORIDA'S GROUNDWATER

Florida enjoys a humid, subtropical climate throughout much of the state (Henry, 1998).
Rainfall, in the region of the major springs (Figure 1), ranges from 127 cm (50 inches) to over
152 cm (60 inches) per year. As a result of the climate and the geologic framework of the state,
Florida has an abundant supply of fresh groundwater. Scott (2001) estimated that more than 8.3
billion cubic meters [2.2 quadrillion (2.2 x 1012) gallons] of freshwater are contained within
Florida's aquifers. However, only a very small percentage of freshwater is available as a
renewable resource for human consumption.

The Florida peninsula is the exposed portion of the broad Florida Platform. The Florida
Platform, as measured at the 200 meter (more than 600 ft) below sea level contour, is more than
483 km (300 miles) wide. It extends more than 240 km (150 miles) westward under the Gulf of
Mexico, and more than 113 km (70 miles) under the Atlantic Ocean. The present day Florida
peninsula is less than one half of the total platform.

The Florida Platform is composed of a thick sequence of variably permeable carbonate
sediments, limestone and dolostone, lying on older igneous, metamorphic and sedimentary rocks.
The Cenozoic carbonate sediments may exceed 1,220 m (4,000 ft) thick. A sequence of sand, silt
and clay with variable amounts of limestone and shell overlie the carbonate sequence (see Scott
et al, 1991 and Scott, 1992b for discussion of the Cenozoic sediment sequence and the geologic

FLORIDA GEOLOGICAL SURVEY

structure of the platform). In portions of the west-central and north-central peninsula and in the
central panhandle, the carbonate rocks, predominantly limestone, occur at or very near the
surface. Away from these areas, the overlying sand, silt and clay sequences become thicker. As
the rocks sediments compacted and were subjected to other geologic forces, fractures formed.
These fractures allowed water to move more freely through the sediments and provided the
template for the development of Florida's many cave systems.

There are three major aquifer systems in Florida, the Floridan, the intermediate and the
surficial aquifer systems (Southeastern Geological Society, 1986; Scott et al., 1991). The
Floridan aquifer system (FAS) occurs within a thick sequence of permeable carbonate sediments
(see Miller, 1986 and Berndt et al., 1998 for discussion of the FAS). In some areas, it is overlain
by the intermediate aquifer system (IAS) and the intermediate confining unit (ICU) which
consists of carbonates, sand, silt and clay. The surficial aquifer system (SAS) overlies the IAS
(or the FAS where the IAS is absent), and is composed of sand, shell and some carbonate. The
vast majority of Florida's springs result from discharge from the Upper Floridan aquifer system
(UFAS), a subdivision of the FAS as discussed by Miller (1986).

Typical natural recharge to the FAS originates as rainwater. As the acidic rainwater
percolates downward to the FAS, it is made slightly more acidic by carbon dioxide from the
atmosphere and organic acids in the soil. Once in the FAS, the groundwater dissolves portions
of the limestone and enlarges naturally occurring fractures. The dissolution enhances the
permeability of the sediments and forms cavities and caverns. Sinkholes are formed by the
collapse of overlying sediments into the cavities. Occasionally, the collapse of the roof of a cave
creates an opening to the land surface. See Lane (1986) for a description of sinkhole types
common in Florida.

Recharge to the FAS occurs over approximately 55 percent of the state (Berndt et al.,
1998). Recharge rates vary from less than 2.54 cm (one inch) per year to more than 25.4 cm (10
inches) per year. Water entering the upper portion of the FAS eventually discharges from a
spring. The water has variable residence times. Katz et al. (2001) and Katz (2004) found that
water flowing from larger springs had a mean groundwater residence time of more than 20 years
and may reflect the mixing of older and younger waters.

Florida's springs occur primarily in the northern two-thirds of the peninsula and the
central panhandle where carbonate rocks are at or near the land surface. Most of these springs
produce water from the UFAS which consists of sediments that range in age from Late Eocene
(approximately 36 38 million years old [my]) to mid-Oligocene (approximately 33 my).
Miocene to Pleistocene sediments (24 my to 10,000 years) often are exposed in the springs.

The geomorphology of the state, coupled with the geologic framework, controls the
distribution of springs. The springs occur in areas where karst features (for example, sinkholes
and caves) are common, the potentiometric surface of the FAS is high enough and the surface
elevations are low enough to allow groundwater to flow at the surface. Springs generally occur
in lowlands near rivers and streams. There are a number of springs known to flow from vents
within river channels and many more are thought to exist. Hornsby and Ceryak (1998) identified
many newly recognized springs in the channels of the Suwannee and Santa Fe Rivers. Springs

BULLETIN NO. 69

that have yet to be described have been found within the Apalachicola River between Gadsden
and Jackson Counties (H. Means, Florida Geological Survey, personal communication, 2004).

Weather and climatic events affect the appearance of spring water. For example, during
periods of higher than normal precipitation, such as hurricanes, some springs may reverse flow.
When this occurs, stream water flows into the aquifers. During these times, spring water often
has a dark appearance because of the presence of tannins from surfacewater sources. Once
stream levels drop enough, the dark waters again reverse flow. When this occurs, discharge
becomes much clearer. Dryer periods also affect the appearance of springs. For example, during
1998 2002, Florida experienced a major drought with a rainfall deficit in places totaling more
than 127 cm (50 in) (Verdi et al., 2006). The resulting reduction in recharge from the drought,
along with the normal withdrawals, caused a lowering of the potentiometric surface in the FAS.
Many first magnitude springs experienced a significant flow reduction. Some springs ceased
flowing completely. The appearance of the springs also changed as river and lake levels
declined reducing the size of the spring-water body and exposing sediments along the banks.

QUALITY OF GROUNDWATER AND SPRING WATER

Natural Factors Affecting Groundwater and Spring-Water Quality

Most of the Florida land mass is a peninsula that is surrounded by saltwater. Relict
saltwater also underlies the entire state. The reason for this is that the Florida Platform consists
of carbonate rocks that were deposited in a shallow ocean. At the time of deposition, saltwater
existed in their intergranular pore spaces. Gradually over geologic time, sea level was lowered
relative to its position when the carbonate sediments were deposited. Through compaction and
down warping of sediments on both sides of the Platform, a series of complex fracture patterns
developed. The patterns are often reflected at land surface and have actually influenced the
pathways of many of Florida's streams.

Over geologic time, as sea level lowered, the central portion of the Florida Platform was
exposed to the atmosphere. As rainfall percolated downward it eventually replaced the upper
portion of saltwater in the developing aquifers with a freshwater "lens." Today, the irregularly
shaped "lens" is generally thickest in the central portion of the state, where it is over 610 m
(2,000 ft) thick (Klein, 1975). It becomes narrow toward Florida's coastline. The base of the
"lens" is typically a transitional rather than a sharp boundary. Groundwater in the deeper portion
of the "lens", and along the coasts, is mixed with saltwater and has relatively high concentrations
of saline indicators such as sodium (Na), chloride (Cl), and sulfate (SO04).

Water discharging from Florida's aquifer systems and springs has its primary source from
rainfall. Much of the rainfall reaching land surface flows overland to surfacewater bodies,
evaporates, or is transpired by plants. However, a portion of the rainfall percolates downward
through the sediments, or enters sinkholes, where it recharges the aquifers. During its travel
downward from land surface to the water table, and during residence within Florida's aquifer
systems, many factors affect the water chemistry.

FLORIDA GEOLOGICAL SURVEY

A long residence time may allow sufficient time for chemical reactions between the water
and the aquifer rock. As such, water chemistry reflects the composition of the aquifer rock.
Typical residence times range from less than several days (in secondary produced caverns and
sinkholes) to centuries (Hanshaw et al., 1965).

A second factor affecting groundwater chemistry is flow path, which is the length and
depth of the path that the groundwater follows as it flows through an aquifer (Upchurch, 1992).
In general, shallow, short flow paths (which are characteristic of the SAS) result in shorter
residence times for chemical reactions to take place. Consequently, the total dissolved solid
(TDS) content is less than in longer flow-path systems. If the flow path is long (on the order of
tens of kilometers), such as commonly occurs in the FAS, reactions between rock and water
become more probable and the TDS content of the water would be greater as a result of
continued rock-water chemical reactions. Because of the residence time and the flow paths of the
groundwater within an aquifer, the quality of spring water is typically reflective of the
interactions of the major rock types in the aquifer and the groundwater itself.

A third factor which is of particular interest is intergranular porosity (pores through
which water passes between the individual rock matrix grains). Even though Florida's aquifers
have large, secondary cavernous pores spaces, most of the pores tend to be small (Upchurch,
1992). Fortunately, whenever the pores are very small, they act as filters for microbes, small
organic substances, and clay minerals. In general, this results in naturally filtered groundwater
that is very pure and desirable for both drinking water and recreation. Unfortunately, some
pollutants are not always removed and our aquifers can become contaminated.

Differences in Spring- and Well-Water Quality

The processes controlling the water quality in wells is very similar to those controlling
spring-water quality with at least one major difference. Wells are often drilled to production
zones as close to land surface as is economical. This is the situation for the wells used in this
study, which are for the most part monitoring wells. Monitoring wells tend be shallow (median
depth z 80 feet (24 m) (Appendix C). Most water in these shallow wells represents young,
recently recharged water. On the other hand, because springs are major discharge points, spring-
water can be considered to be an integrator of water from the entire springshed. Spring water is
a mixture of young, shallow, freshly recharged water and older water from the deeper portions of
the aquifer. For this reason, spring water tends to be older than the relatively shallow water
found in the monitoring wells used in this study.

Indicators of Groundwater and Spring-Water Quality Problems

Spring water, while it resides in the aquifer, is considered to be groundwater. However,
once spring water exits from the spring onto the earth's surface, it is considered to be surface
water. Because of this change, the question arises whether regulators should apply groundwater
or surfacewater quality standards to the water. Primary and secondary standards with maximum
contaminant limits (MCLs) may exist for an analyte while the water is considered groundwater,
but differ for surface water; or vice versa. Drinking water standards are protective of human

BULLETIN NO. 69

health while surface water criteria are protective of aquatic biota. Although several analytes fall
into this category, Nitrate (NO3 + NO2 as N), and hereafter abbreviated NO3, is a good example.
Based on drinking water criteria, nitrate has a groundwater threshold value of 10 mg/L (Florida
Department of Environmental Protection, 1994). However, no numeric nitrate criteria exist for
surface water, other than Class I surface water which is used for drinking water. The FDEP is
currently developing criteria for spring water. Until legal numeric criteria are established for
nitrates, it should be understood that any reference to threshold values in the following text
simply infers potential water-quality problems.

One of the more disturbing aspects about Florida's groundwater quality has been the
documented steady increase of nitrate over the past several decades (Jones et al., 1996;
Champion and DeWitt, 2000; Means et al., 2003). An example is displayed in Figure 5 (From
Scott et al., 2004). It shows that nitrate concentrations have a greater than 19-fold increase in
nitrate concentrations in 13 selected first-magnitude springs (Alexander, Chassahowitzka Main,
Fanning, Ichetucknee Main, Jackson Blue, Madison Blue, Manatee, Rainbow Group composite,
Silver Main, Silver Glen, Volusia Blue, Wakulla, and Wacissa #2 Springs) between the 1970s
and 2000.

The natural background nitrate concentrations in Florida groundwater are less than 0.05
mg/L (Upchurch, 1992). During the 2001-2002 time frame, the FGS sampled 125 spring vents.
Of the 125 spring vents sampled, none had nitrate concentrations exceeding the 10 mg/L
threshold for Class I surface and drinking water. Fifty-two of the spring vents sampled had
nitrate concentrations exceeding 0.50 mg/L (42 percent) and 30 (24 percent) had concentrations
greater than 1.00 mg/L. Thus, over 40 percent of the sampled springs had at least a ten-fold
increase in nitrate concentrations above background and approximately one quarter of them had
at least a 20-fold increase. The elevated nitrate concentrations may adversely affect the aquatic
ecosystem in springs and spring runs. Further research is still needed and is currently being
sponsored by the Springs Initiative Program. The FDEP is aware of the nitrate issues and has
worked with other governmental agencies to develop a series of steps to reduce nitrate
concentrations in groundwater and springs in the middle Suwannee River Basin where many of
Florida's springs are located (Copeland et al., 2000). The FDEP Bureau of Watershed
Management and the Florida Department of Community Affairs are active in coordinating the
development of spring protection measures.

Another groundwater quality concern is the influence of saline water. Several springs
have concentrations of chloride (Cl; a saline indicator) exceeding the 250 mg/L threshold for
drinking water. Springs with this type of water tend to be located along Florida's coast and along
the St. Johns River. The ultimate source of the saline indicators is from naturally occurring saline
water within the FAS (Klein, 1975), or from sea water near Florida's coasts. When the
concentrations of saline indicators are increasing, it may be the result of: (1) natural
circumstances such as drought, (2) the consequent upcoming of groundwater within the FAS, or
(3) lateral intrusion of salt water due to increased groundwater pumping.

Enterococcus and total coliform bacteria represent a third concern. It is generally
believed that these bacteria originate in fecal matter from warm-blooded animals (Jelinkova
and Rotta, 1978). Total coliform concentrations in several springs has exceeded the

FLORIDA DEPARTMENT OF ENVIRONMENTAL PROTECTION
Herschel T. Vinyard Jr., Secretary

LAND AND RECREATION
Erma Slager, Acting Deputy Secretary

OFFICE OF THE FLORIDA GEOLOGICAL SURVEY
Jonathan D. Arthur, State Geologist and Director

ADMINISTRATIVE AND GEOLOGICAL DATA MANAGEMENT SECTION
Jacqueline M. Lloyd, Assistant State Geologist

Traci Billingsley, Business Manager
Doug Calman, Librarian Specialist
Brian Clark, Environmental Specialist
Jan DeLaney, Environmental Supervisor

Kerry Pond*, Archivist
Sara Sayers, Secretary Specialist
Carolyn Stringer, Operations Manager
Keith Wood*, Computer Programmer Analyst

GEOLOGICAL INVESTIGATIONS SECTION
Harley Means, Assistant State Geologist

Kevin Burdette*, Geologist
Seth Bassett*, Environmental Specialist
Ken Campbell, Professional Geologist
Bob Cleveland, Engineer Specialist
Adel Dabous*, Environmental Specialist
Andrew Flor*, Geologist
Rick Green, Professional Geologist
Eric Harrington, Engineering Technician
Ron Hoenstine, Professional Geologist
Jesse Hurd, Laboratory Technician
Clint Kromhout, Professional Geologist

Michelle Ladle*, Geologist
David Paul, Professional Geologist
Dan Phelps, Professional Geologist
Frank Rupert, Professional Geologist
Guy Richardson, Engineering Technician
Andy Smith, Professional Geologist
Wade Stringer, Engineering Specialist
Eric Thomas, Engineering Technician
Farman Ullah*, Geologist
Christopher Williams, Professional Geologist

HYDROGEOLOGY SECTION
Rodney DeHan, Environmental Administrator

Alan Baker*, Professional Geologist
Scott Barrett Dyer*, Environmental Specialist
Katy Etheridge*, Laboratory Technician
Cindy Fischler, Professional Geologist

Tom Greenhalgh, Professional Geologist
Paul Hansard, Environmental Consultant
Chantel lacoviello*, Environmental Specialist
Stewart Norton*, Environmental Specialist

* Temporary or student employee

FLORIDA GEOLOGICAL SURVEY

drinking water standard of four colonies per 100 ml (Florida Department of Environmental
Protection, 1994). However, it has been determined that these bacteria can complete their normal
life-cycle outside of warm-blooded animals, especially in environments found in parts of Florida
(Fujioka and Byappanahalli, 2004), thus the concentrations of fecal coliform may not necessarily
represent a direct link to warm-blooded animal pathogens. Further research is needed before
definitive conclusions can be made regarding the source of fecal bacteria. Another concern is
concentrations of enterococcus and fecal coliform bacteria with regards to swimming. The
Florida Department of Health has set beach swimming standards and advisory thresholds for
both organisms. To date, exceedances of the standards and thresholds in springs have not been a
problem. Nevertheless, many residents swim in spring runs and these bacteria are a concern.

SPRING SELECTION PROCESS

Very little spring-water quality sampling, mostly by the USGS, occurred until the 1940s.
In 1947, the FGS published its first edition of "Springs of Florida" (Ferguson et al., 1947) which
documented the water quality in the major springs of Florida. The document was revised in 1977
and many previously undocumented springs were sampled (Rosenau et al., 1977). It should also
be noted that during the 1970s, the three northern water management districts were formed.
They were the NWFWMD, the SRWMD, and the (Saint Johns River Water Management District
(SJRWMD). Within a few years, these WMDs, along with the USGS and the already established
SWFWMD, occasionally collected spring-water quality samples. By the 1990s, the NWFWMD,
SRWMD, SJRWMD, and SWFWMD had established periodic to regular sampling, often with
the assistance of the USGS, of springs within their jurisdiction.

Partially due to the sampling efforts of the WMDs, in the 1990s it became apparent that
the water quality in some of Florida's springs was deteriorating. For this reason, in 1999 the
Secretary of the Florida Department of Environmental Protection directed the formation of a
multi-agency Florida Springs Task Force to provide recommendations for the protection and
restoration of Florida's springs. In late 2000 the Task Force made recommendations for the
preservation and restoration of Florida's springs to the Secretary, and in 2001 the Florida
Legislature passed the Florida Springs Initiative. The Initiative authorized funds for FDEP to
begin investigating the status of Florida springs and develop strategies for protecting them. As a
result of the Initiative, the four WMDs, FDEP, and the FGS have cooperated to monitor Florida's
springs.

The Springs Initiative has been responsible for the collection of spring-water quality
since 2001. Beginning with that year, much of the data used in this report were obtained from
Springs Initiative-sponsored samples. Methods of evaluating the data used in this report can be
used in the future to analyze the spring data currently being generated as a result of the Springs
Initiative. In the meantime, data of spring-water quality collected as part of WMD spring
sampling programs were used for this interpretative report.

The FGS requested spring data from each of the four northern WMDs in order to analyze
spring-water quality and quantity for trends. The districts delivered available data to the FGS in
2002 and 2003. It was soon discovered the WMDs had only sporadically sampled their springs

BULLETIN NO. 69

through the 1980s. However, beginning in the early 1990s each district had begun to sample
springs in a semi-consistent manner. Even though data do exist for many springs, only 58
springs (Figure 6) were ultimately included in the analysis; one from the NWFWMD, 14 from
the SRWMD, 15 from the SJRWMD, and 28 from the SWFWMD. Selection was determined
based on the consistency of the data. As a working definition, we considered consistency to be
the longest string of data in terms of time, along with the greatest number of analytes. We also
wanted the largest number of springs to be included that met our concept of consistency. With
these criteria in mind we determined that the time period from January 1991 through December
2003 represented the time in which the most consistent data existed for the greatest number of
springs. We realize that there are several springs that have decades of data. We also realize that
since commencement of the Springs Initiative, many springs now have data, but the time
sequences are short. As a result, our data interpretations are valid only for the 1991-2003 time
frame.

A discussion of analytes evaluated and frequencies of sampling will be discussed later.
Figure 6 displays the location of the included springs in the analyses. A list of the names of
springs, along with location information, can be found in Appendix D.

WELL SELECTION PROCESS

In 1983, the Florida Legislature passed the Water Quality Assurance Act (Florida
Statutes, 1983, Chapter 403.063). As a result, FDEP, with the assistance of the five water
management districts, plus several counties (Alachua, Broward, Collier, Lee, Miami-Dade, and
Palm Beach) established extensive groundwater monitoring networks. The purpose was to
document both ambient groundwater quality conditions (Background Network) and to detect
changes in Florida's groundwater quality resulting from the effects of various land uses and
potential sources of contamination (Very Intense Study Area Network [Scott et al., 1991]). Both
networks were in operation until 2000. A major subdivision of the Background Network was the
Temporal Variability (TV) Network. The TV Network consists of a series of strategically-
located Background Network wells scattered throughout the state. They are sampled on a
monthly to quarterly frequency.

Beginning in 1996, FDEP began a major redesign of its water resource monitoring
efforts. The purpose of the redesign was to characterize the environmental conditions of
Florida's water resources and to determine if those conditions are changing over time. The
revised network (The Status Network) became operational in early 2000. A detailed description
of the Status Network is presented by Copeland et al. (1999). Throughout the redesign process,
the TV Network only had minor modifications. The stated purpose of the redesigned network is
to evaluate temporal variability of Florida's groundwater quality and to determine whether
concentrations of the sampled analytes are increasing or decreasing over time. The TV Network
consists of 46 wells (Figure 7); 25 wells monitor confined groundwater and 21 wells monitor
unconfined groundwater. The wells tap each major aquifer system and are scattered throughout
each of Florida's five WMDs. As can be seen in Figure 7, some of the well locations represent

FLORIDA GEOLOGICAL SURVEY

N

S

Legend
Springs
NWFWMD
SFWMD
S SJRWMD
S SRWMD
S SWFWMD

"%- %.

'*
;...-&

.A~: ~

Miles

0 30 60 90

Kilometers
0 70 140 210

Figure 6. Location of springs analyzed in this report. (A list of the spring names can be found in
Appendix C.)

BULLETIN NO. 69

N

S

Legend
Wells
NWFWMD
SFWMD
M SJRWMD
I SRWMD
I SWFWMD

., -

Miles
0 50 100 150

Kilometers
0 75 150 225

Figure 7. Location of Temporal Variability Network (TVN) wells. (A list of well identifiers can
be found in Appendix D.)

FLORIDA GEOLOGICAL SURVEY

clusters of wells. Wells that monitor confined ground water are sampled quarterly (thought to
have less temporal variability), whereas wells that monitor unconfined groundwater are sampled
monthly.

With respect to the WMDs, the NWFWMD has eight wells, the SRWMD has 10 wells,
the SJRWMD has nine wells, the SWFWMD has 11 wells, and the SFWMD has eight wells in
the TV Network that had significant data for analyses. A list of well names can be found in
Appendix C, along with well construction data.

METHODS

This report uses a relatively simple methodology to determine the condition of spring and
groundwater quality. Most analyses boil down to the single straightforward question, "are
conditions getting better, getting worse, or remaining the same?" Though this report is based
upon several statistical procedures, all address this single question.

With this simple objective in mind, additional elaboration is required to expand upon the
connection with actual statistical tests and methods. First, since "better" and "worse" are
subjective and qualitative questions, an approach that will quantify them is needed. Thus, a
somewhat more objective and quantitative form of the above question becomes, "for the 1991-
2003 period of record, are the indicators decreasing, increasing, or remaining the same?" This
frames the question of quality in the terms of changing quantities between two end-points (i.e.,
the start and finish of a time period of interest); changes in quantities, such as flow and loading,
can be objectively tested in a variety of ways.

In order to test quantities, the last remaining questions are: (1) which quantities and, (2)
over what period of time? These further conditions must be defined. The first question is which
quantities? For this report, as many indicators as possible were tested. This allowed the authors
to ask questions of the largest possible scale; limiting the number of indicators only limits the
possible number of observations and maximizing the number of observations allows the most
comprehensive view of changes that might be of concern. Second, for the quantities examined an
increase or decrease in concentration must be addressed over a time frame. Therefore, in order to
maximize the effectiveness of the analysis, the longest possible time series was chosen for as
many springs as possible. In summary, the choice was for the longest possible time frame for
data il ith the highest quality, for as many indicators possible, and for as many springs as
possible. Laboratory and collection methodologies have varied over the last several decades in
the state of Florida. Variations include not only differences among WMDs, but even use of
different laboratories by the same district, changes within laboratories, incomplete sampling
intervals due to varying purposes and other reasons. Because of this, the earliest starting point for
which data quality could be uniformly assumed to be high (in this case 1991) was chosen; this
created the longest possible time series for analysis (1991-2003) for as many springs as possible.

Regarding the second question, for this report, we chose trend analysis to evaluate a
given time series (between 1991 and 2003) for linear trends. Note that Urquhart and Kincaid
(1999) mentioned that trends may deviate from strict linearity. Nevertheless, they mentioned

BULLETIN NO. 69

that if a trend is present, a linear trend will be present, regardless of the type of mathematical
structure of the trend, e.g. cyclic, episodic, or a stair-step look.

For this report, we were not only interested in detecting the presence of a trend we were
also interested in a statistical method that was relatively insensitive to missing sampling points
(e.g., gaps in data series), outliers, and, data that seldom had normal (Gaussian) distributions.
Based upon these reasons, our choice for analysis was the non-parametric, Mann-Kendall (MK)
test for trends. Discussions of the MK test and other statistical procedures used in the study,
including the corresponding assumptions, are found in Appendix E.

Our last clarification involves interpretation of trends; not all increases are bad nor are all
decreases good. For example, a decrease in nitrate is desirable and is considered to be good. On
the other hand, a long-term decrease in flow is not desirable, since it may indicate an overuse of
the resource. Thus, it can be considered to be bad. Another example, an increase or decrease in
pH may not be considered to be good (if it is extreme), since this analyte is best defined by an
optimal middle range; being far outside that range on either side is bad. The point is that change,
in one direction or another, can be tested and the result has implications regarding the
improvement or degradation of the system in question.

Definition of Trends

Natural systems in general undergo two main types of change: cyclic and linear (A and B,
top of Figure 8). Cyclic change is common in nature. Two common examples of cyclic changes
include diurnal and seasonal changes. Natural changes can also be linear, moving conditions
from one state to another without returning to the original state. The focus of this report is to
document linear trends in water quality and quantity. It is also assumed that trends in certain
analytes are most likely anthropogenic, rather than natural in origin. In this case, three possible
linear trend scenarios can be tested. In each case, a chemical component of a groundwater system
(whether spring or well) can be plotted as a concentration against time (Figure 8, bottom). The
first scenario (on the left) is that the system is increasing in concentration for a particular analyte
(for which the symbol, "+", will be used in this report). One case could be phosphorus. Over a
period of interest, change of concentration can be tested at a specific level of confidence (e.g., at
a 95 percent confidence level, or an a level of 0.05). This means that by the end of the time
period, the concentration was high enough to warrant the designation of being higher than
expected by chance fluctuation alone. Such values are marked as being highly unlikely to have
occurred unless notable changes to the system were introduced. In the opposite case (on the
right), the concentration could have decreased significantly (represented by "-"). Such a trend
suggests a substantial change to the physical environment and would therefore be recorded. The
third scenario (middle chart) is that neither case was observed. As will be detailed below, this is
not a positive statement affirming uniform conditions for the system in question; rather it is a
general category for all conditions not classified within the former two situations. This is a
default option and it is likely that a number of valid trends that could escape detection and be
included in this scenario.

FLORIDA GEOLOGICAL SURVEY

/

3 observations for trends:
Increasing Trend (+) Unable to confirm

Decreasing Trend (-)

Time

Figure 8. Illustration of three options for water-quality trends. Trends can either increase (+), decrease (-), or
otherwise cannot be confirmed. Depending on the analyte, the interpretation is that the system is getting better,
getting worse, or remaining the same. Taking the example of phosphorus, if the trend is increasing, the situation is
getting worse. If the trend is decreasing, it is getting better. Finally, if neither, nothing can be confirmed. All
analyses in this report are assigned to one of these three observations.

Problems with Trends

Trend analyses were largely straightforward and posed few problems. Visual analysis of
time series plots showed that the majority of significant trends were based on a large amount of
data that indeed demonstrated an obvious tendency. However, several exceptions arose and their
handling is addressed in the following sections.

"Remaining the Same" Possibility of Missed Trends

The last case scenario in the phrase "getting better, getting worse or remaining the same"
leaves a question as to the identity of the last category. Note that the last observation-
"...remaining the same"-cannot be addressed statistically. It is, therefore, considered the
alternative case to the situation of an increasing or decreasing trend. Because of this
"... remaining the same" amounts to a catch-all for all remaining observations (i.e., trends that
neither increased nor decreased). Though simple in principle, a clarification should be stated.
Within this last category remains interesting, important, and valuable information-cycles,
interesting structure, nonlinear trends, or other phenomena. More problematic, it is likely the
analyses conducted here "missed" a number of trends (due to the strict confidence limit).

A. Cycle
6
4-
2
-20-

0 5 10 15 20 25
Time (years)

B. Trend
30
20 i------- --------;--------------

. 10 .---.................. -.............. .................... ..................- .....--- .......-

0
0 5 10 15 20 25
Time (years)

Variable

BULLETIN NO. 69

However, it is important to state once again that the purpose for this study was neither to find all
the trends possible, nor to find the largest number of trends; rather, the purpose was to identify
all the trends that could be confidently, statistically labeled as such. Other studies employing
greater power (i.e., ability to detect more trends) could, and probably should, be conducted. But
since this is the first such statewide analysis of water-quality trends, the goal was to minimize the
number of false trends-while maximizing the number of true trends-in order to get the best
picture of where the clearest problems exist.

Outliers

Statisticians often encounter data that lies outside of an expected range of values. The
reasons for this may include data transmission errors, failed laboratory analyses, contaminated
samples, and sometimes accurate data recording unusual situations-causes are not always
visible to analysts. This report was no exception. Technically, there are really only two ways of
dealing with such data. One is to set arbitrary guidelines in advance and handle the data in
accordance. This may include removing outliers that occur above or below a certain accepted
range, e.g. adjust the data. The other approach is to include all outliers in the dataset and
analyze the data regardless. The rationale is that well-maintained data sometimes records outliers
but, with sufficient data, the effects will be minimal. This report chose the latter option and
included all data in all analyses-none were discarded. Their presence was accounted for and
accommodated in several ways. The first was simply by the choice of analysis. Nonparametric
statistics are relatively insensitive to the influence of extreme data points (outliers).

A large number of "bad" data points can still influence even a nonparametric analysis.
Cross-checking results and examining raw data can assist with this judgment. In order to
compare and check the influence of outlying data, every nonparametric statistical trend test [the
Mann-Kendall (MK) will be discussed later] result was checked against a linear regression
parametricc test) of the same data. Further, both analyses were conducted with different
statistical packages: Minitab (Minitab, 2003) for MK and S-PLUS (S-PLUS, 2003) for cross
checking regression analyses. Visual examination of each individual time series was conducted
to corroborate the results of the statistical tests; suspicious data sets were re-analyzed.
Inspections revealed that in the vast majority of cases, reported statistical trends were composed
of time series that showed clear visual trends. Comparison of MK results to linear regressions
(though parametric) showed surprising similarity. Not only did the non-parametric MK results
closely match the parametric analyses, but both were surprisingly unaffected by outliers; thus
providing strong confirmation that both the data was of high quality and that it gave robust
signals.

Detection Levels

Though the data used in these analyses were the best available in terms of quality
assurance, other factors had to be considered. The analysis of outliers demonstrated that
consistency of data handling, laboratory reporting, and subsequent quality assurance was good.
Yet an additional issue surfaced in the plotted time series: the effect of laboratory detection
limits. For statistical purposes, if a sample's concentration was below the laboratory's method
detection limit, it was considered to be the detection limit. For example, where improvements in

FLORIDA GEOLOGICAL SURVEY

laboratory methodologies over time lowered the minimum detection value for several analytes,
trend analyses detected significant downward trends where no such trend existed. Indeed, a
number of time series plots revealed that a number of trends-statistically significant by MK
tests-were actually the artifact of such "stair step" patterns trending down over time (Figure 9).

All data series, therefore, were checked visually for such spurious results. Those data
series found exhibiting such results were removed from consideration in the final analyses.
These were assigned the designation "DL" (detection level) in result tables (e.g., "plus-minus"
charts which will be discussed later); trends created by detection level artifacts were removed
from further analysis.

Well 1943: Example of Detection Limits

U.UU I I I I I
2/12/1990 11/8/1992 8/5/1995 5/1/1998 1/25/2001 10/22/2003
Date

Figure 9. Example of a spurious trend. Detection limit changes can
generate the appearance of false trends. All time series for all analytes
were visually checked for aberrant results since visual inspection was
was necessary to identify artifacts.

HII EEElllI

BULLETIN NO. 69

Sparse Data

The quantity and consistency of existing data varied widely depending on sampling
agency, analyte, and location of springs or wells. Given the amount of data used for the time
series studied herein, a distinction should be made concerning how the quantity of data affected
its quality. First, all reported analyses had sufficient data for time series analysis. Sufficient data
constituted a minimum of 10 points for the entire series. Many software packages would not
generate statistical results without a minimum set of points-which was often 10 values. At the
same time, there is a difference between what constitutes a sufficient amount of data and how
that (sufficient) quantity is structured through time. The former issue concerns whether an
analysis could be conducted while the latter has implications for the reliability of the
interpretation. On one end of the spectrum, some locations only had 10 values, while at the other
some had in excess of 100 values. As it turns out, considering both springs and wells, the
median number of data for the 1991-2003, Sequence A, time frame was 38.

Long-term, consistent data collection is an ideal situation for analysis. However, most
data sets were between the extremes of a lot or too little. Much of the data used here can be
described by the term "sparse data" which we use to mean there are not very many data points in
the time sequence but there were a minimum of ten. Often the spring data were collected for
some other purpose than for time-series analysis having little structure at all. This results in
"messy data." Messy or disorderly data includes missing values, outliers, transcription errors, or
extreme and skewed results. Simply stated, a high proportion of time sequences have varying
amounts of missing data. The missing data hinders reliable data interpretation. One example of
"messy data" is nitrate concentrations at Wakulla Spring (Figure 10). An example is as follows.
Suppose a large number of data points exist at the beginning of the time series, nothing in the
middle, and one point at the opposite end of the series. Also, suppose a trend is detected. The
problem with such a trend is that although it is statistically valid, it may be entirely dependent
upon the single point at the one end of the series. If such a trend is labeled valid, then poor
judgment was used. The best interpretation for a trend exists when there is an abundance of
points sampled consistently for the longest period of time.

Time gaps in data series were the most common problem. In a number of cases data
collected early in the time series were followed by one or more data collection gaps of varying
temporal duration. Such trends are dependent upon the connection of two (occasionally more)
clusters of data. Though the trends may be valid, they are not ideal; this example underscores the
necessity for sampling agencies to implement consistent collection plans over the long term.
Though the data can often be used, its utility can be challenged, or considered suspect. The
reason is that the value of any individual data point is a function of the number and reliability of
nearby data points to which it can be compared over the long term. Data that are sparse,
inconsistently collected, or have large time gaps are substantially less valuable than a consistent,

FLORIDA GEOLOGICAL SURVEY

Wakulla Spring Time Sequence A (1991-2003)

1.1 * U

1.0 U

o" 0.8 y ----
0-

0.6

0.5

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Date
Figure 10. Example of sporadic, unsystematic, and incomplete sampling.
Only seven points were collected in the final eight years of this Wakulla Spring
study. Sparse, inconsistent sampling after 1994 meant the trend seen here was
dependent on relatively few points collected in 2000. Though the trend is statis-
tically valid, this is an excellent illustration of the need for consistent long-
term data collection. Ironically, though budget issues are often responsible for
gaps in sampling, note that missing data greatly reduces the value of points
remaining. Note no data were obtained after 2001 even though Sequence A
continued through 2003.

well-documented time series. Sparse data collection was a significant issue in several notable
springs. For example the trend for nitrate at Wakulla Spring for Sequence A included a six-year
"gap" (1994-2000) during which there was only one sample collected. Although the statistical
conclusion that nitrate concentrations at Wakulla were decreasing is valid (to be discussed later),
with the lack of data during the 1994-2000 time frame, some may doubt the interpretation of a
decreasing trend.

Incomplete data sets existed for many analytes and indicators. Time series for some
analytes (e.g., iron) only had a handful of points over the 13 years. For such very small sets,
trend analysis was meaningless and these were excluded from analysis.

ANALYTES AND INDICATORS

A total of 48 chemical constituents and indicators, with a period of record 1991-2003,
were analyzed for this study. A list of analytes and their corresponding STORET codes can be
found in Appendix F. Data were obtained from several different sources. The state water
management districts offered the most information, followed by FDEP and the USGS. Each
agency used their respective sampling and analysis procedures under whatever guidelines that
were being followed for that particular period of time. This complicated the statistical analyses.
However, identification of useful data led to a field of 48 different analytes of water quality with

BULLETIN NO. 69

a temporal span of 13 years. Table 3 lists, in alphabetical order by column, all of the analytes
examined in this study.

Table 3. Analyte and Indicator List.
Analyte/Indicator
Color Depth to Groundwater Total Chloride
Dissolved Alkalinity Dissolved Phosphorus Total Dissolved Solids
Dissolved Calcium Dissolved Phosphate Total Fluoride
Dissolved Chloride Dissolved Sulfate Total Kjeldahl Nitrogen
Dissolved Fluoride Dissolved Strontium Total Magnesium
Dissolved Iron Enterococci Bacteria Total Nitrate
Dissolved Potassium Fecal Coliform Total Nitrate Nitrite
Dissolved Magnesium Mean Daily Flow Total Organic Carbon
Dissolved Manganese pH Total Phosphorus
Dissolved Sodium Residuals Total Phosphate
Specific Conductance,
Dissolved Ammonia Field Total Potassium
Specific Conductivity,
Dissolved Nitrate Lab Total Strontium
Dissolved Nitrate,
Nitrogen Temperature Total Sulfate
Dissolved Nitrate Nitrite Total Alkalinity Turbidity (Hatch Meter)
Dissolved Organic Carbon Total Bicarbonate Turbidity, Field Sampling
Water Level (feet relative to mean sea
Dissolved Oxygen Total Calcium level)

Sample Collection and Laboratory Analyses

Because many different agencies and laboratories were used to collect and analyze
groundwater samples, unwanted variability was potentially introduced that affected the trend
analyses. At a minimum, potential variability was introduced by: (1) different sampling
personnel, techniques and equipment, (2) sample transport from the field to the laboratory, (3)
environmental and laboratory contamination, (4) concurrent use of several analytical
laboratories, and (5) varying methods of reporting results.

For additional information on the analytes, including abbreviations and units for those 27
analytes that had detectable trends, see Table 4. Spring and well water samples were collected

FLORIDA GEOLOGICAL SURVEY

Table 4. Analytes and Indicators Displaying Trends
Analyte Abbreviation Unit of Measure
Alkalinity Alk mg/L
Ammonia/Ammonium NH3 or NH4 mg/L
Calcium Ca mg/L
Chloride Cl mg/L
Discharge or flow cfs or cubic feet per second
(ft3/sec); one cfs = 0.028 cubic
meters per second
Dissolved Oxygen DO mg/L
Fluoride F mg/L
Iron Fe gg/L and mg/L
Magnesium Mg mg/L
Nitrate as N NO3 or NO3 + NO2 mg/L
Nitrogen (total) N mg/L
Orthophosphate as P PO4 mg/L
pH
Potassium K mg/L
Sodium Na mg/L
Specific Conductance SC kS/cm at 25 C
Stage feet above datum

Strontium Sr gg/L
Sulfate SO4 mg/L
Temperature (of water) Temp C
Total Dissolved Solids TDS mg/L
Total Kjeldahl Nitrogen TKN mg/L
Total Organic Carbon TOC mg/L
Total Phosphorus P mg/L
Total Suspended Solids TSS mg/L
Turbidity Turb NTU (Current)*
SJTU (Historic)
Water Level/Stage WL(msl) feet above mean level (1988)
*JTU and NTU are approximately equivalent, though not identical
msl = mean sea level (National Geodetic Vertical Datum, 1988)

by several agencies and a private company for the SRWMD. Regarding springs, the agencies
include the SRWMD (plus its subcontractor), the SJRWMD, the SWFWMD, and the USGS. For
wells, samples were collected by the NWFWMD, the SRWMD and its subcontractor, the

BULLETIN NO. 69

SJRWMD, the SWFWMD, the SFWMD, and the FDEP. In addition, multiple analytical
laboratories were used to process the samples.

For the 1991-2003 time sequence, spring sampling and analyses faced all of the potential
aforementioned problems. During the same time period, especially during the early days of the
operation of the TV Network, well monitoring encountered many of the same problems that
spring monitoring encountered. The TV Network is operated by the FDEP and by the mid
1990s, the FDEP reduced a considerable portion of unwanted variability by adopting a policy of
using a standardized sampling protocol, a standardized method of sample transport, a single
analytical laboratory, and a standard set of analytical methods and reporting protocols. It is
hoped that one day, spring monitoring throughout Florida will also adopt similar protocols that
will reduce variability.

In spite of the potential variability, not all is negative. For all water samples and data used
in this report, each corresponding sampling agency and/or analytical laboratory has an
individually-approved quality assurance/quality control (QA/QC) plan on file with FDEP.
Regarding QA/QC, the contact for each WMD, FDEP, and the USGS are found in Appendix G.

It should be noted that by 2001, in an effort to achieve standardization, the FDEP adopted
a recommended method for spring-water quality sampling. An overview of the protocols is
found in Scott et al. (2004). The TV Network is managed by the Watershed Monitoring Section
(WMS) of the FDEP. It recently produced an overview of its well water sampling protocols
(Florida Department of Environmental Protection, 2003).

Analytes used in this study

Multiple agencies collected water-quality samples for this publication; however one
agency may have sampled one analyte, while another agency sampled a similar analyte that was
closely related to the first. This was quite common for the analytes nitrate, ammonia, phosphate,
phosphorus, magnesium, sodium, potassium and chloride. Most often, the difference was
between the collection of the dissolved (filtered sample) and total (unfiltered sample) form of the
analyte. It would be preferable if sufficient data in both the dissolved and total forms of these
analytes were available. Unfortunately, it was not always the case. It was decided to combine
the total and dissolved forms because of the importance placed on nutrients in order to obtain a
time series with a sufficient number of data values. We do not recommend this procedure in the
future because it would be better to use one or both of the forms in conducting statistical
analyses. In the recommendations section (discussed later) we recommend a more consistent set
of analytes be used in the future. Nevertheless, for this study, we occasionally used a combined
surrogate form of nitrate, ammonia, phosphate, phosphorus, magnesium, sodium, potassium and
chloride. We did this solely for the purpose of obtaining a sufficient amount of data necessary
for data analyses.

Grouping of Analytes

For convenience, and in an effort to better understand groundwater quality trends, the
analytes (or indicators) were divided into several groups. They are: (1) Field, (2) Rock-matrix

FLORIDA GEOLOGICAL SURVEY

or Rock, (3) Saline or Saltwater, (4) Nutrient, and (5) Other analytes. However, because of
occasional chemical complexities, many analytes are grouped into more than one category.
Table 5 lists them by group. Note that analytes in the table only refer to those that displayed
trends. A detailed description of each analyte is found in Appendix F.

Table 5. Analyte Groups
Rock-Matrix Saline or
Field (Rock)* saltwater Nutrient Other
Discharge Alk Ca Ca and Mg TSS

DO Ca Cl
pH F K
SC Fe Na
Temp
K SC
WL(msl) or Mg S04
Stage
PO4 and P TDS
SC WL(msl) or
Stage
SO4
Sr

Turb
TOC

*Light gray indicates common rock and saline-related indicators while dark gray shows common nutrient analytes.

Descriptions of Analyte Groups

Each analyte represents a measure or variable that can be used to assist in judging the
overall health of Florida's groundwater. Field analytes such as discharge, water level, and flow
describe quantity, but they can also greatly affect quality. The rock analytes suggest upcoming of
water from deep within Florida's aquifers. The saline analytes suggest intrusion or upcoming of
water from the deep portions of our aquifers, and the nutrient analytes are those that stimulate
biological growth or are present as a direct result of biological activity.

Field Analytes

Field analytes represent a grouping for convenience. Measurements of field analytes
were conducted prior to collecting samples for laboratory analyses. The analytes in this group
that were used for trend analyses include: discharge (or flow), dissolved oxygen (DO), pH,
specific conductance (SC), water temperature (Temp), and water level [water level relative to
mean sea level (msl) based on the North American Vertical Datum (NGVD) of 1988)].

Rock-Matrix Analytes

Rock-matrix analytes are those indicative of the rocks making up an aquifer. Because of
natural rock weathering, water that has had a long residence time in an aquifer system has a

OI ,

BULLETIN NO. 69

greater probability of having a high concentration of dissolved rock matrix material. Rock
indicators include: alkalinity (Alk), calcium (Ca), magnesium (Mg), plus to a lesser extent,
fluoride (F), iron (Fe), pH, potassium (K), strontium (Sr), sulfate (SO4), phosphorous (P),
orthophosphate (P04) and SC. Since phosphate and phosphorus are often found in the mineral
fluorapatite, these two analytes are also included in the rock-matrix group.

Saline or Saltwater Analytes

Saline analytes are those associated with salts within either connate water or seawater.
Connate waters are those waters trapped within the sediments at the time of their deposition.
Since the original sediments were deposited in a marine environment, the pore spaces contain
very old saltwater. Saline analytes are obviously also found in the seawater located along
Florida's coasts. The major difference is the age of water. High concentrations of saline
analytes are often an indication of horizontal saltwater encroachment. However, they can also be
an indication of encroachment of highly mineralized water from the deeper portion of Florida's
aquifers, below the fresh-water "lens". The encroachment can be caused by the depletion of the
less dense fresh-water "lens" during a very dry period (e.g. a drought), or by the upcoming of
connate water during periods of heavy groundwater withdrawals. Pumping of groundwater
increased during dry periods and this process exacerbated the apparent intrusion process. Saline
analytes include: calcium, chloride, potassium, sodium (Na), specific conductance, sulfate, total
dissolved solids (TDS), plus water level (MSL) and stage.

Nutrient Analytes

Nutrients represent naturally occurring compounds or elements that are essential for the
growth of living organisms. However, if found in high concentrations, over-enrichment of
nutrients eutrophicationn) in a body of surface water can lead to an overgrowth of plant life
(including algae) and possibly a loss of dissolved oxygen. For this report, nutrient analytes
include: organic carbon, phosphate, phosphorus, a series of nitrogen related species, and to a
lesser extent, Mg, Ca, K, and sulfur in the form of sulfate. The nitrogen related species include
nitrogen, ammonia, total kjeldahl nitrogen, nitrate, and nitrite.

Other Analytes

Analytes in the "other" category do not fit in any of the other four categories. They
represent a miscellaneous group. For trend analyses, the analytes included in this group are
suspended solids, and turbidity.

DATA

The original data were from several sources. The data used for the trends analyses
discussed in this document are in Appendix H.

FLORIDA GEOLOGICAL SURVEY

Data Sources

The majority of the water-quality data from the springs were collected and analyzed by
the water management districts. The data for wells were obtained from the FDEP Watershed
Monitoring Section.

Data Verification

The analysis of the data was verified several times with processes including internal and
external reviews in addition to repeat analyses by each author. The internal review consisted of
audits performed by two of the authors (N. Doran and A. White). These audits included hand-
eye verification of every analysis figure for accuracy. Repeat calculations were performed and
compared with the first value made using new values calculated from the original data. When
errors were found, the data were recalculated by at least two of the co-authors and then replaced.

The external verification was conducted through multiple meetings with WMD staff.
During these meetings many of the actual samplers and initial compilers of the data were present.
Two rounds of discussion took place; once before this document was compiled and again as it
neared completion. These meetings lasted for several hours and many comments were made on
procedures and verification policies. Each concern was subsequently addressed and is exhibited
in the subsequent sections of this document.

Data Preparation

Preparing the data for analysis included addressing the problems of seasonality, missing
values, duplicate data, censored data and detection limits. The data variation caused by seasonal
cycles increases the difficulty of detecting long-term trends. This problem can be alleviated by
removing the cycles before applying tests or by using tests unaffected by the cycles (Gilbert,
1987).

Missing values (i.e., samples that were never collected) cause their own special
difficulties for analysis. For example, suppose 12 monthly water samples were scheduled to be
collected from a selected well in a given year. Suppose that for a variety of reasons, only 10
were actually collected. Thus, the well had two missing values for each indicator sampled.
Unless otherwise stated for the statistical analyses, missing values were treated as if they were
never collected. For example, if only 10 samples were collected, then descriptive statistics were
based on 10, instead of 12 samples.

Duplicate data resulted from two samples collected from the same spring or well
consecutively. The two samples were then labeled as representing two different sampling
events and sent to a laboratory for analyses for the same set of analytes. The purpose of
duplicates is to evaluate the internal precision of a laboratory. For statistical analyses, it was
decided that the primary sample, collected first in the time sequence, would be used. The second
duplicate sample was only used for quality assurance evaluations.

BULLETIN NO. 69

The minimum detection level for analytes from analytical laboratories can cause
environmental data to be censored. That is, the distributions are truncated at their lower ends
near the laboratory detection level. As stated earlier, for statistical analyses, all data reported as
"Below Detection Level (BDL)" were arbitrarily set at the detection level. In addition, it should
be noted that for a given analyte, over the period of record, the laboratory detection levels
changed, giving multiple detection limits.

Time Sequences

Data for analyses were segmented into three time sequences: Sequence A (1991-2003),
Sequence B (1991-1997), and Sequence C (1998-2003). The first sequence spanned the entire
sampling period, January 1, 1991 to December 31, 2003. The two smaller time sequences were
used to assist in identifying and evaluating shorter-term trends (five to six years).

Within the time sequences, each analyte needed to have a minimum of 10 data points in
order for any statistics to be performed. In addition to the minimum number of 10 data points, it
was arbitrarily decided that for Sequence A at least three data points from Sequence B and at
least three data points from Sequence C needed to be present. If Sequence A lacked this
additional criterion, then no analyses were performed on the sequence. As an example, suppose
a spring had 15 data points, 12 in Sequence C and three in Sequence B. An analysis for trend
was conducted for Sequence A, and C, but not B. If a spring only has 14 data points, 12 in
Segment C and two in Segment B, then no analyses was performed for Sequence A nor
Sequence B. However, the statistical analysis was conducted for Sequence C. A question
arises, are only three data points sufficient to represent the time Sequence B or C within
Sequence A? It certainly is not desirable and is an example of "messy" data. This situation was
considered to be sufficient for trend analyses because this study represented the first statewide
analyses for trends. Fortunately, this was not a common situation and, hopefully in the future,
available data will be less "messy."

Data Used for Analyses and Explanation of Appendices

All data presented in this report represent a collaborative effort among the five water
management districts, the U.S. Geological Survey and the Florida Geological Survey for spring
data, plus Alachua, Palm Beach, Broward, Miami-Dade, Lee, and Collier Counties for well data.
This is significant since each sampling agency has its own agenda resulting in different reasons
for the collection of a particular analyte.

Resultant data for both springs and wells can be found in Appendix H. The appendix
contains the actual concentrations for the analytes measured. The state is broken down into three
regions, Northwest, Central, and South Florida. Within each regional folder, the data are placed
in their respective WMD. Missing data were noted with an asterisk. In the folder, the results of
the MK analyses along with the corresponding n (number of data points) and the Sen Slope (SS)
for each spring and well (to be discussed later) can also be found. The format is similar to that
within the data folder. Finally, plus/minus charts (to be discussed later) are also included.

FLORIDA GEOLOGICAL SURVEY

The data in the statistical analysis folder are temporal data. A numerical value of -9999
was included to maintain the order of the spread sheet. A value of -9999 can either mean that
data are missing or it can mean that there are an insufficient number of samples to perform the
statistical analyses (procedures to be discussed later).

The remaining data were placed in tables. The tables contain the following information:
(1) the identification (ID) that names the spring (or well), (2) its location in latitude and
longitude, (3) the time sequence, (4) the dates for which samples were obtained, (5) a p-value for
significant increase, (6) a p-value for significant decrease, (6) the total number of samples within
the sequence, (7) the calculated SSs, and (8) the trend results. With regards to the results, the
tables indicate whether there was a significant increase (UP), decrease (DOWN), or no evidence
of trend. Throughout this report an upward trend will be designated with either an up arrow (1)
or a plus sign (+). A downward trend will either be designated with a down arrow () or a
negative sign (-).

INFORMATION GOALS AND DATA ANALYSIS PROTOCOLS

Information Introduction

The purpose of data analyses was to document water-quality trends in Florida's springs
and wells for the period 1991 2003. Prior to evaluation, a list of information goals was
developed. The goals were then turned into specific questions for which statistical procedures
could be used in an attempt to answer them. The questions are listed below and are followed by a
discussion of the statistical procedures used in this report. A more detailed discussion of all
statistical procedures used in this report can be found in Appendix E. 1 and E2. Minitab Release
14 (Minitab, 2003) and S-PLUS 6.2 Professional Edition (S-PLUS, 2003) were used for all
analyses. The six questions were:

1. What were the statistical distributions for each of the sampled analytes?
2. For Sequence A (the longest time sequence), for each analyte, and for each
spring or well, was seasonality present?
3. For each sequence, for each analyte, and for each spring or well, were
linear time series trends present?
4. If trends were present, what were their slopes?
5. For springs or wells with detectable trends, were they spatially related?
6. If evidence was found to indicate that the degrading trends were man-induced,
what are plausible solutions and recommendations?

Overview of Statistical Analyses Procedures

Descriptive Statistics

Descriptive statistics were produced for each analyte at each spring and well (station) for
the longest time sequence. The descriptions can be found in Appendix I. For each sampled
station for Sequence A, the tables list the analyte (or indicator), the measurement unit, the

BULLETIN NO. 69

number of samples collected, the number of samples with concentrations below the laboratory
detection level (BDL), the minimum value, the first, second and third quartiles, and the
maximum value. The first, second and third quartiles (Q1, Q2, and Q3) correspond to the 25th
50th (median), and 75th percentile respectively. An example of the descriptive statistics is
presented in Table 6. For a given analyte, the reported minimum concentration value in the table
often reflected the minimum detection level reported by the analytical laboratory.

Table 6. Example of Descriptive Statistics Table
(Sequence A; January, 1991 -December, 2003)
Analyte Meas. Num. Num. Min Q1 Median Q3 Max.
Unit Samp. BDLs Value Value Value Value Value
NO3 mg/L 30 7 0.05 0.09 1.00 2.00 10.30
P04 mg/L 29 4 0.05 0.10 0.10 0.15 1.30

Kruskal-Wallis, Mann-Whitney, and Wilcoxon Rank Sum Tests

Seasonality can be thought of as periodic fluctuations or cycles. As an example, Figure
11 displays monthly water temperatures for an imaginary well during the 1992 calendar year.
Not surprisingly the temperature is highest during the summer and lowest during the winter
months, indicating that for temperature there exists a one year cycle.

Figure 11. Monthly water temperatures plotted over the 1992 calendar
year for an imaginary well.

Cycles are not restricted to calendar years. They can occur over virtually any length of
time. Figure 12 displays an example of a cycle longer than one year. In the example, the
concentration of an imaginary analyte has a six year cycle or season. Depending on the variable
of interest, it may or may not have been influenced by cycles whose frequencies are longer than
13 years; Sequence A was 13 years in length (1991-2003). It is difficult to make that

Monthly Water Temperatures
(Calendar Year 1992)
22.6-

22.5

22.4-

I 22.3-

. 22.2-
E
22.1-

22.0-
21.9-
1/1/1992 3/1/1992 5/1/1992 7/1/1992 9/1/1992 11/1/1992 1/1/1993
Date

FLORIDA GEOLOGICAL SURVEY

determination of the cycle's influence on the analyte. With this in mind, the authors were
concerned with the influence of shorter term cycles on Sequence A. Since ground-water samples
were collected on a quarterly and monthly basis while springs were sampled either on a quarterly
or quasi-quarterly fashion, it was decided to determine if cycles in those frequencies were
present in the data.

6

"-" 2
0

S-2 -

o -4 -
-6 -11
0 5 10 15 20 25
T im e (y ears)

Figure 12. Example illustration of seasonality with a six-year cycle.

For each spring or well for Sequence A, the presence of seasonality for each sampled
analyte was determined using a Kruskal-Wallis (KW) test (Hollander and Wolfe, 1973; Gilbert,
1987). Quarterly and monthly seasonality tests were conducted because stations were generally
sampled quarterly and occasionally monthly. It should be noted that monthly seasonality tests
could only be conducted if samples were collected on a monthly or quasi-monthly basis. For the
most part, monthly samples were only collected for 24 of the 46 wells and only for field analytes.
On the other hand, quarterly samples were obtained on the remaining wells and quasi-quarterly
samples were collected on most of the springs. The quasi-quarterly sampling by the WMDs and
the arbitrary seasonal breakdown was as follows: (1) December February, (2) March May, (3)
June August, and (4) September November. It should be noted that as we conducted the
analyses for trends, we found that, based on the four arbitrary seasons, most analytes did not
display significant seasonality. We recognize that in the future, with the acquisition of additional
data and with additional trend analyses, a better breakdown may be discovered. Nevertheless,
for this analysis exercise, the KW test was used to compare the distribution of two or more
populations (seasons) by indirectly comparing their median values during each season as defined
by this study. If we had defined only two seasons, the KW test is equivalent to a Mann-Whitney
(MW) test (Conover, 1999). Both tests are discussed in greater detail in Appendix E. It should
also be noted that the results of the MW test are identical to another very similar test; the
Wilcoxon rank sum test (WT) (Conover, 1999). The WT test was occasionally used during this

BULLETIN NO. 69

study because some of the statistical software used included the WT test rather than the MW test.
Conover (1999) discusses the WT test in detail. Results of the WT test are identical with those of
the MW test.

Consider a situation in which one wants to determine if two populations have the same
statistical distributions for a given season and samples are therefore obtained for each season.
For the MW test, the null hypothesis is that the median of the two populations are the same while
the alternate hypothesis is that they are not. The two samples are combined into a single ordered
sample from smallest to highest. Each observation is then assigned a rank without regard to
which sample it originally came from. The sum of the ranks assigned to those values from one
of the populations is then generated. If the rank sum of the corresponding population is very
small (or very large), there is an indication that the values from one population tends to be
smaller (or larger) than the values from the other. If so, the distributions of the two populations
are not equal. If the rank sums of the two populations are not equal, neither are their medians.

Returning to the KW test, it compares the distribution of more than two populations (e.g.,
seasons). For this report, each test was two-sided. The null hypothesis is that the median
concentration of an analyte sampled in any season is equal to the median of the remaining
seasons. The alternate hypothesis is that the median concentration for at least one season is not
equal to the others. Under the latter scenario, it is assumed that seasonality does exist. For
quarterly data, tests were conducted assuming that each quarter was a season. For monthly data
sets, tests were conducted assuming that each month was a season. The level of significance was
preset to a = 0.05.

For example, 38 temperature samples were collected at Weeki Wachee Main Spring for
time Sequence A. However, data were not available for the period 1991 through most of 1993.
Data were available for the 1993-2001 time frame. All samples were sampled on a quarterly
basis; nine in each of seasons (1), (2), and (4), plus ten in season (3). The KW test compared the
median values for each of the four seasons and, based on the test, it was concluded that the
median of at least one season did not equal the other medians. Thus, it was concluded that
quarterly seasonality does exist for the spring with respect to temperature. Since monthly data
were not available, no conclusion could be made regarding monthly seasonality. Results for
these analyses are found in Appendix J.

Deseasonalized Data

If seasonal cycles were present in the data, the data were deseasonalized using a method
presented by Intelligent Decision Technologies (1998). Although most measurements of central
tendency used in this report pertain to the medians, means were used (Sen, 1968) in the
deseasonalization transformation equation (Intelligent Decisions Technologies, 1998). The Sen
method subtracts the mean of the corresponding season from each datum and then adds the
overall average (mean) of the sequence back to the original datum. For example, suppose 10
years of quarterly data were collected at a site for chloride. Suppose the overall mean of the data
for the 10 year period was 1.0 unit while the mean of the winter quarter was 0.2 mg/L. Now
suppose a concentration for a particular winter quarter sample was 1.2 mg/L. In mg/L, the
corresponding transformed, deseasonalized datum becomes:

FLORIDA GEOLOGICAL SURVEY

x = (original) [seasonal (winter) mean] + (overall mean) = (transformed x)
x = 1.2 mg/L 0.2 mg/L + 1.0 mg/L = 2.0 mg/L.

Mann-Kendall Test

Gilbert (1987) stated that the Mann-Kendall (MK) test can be viewed as a nonparametric
test for zero slope of the linear regression of time-ordered data versus time. Given that it is a
nonparametric technique, it does not depend on an assumption of a particular underlying
distribution. The test identifies correlations in data through temporally ranking the data and then
determining the number of times the concentration goes up or down relative to the previous time
step. It only uses the relative magnitudes of the data rather than their measured values.

Data reported as trace or below the minimum detection level (MDL) were used by
assigning a common value that was smaller than, or equal to, the smallest measured value in the
data set. For this report, below detection level (BDL), was assigned an arbitrary value equal to
the detection level (DL).

Once the seasonality tests were completed (results found in Appendix J), each analyte
was tested for a linear trend using the MK test (a = 0.05) for each time sequence. A macro
program was used for the analysis while working within Minitab [Appendix E. 1)]. However, if
data were insufficient (n < 10), the MK test was not conducted. For this exercise, we always
used a one-sided test. The reason was that we had a preconceived idea as to whether or not a
downward (or upward) trend was an indication that conditions were getting worse (or better).
The results of the MK tests are found in Appendix K.

Seasonal Kendall Test

A common test used in the analyses of time series is the Seasonal Kendall (SK) test
(Gilbert, 1987). It is an adoption of the MK test, and can be used if there is seasonality in the
data. The SK test is the technique of choice. Unfortunately, it has a set of requirements that
were not obtainable. Miller et al. (2004) mentioned that the test requires that the percentage of
censored data (e.g. data reported as BDL) be no more than about five percent. In addition,
Miller stated that there should only be one censoring level. This latter requirement was not
obtainable because our data were obtained from agencies operating independently of each other.
The agencies used multiple laboratories with multiple detection levels, which amounted to
multiple censoring levels. Thus, the SK test was not used in this investigation. In the future, as
better and more consistent data are obtained, the SK test is the recommended test.

Sen Slope

If a trend was found to exist for either non-seasonal or seasonal data, its corresponding
slope was determined using a Sen Slope (SS) estimator (Sen, 1968; and Gilbert, 1987). The
estimator measured the median difference between successive concentration observations over
the time series. The SS was used only to measure the magnitude of the slope. It was not used as
a hypothesis test. Results are found in Appendix K.

BULLETIN NO. 69

Sign Test

For each analyte exhibiting a trend, a map showing the location of the corresponding
station was created. In addition to statistical evaluations, visual estimates were made as to
whether clusters of corresponding upward or downward trends existed. Associations with depth,
land use, and other relationships were evaluated. The last statistical procedure used was the sign
test (Sullivan, 2004). The sign test is a relatively simple procedure to conduct. It was used to
determine if a significant number of stations demonstrated upward or downward trends over a
geographical region. Note that one spring in the SWFWMD was not used in the analyses. The
reason will be discussed later.

As an example, suppose during Sequence A, 29 of the 57 springs displayed an upward
trend for nitrate. Can it be concluded that there exists an upward statewide trend? What if 40 (or
45) of the springs demonstrated upward trends? If one thinks of the causes of trends in
individual stations as being random processes, we could expect about half of the springs to have
upward trends, while about half should have downward trends. On the other hand, if a large
proportion of the springs had upward trends, we might be suspicious that one or more
phenomena were affecting springs and causing the upward trends over a region. Finally, if an
extremely large proportion of the springs demonstrated upward trends, we would become even
more confident that the phenomena were affecting the upward concentrations over a region.

For the sign test, one assigns a (+) value if there is an upward trend and a (-) value if
there is a downward trend. Sullivan (2004) stated that zeros add nothing to the test and therefore
should be eliminated from further analysis. Thus, all springs demonstrating no trend were
assigned a value of zero (0) and were eliminated from further analyses. The test simply
compares the proportion of + values to the values. For this exercise, a was preset to 0.05 for
the level of significance for these evaluations.

Caveats and Assumptions

It should be noted that this study was not set up as a designed experiment. We took
existing data and attempted to evaluate them. As a consequence, there were many less-than-
perfect situations that we needed to address in order to conduct the statistical analyses related to
this project. Whenever one takes existing data which were originally collected with a variety of
goals in mind and attempts to evaluate them with a new set of objectives, problems should be
expected. For example, R.A Fisher, a statistician sometimes referred to as the "Father of
Modern Sitaisi \" (Sullivan, 2004), once stated, "To call in the statistician after the experiment
is done may be no more than asking him to perform a postmortem examination: he may be able
to say what the experiment died of This quote is appropriate for our study. We faced many
unpleasant situations with regard to the data analyses.

One of the major sources of problems pertained to the assumptions of the statistical
procedures (see Appendix E). Generally these tests assumptions are: (1) the measurements are
mutually independent, (2) the observations are random, (3) the populations are continuous, and
(4) the scales or measurements are at least ordinal. For the sign test, the assumptions are slightly

FLORIDA GEOLOGICAL SURVEY

different: (1) the stations are mutually independent, (2) the measurement scale is at least ordinal,
and (3) if the probable outcome of a sign is (+) or (-), for one station, the same is true for all
other stations. With the exception of the independence issue, the assumptions were valid. The
issue of dependency will be addressed during the discussion of the results of the study.

RESULTS

The trend results of every spring and well for each analyte can be found in Appendix K.
What follows is a set of examples from selected springs and wells. Our purpose is to give the
reader a generalized idea about the behavior of analytes during the study period. Although some
discussion regarding the causes of trends at an individual spring will be discussed, the emphasis
of this report is on regional and statewide trends. A discussion regarding the possible sources of
the analytes and the most probably causes of trends can be found in Upchurch (1992) and
Appendix B2. The trend results are divided into both springs and wells by water management
district. There were no springs analyzed in the SFWMD. Thus springs were geographically
divided into the NWFWMD, SRWMD, SJRWMD, and SWFWMD. A note is needed regarding
the relationship between time-series figures and Sequences. If sufficient data were available,
time series analyses were generated for each Sequences A, B, and C. However, if data were
missing for the front or back end of the Sequence, the corresponding figures still cover the entire
sequence. As an example, the first series discussed is for magnesium in Wakulla Spring during
Sequence A (1991-2003). Unfortunately, no data exists for the last two years of the sequence.
Nevertheless, the figure displays the entire sequence. This is true for all time series discussed.

Springs

Northwest Florida Water Management District

In the NWFWMD, only Wakulla Spring (Figure 13) had sufficient data for analyses for
this study. Wakulla was sampled through a piece of tubing placed into a major conduit of the
spring. Thus, the samples are considered spring-water samples. However, for years, the FDEP
has, for administrative purposes, considered the tubing to be a well (Well 67 in the Temporal
Variability Network). Since FDEP considers the station to be a well and the fact that the tubing
taps a spring vent, the station for this report was analyzed both as a spring and a well. Stage data
were collected at the spring vent, and stage was used in lieu of water levels.

Rock and Saline Analytes, Nutrients, and Flow

Rock-matrix analytes included cations such as calcium and magnesium. Wakulla Spring
shows an increase in dissolved magnesium over time Sequence A (1991-2003). Increases in
magnesium and specific conductance (SC) are illustrated in Figure 14. The time series for
magnesium at Wakulla, like many analytes, showed inconsistent sampling over the period of
record. In this case the time period from 1994 to 2000 contained only one point. For the
given data, the MK

BULLETIN NO. 69

= ..-.
L n

Legend

Sakulla

Springs .
NWFWMD
SRWMD

Miles N
10 20 40 60
Kilometers W* E
15 30 60 90 S

Figure 13. Location of Wakulla Spring within the NWFWMD.

test confirmed an increasing trend (p < 0.05). For almost every time-series figure, the median
value of the second half of the sequence was compared to median of the first half, using the
Wilcoxon rank sum test (WT). This enabled us to not only determine the slope of the trend, but
to also better evaluate magnitude of change during the time series. There was a significant
change over the period of record (p < 0.05). Whether by analysis of trends, or comparison of two
halves of the time sequence, the latter half of the study revealed elevated values of dissolved
magnesium. The missing data from the intervening years in the magnesium time series (Figure
14, top) appear to be accounted for in the time series for SC (Figure 14, bottom). Wakulla Spring
showed a clear increase in SC over time. The probable causes for the increases in magnesium
and SC, will be discussed later on a districtwide and statewide perspective.

Figure 15 (top) displays a trend for nitrate-nitrite concentrations in Wakulla Spring for
the 1991-2003 time frame. The time series shows noticeable data gaps from 1995 to 2000 and
again during 2001-2003. In the Wakulla Basin, Chelette et al. (2002) indicated that there are
several significant sources of nutrients. These include effluent from a large spray field, fertilizer
application, and numerous onsite waste disposal treatment sites (OSTDS) within the basin, and
up-gradient of the spring. Fortunately, it appears, since 1991, the concentration of nitrate has
significantly decreased. Nitrate in the form of dissolved nitrate-nitrite declined (Figure 15, top).

FLORIDA GEOLOGICAL SURVEY

Wakulla Spring Time Sequence A (1991-2003)

a^^m ^
^--^^^ a

10.5

10.0-

E 9.5 -

S9.0

8,5

8.0

7,5

MK p-value=0.0002 na= 15 nc= 6
WTp-value=0.0112 SS= 0.0750

Wakulla Spring Time Sequence A (1991-2003)

* .

a
a,

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Date
MK p-value <0.0001 nB = 52 nc=53
WT p-value <0.0001 SS =0.52587

Figure 14. Increasing rock analytes at Wakulla Spring. Magnesium
(top) and specific conductance (SC) (bottom) have upward trends
(p < 0.05). Tests include MK for time series trends, (WT) on sequences
B and C (first and second half of study), plus SS calculations on the
rates of changes.

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

300

()

200

150
150-

BULLETIN NO. 69

Wakulla Spring Time Sequence A (1991-2003)

11 -

1.0

3 0.9

o"
z

d o7

0-6

0.5

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Date
MK p-value = 0.024 nB = 15 nc = 6
WT p-value= 0.035 SS =-0.01

Wakulla Spring Time Sequence A (1991-2003)

7-

*5

4
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
EU

1991 1992 1993 1994 1995 1996 1997 199B 1999 2000 2001 2002 2003

MK p-value -0.0005 n= 52 nc=53
WT p-value -0.0412 SS =-0.0063

Figure 15. Decreasing nitrates and water levels at Wakulla Spring.
Dissolved nitrate (top) and water levels (bottom) had significant trends.
Tests (p = 0.05) included MK and WT. WT compare medians of the
first and second halves of the study. (1.0 m = 3.3 ft)

N U

U
U
U U

U U
U I
C

Y
U

U
U

U

FLORIDA GEOLOGICAL SURVEY

MK and WT tests indicate that whether by trend analysis or comparison of the first and second
half of the time sequences, concentrations of nitrogen decreased. Loper et al. (2005) suggested
that the decreasing nitrate concentrations were due to lowered concentrations of effluent from a
large spray field located within 16 km (10 mi) of the spring. Figure 15 (bottom) illustrates a
significant drop in stage level.

Suwannee River Water Management District

Figure 16 displays the locations of the 15 springs located in the Suwannee River Water
Management District (SRWMD) used in this report. The spring names and the abbreviations are
found in Table 7.

Figure 16. Location of springs within the SRWMD.

BULLETIN NO. 69

Table 7. Suwannee River Water Management District Spring Names and Abbreviations.
Spring Abbreviation Spring Abbreviation
Alapaha Rise ALR Poe Spring POE
Gilchrist Blue Spring GIL Blue Ruth/Little Sulfur Springs RLS
Fanning Spring FAN Rock Bluff Spring RKB
Hart Spring HAR Royal Spring ROY
Homsby Spring HOR Suwannee Blue Spring SBL
Lafayette Blue Spring LBS Telford Spring TEL
Little River Spring LRS Troy Spring TRY
Manatee Spring MAN

Many of the springs are located along the Suwannee River, along a section of the river
which is roughly perpendicular to the coast, at least in its lower stretch. Numerous trends were
noted along an approximate south to north, or lower to upper, river direction. (Specific results are
found in Appendix K).

Rock-Matrix and Saline Analytes

Calcium, magnesium, and sodium increased strongly in the SRWMD over Sequence A.
Increases were particularly strong in the latter half of the study (Sequence C). Increasing trends
were dominant for several analytes. Magnesium and sodium had significant increases in eight of
14 springs with no decreases. Calcium increased in nine springs. Examples of increases in the
rock indicators are shown in Figures 17-20. Figures 17 and 18 illustrate trends in calcium for
four springs over Sequence A while Figures 19 and 20 demonstrate patterns for magnesium for
four springs over Sequence A or C, depending on the spring.

Note that for many of the time-series figures that compared two stations, the vertical
scales do not coincide. By keeping the vertical scales constant, occasionally the variability of
one graph became so small that you could not see it. In the end, we decided it was better to use
inconsistent vertical and to emphasize variability over time.

Changes in calcium in springs for the lower Suwannee River are similar to changes in
springs farther north. Fanning Spring (FAN) and Gilchrist Blue Spring (GIL Blue) illustrate two
of the nine springs that exhibited an increase in calcium (Figure 17). In addition to calcium,
FAN showed significant increases in other rock-matrix and saline indicators including alkalinity,
chloride, potassium, magnesium, sodium, and specific conductance. The time series plot in
Figure 17 shows a gradual increase in calcium from 60 mg/L in 1995 to approximately 80 mg/L
in 2003; the gradual increase had a low variance around the best fit line. When data for
sequences B and C were compared, Sequence C data had clearly higher medians (WT test p-
value, <0.0001, illustrated by box plots in inset figure in bottom corner). Like FAN, for
Sequence A, calcium concentrations in GIL Blue increased. The initial concentration was about
50 mg/L and ended with 65 mg/L; both springs increased in concentration by approximately 20
mg/L. GIL Blue also had many other analytes with upward trends that mirrored FAN: alkalinity,
chloride, magnesium, sodium, and specific conductance. Springs farther north (Figure 18) had
similar looking trends to those springs located farther south (Figure 17), although the overall
trends in other analytes were different. Suwannee Blue Spring (SBL) and Troy Spring (TRY)
together had increases in only

FLORIDA GEOLOGICAL SURVEY

Fanning Springs Time Sequence A (1991-2003)

M.N Mi
- ,,,.,,

-a

0
S40-

20 -

0 -
1/1/1995

5/22/1997

10/11/1999
Date

3/1/2002

MK p-value =0.0005 nA= 10 nB= 40
WT p-value <0.0001 SS =0.2694

Gilchrist Blue Spring Time Sequence A (1991-2003)

*.

1993 1994 1995
MK p-value =0.0005
WT p-value =0.0563

1996 1997 1998 1999 2000 2001 2002
SS =0.2694 Date
nIA 8 nB= 33

Figure 17. Increasing rock analytes at Fanning and Gilchrist Blue
Springs. Fanning (top) and Gilchrist Blue Springs (bottom) had
significant increases in calcium. Tests (p < 0.05) included MK for
for trend, WT on Sequences B and C, plus SS calculations on the rate
of change. Beginning and ending sampling dates for the two springs
are not the same.

7/21/2004

I I I I I I

BULLETIN NO. 69

Suwannee Blue Spring Time Sequence A (1991-2003)

40 -
4/1/1997
MK p-value
WT p-value

12/4/1998 8/8/2000 4/12/2002
<0.0001 SS 0.2565 Date
<0.0001 nA -23 nB -24

Troy Spring Time Sequence A (1991-2003)

70 -

60 -

- 50

40 -

8/5/1995 5/1/1998
<0.0001 SS 0.1782 Date
=0.0006 nA 18 nB= 42

1/25/2001 10/22/2003

Figure 18. Increasing rock analytes at Suwannee Blue and Troy Springs.
Suwannee Blue (top) and Troy Spring (bottom) had significant increases in
calcium. Tests (p < 0.05) included MK for trend, WT on Sequences B and C,
plus SS calculations on the rate of change. Beginning and ending sampling
dates for the two springs are not the same.

12/16/2003

* *~ .*

t .

11/8/1992
MK p-value
WT p-value

FLORIDA GEOLOGICAL SURVEY

Manatee Spring Time Sequence A (1991-2003)

10/1/1997

2/2/2000

6/5/2002

MK p-value <0.0001 SS 0.0351 Date
WT p-value <0.0001 ni = 22 n2 = 22

Hart Springs Time Sequence C (1998-2003)

35 1
5/1/1996
MK p-value
WT p-value

4/12/1998 3/24/2000
=0.0017 SS 0.066667 Date
=0.006 ni = 9 12 = 9

3/5/2002

Figure 19. Increasing rock analytes at Manatee and Hart Springs.
Manatee (top) and Hart Springs (bottom) had significant increases in
magnesium. Tests (p < 0.05) included MK for trend, WT on the first (1)
half and the second (2) half of both time series. Beginning and ending
sampling date for springs are not the same.

6/1/1995

10/6/2004

60

55-

-
E50

I-
45-

40

2/15/2004

III

* 0

BULLETIN NO. 69

Poe Spring Time Sequence C (1998-2003)

2 -
6/26/1997

MK p-value
WT p-value

4/2/1999 1/7/2001 10/15/2002 7/22/2004
Date
0.0001 SS 0.0830
-0.0001 ni 22 n2 22

Lafayette Blue Spring Time Sequence A (1991-2003)

9/11/1997

MK p-value <0.0001 SS = 0.0745
WT p-value =0.0082 ni=22 n2-22

10/13/1999
Date

11/13/2001

12/16/2003

Figure 20. Increasing rock analytes at Poe and Lafayette Blue Springs.
Poe (top) and Lafayette Blue Spring (bottom) had significant increases in
magnesium. Tests (p < 0.05) included MK for trend, WT on the first and
second half of both series, and SS calculation on rate of change. Beginning
and ending sampling dates for the springs are not the same.

16

14

--
10 12

10

0 A --
"* ~~ ~ *' '-- -

*

8/11/1995

FLORIDA GEOLOGICAL SURVEY

calcium, magnesium and sodium. Unlike FAN and GIL Blue, SBL and TRY exhibited no linked
trends in alkalinity, chloride, potassium, or specific conductance.

Toward the south, Manatee (MAN) and Hart (HAR) (Figure 19) began the time series
with approximately four mg/L of magnesium; by the conclusion of the series, they were at six to
seven mg/L. Further north, Poe Spring (Poe) (Figure 20) began at about four mg/L in 1998 and
rose to about 10 mg/L in late 2002. Lafayette Blue Springs (LBS) (Figure 20)( began at about
eight mg/L in 1998 and rose to about 14 mg/L in 2001.

Flow

The SRWMD consistently collected flow or discharge data at specific spring vents during
the same day that they collected water samples from their springs. However, the SRWMD did
not begin collecting discharge data until the 1997-1998 time frame.

During Sequence C, for the SRWMD, flow rates decreased significantly in 12 of 16
springs. There was not an increase in flow rate in any of the springs during the same time
sequence. In addition, the degree of decrease in flow was sometimes severe. Figures 21-24
illustrate the trends for eight springs starting at the lower end of the Suwannee River and moving
inland and northward.

Springs at the lowest end of the Suwannee River included Fanning (FAN) and Hart
(HAR) Springs (Figure 21). Both springs show substantial drops in flow levels. By the end of the
period of record, flow was reduced to approximately half the levels seen at the beginning of the
time series. FAN's highest recorded flows were near 120 cubic feet per second (cfs), but ended
near 50 cfs. HAR's highest recorded flow was approximately 90 cfs and fell to near 40 cfs at the
end fo the time series.

Upstream from these springs are Rock Bluff (RKB) on the Suwannee and Hornsby
(HOR) Springs on the Santa Fe River (Figure 22). Both displayed even sharper declines in flow.
Rock Bluff went from a high of 50 cfs to under 20; flow was reduced to zero cfs briefly in 2001.
Hornsby showed an even stronger decline: over 200 cfs was measured in 1998 and the flow
reduced to zero cfs during a period starting in early 2000. This was followed by a small recovery
of flow rate in 2003.

Poe Spring, on the Santa Fe River, and Little River Sulfur Spring (LRS) on the middle
Suwannee region had strong declines in flow rate (Figure 23 (Sequence A; but mostly C)). Poe
Spring recorded discharges of 60 to 80 cfs near the beginning of the time series but declined to
near 20 cfs by the end. LRS began the time series with a flow rate near 90 cfs and ended near
20. Decline in flow at LRS closely followed a regression line fit to the date (Figure 23, bottom).

Troy (TRY) and Telford (TEL) Springs both had downward trends in flow. Though flow
at Troy was approximately three times higher than Telford (Figure 24; Sequence A, but mostly
C)), there was a slight increase in flow in mid-1998 followed by a decrease for both springs in
early 2000, and then another slight increase in flow occurred in late 2001. Overall, both springs
seem to show that flow was reduced by at least half, with LRS indicating a reduction in flow by a
third at the end of the time series.

BULLETIN NO. 69

Fanning Springs Time Sequence C (1998-2003)

1/1/1997 5/10/1998 9/16/1999 1/22/2001 6/1/2002
MK p-value <0.0001 SS =0.2773 Date
WTp-value =0.0002 nii =11 n2 11

Hart Springs Time Sequence C (1998-2003)

0 i 1
6/12/1996 5/13/1998 4/12/2000 3/13/2002
MKp-value <0.0001 SS -3.2411
WTp-value 0.0012 nii9 n2 9

2/11/2004

Figure 21. Decreasing flow at Fanning and Hart Springs. Between 1998
and 2001 Fanning (top) and Hart Springs (bottom) had significant de-
creases in flow. Tests (p < 0.05) included MK for trend, WT, plus an
SS calculation on rate of change. Over the period of record, flow at both
springs was reduced significantly. Beginning and ending sampling dates are
not the same. [One cfs equals 0.028 cubic meters per second (cms).]

FLORIDA GEOLOGICAL SURVEY

Rock Bluff Springs Time Sequence C (1998-2003)

1/1/1997
MK p-value
WT p-value

10/2/1998 7/2/2000 4/2/2002
-0.0001 SS =-1.0039 Date
0.0001 n11i14 n2 14

Hornsby Spring Time Sequence C (1998-2003)

1/1/1997

11/5/1998

9/8/2000

7/13/2002

MK p-value =0.0003 SS--3.4692 Date
WTp-value =0.0006 ni= 17 n2 17

Figure 22. Decreasing flow at Rock Bluff and Hornsby Springs. Rock
Bluff (top) and Hornsby Springs (bottom) had significant decreases in flow.
Tests (p < 0.05) included MK for trend, WT, plus an SS calculation. For both
Rock Bluff and Hornsby, flow reduced dramatically. Hornsby Springs flow
stopped for a period after late 2000. Beginning and ending sampling dates are
not the same. (One cfs = 0.028 cms)

1/1/2004

300

200

0
LL

100

0-

5/16/2004

BULLETIN NO. 69

Poe Spring Sequence A (1991- 2003)

11/26/1998

MK p-value <0.0009 SS =-0.925
WT p-value 0.0065 n,-22 n2 22

10/20/2000
Date

Little River Spring Sequence A (1991- 2003)

111/1997 8/9/1998 3/17/2000 10/23/2001 6/1/2003
MKp-value<0.0001 SS =-2.1420 Date
WTp-value=0.0039 n1=15 n2= 16

Figure 23. Decreasing flow at Poe and Little River Springs. Poe (top)
and Little River Springs (bottom) had significant decreases in flow. Tests
(p < 0.05) included MK for trend, WT, plus an SS calculation on rate of
change. For both Poe and Little River flow reduced to about one third by
the end of the series. Beginning and ending sampling dates are not the same.
(One cfs = 0.028 cms)

80-

60-

40-

U

U

U p
U
U
U.
U

U U

U

U

111/1997

9/14/2002

8/9/2004

FLORIDA GEOLOGICAL SURVEY

Troy Spring Time Sequence A (1991-2003)

1/1/1997
MK p-value
WT p-value

8/9/1998 3/17/2000 10/23/2001
,0.0001 SS-0.1782 Date
0.0221 ni= 13 n2-13

Telford Spring Time Sequence A (1991-2003)

8/1/1998

MK p-value <0.0001
WT p-value =0.1547 SS =-0.6674
nl=15 n2-15

3/1/2000
Date

9/30/2001

Figure 24. Decreasing flow at Troy and Telford Springs. Troy (top) and
Telford Springs (bottom) had significant decreases in flow. Tests (p < 0.05)
included MK for trend, WT, plus an SS calculation on rate of change. Over
the period of record, flow at both Troy and Telford was reduced by half. Be-
ginning and ending sampling dates are not the same. (One cfs = 0.028 cms)

6/1/2003

- .

10 -
1/1/1997

5/1/2003

BULLETIN NO. 69

Nutrient Analytes

For the study period, nutrients in the SRWMD had more complex patterns than the
patterns of either the salinity indicators or flow. While some nutrient trends were very strong,
others were not as clear. During Sequence A, o f the 15 springs in the SRWMD, TKN increased
significantly in nine springs (with no decreasing trends). Nitrate appeared to decrease (downward
trend in six springs, while it increased in three springs). At the same time, other nutrients-
phosphorus and phosphate specifically-appeared to increase. For phosphorus, there were five
springs with increasing trends and only one spring indicating a decrease; for phosphate there
were four springs with increasing trends and only one spring with a decreasing trend.

The FDEP has a maximum nitrate standard of 10 mg/L for groundwater and Class I
surface water before considering the water impaired. Both water standards are directed toward
maintaining drinking water quality (Florida Department of Environmental Protection, 1994).
Currently, there is not a numeric standard that is directed toward changes and concentrations of
biota in surface water. However, FDEP has established a non-legal threshold for nitrate and
phosphorus for surface water (Florida Department of Environmental Protection, 2004). The
thresholds were based on a statewide evaluation of chlorophyll concentrations in lakes. The
groundwater-surface water relational assessment (SRA) limit is 0.45 mg/L for nitrate.
Groundwater nitrate concentrations exceeding the 0.45 mg/L limit suggest that there is a
potential for adverse affects on aquatic organisms in the spring runs. Technically, the threshold
level applies only to surface water and there is a need to establish a groundwater to surface water
interaction for the threshold to be relevant. Since springs represent an interaction between
groundwater and surface water, we used the threshold level for comparative purposes. Figures
25 and 26 represent examples of changes in nutrients in springs of the SRWMD. Figure 25 is
an example of a decreasing nitrate trend. Nitrate significantly decreased from 1998-2003 for Poe
Spring. For comparative purposes the SRA was chosen as a fixed reference and is the gray line in
Figure 25. Poe Spring exceeded the SRA recommendations prior to 1999, but then declined to
levels below the SRA. Possible reasons for the decline in nitrate in the Suwannee Basin will be
discussed later.

While nitrate often decreased in the SRWMD, TKN rose significantly. Phosphorus also
exhibited some increasing concentrations. Total phosphorus at Poe Spring (Figure 26) almost
doubled from 1999 to 2003. Phosphorus and phosphate both increased at several springs. An
even greater number of upward trends, however, were seen for TKN. Figure 26 (bottom) shows
an increase in TKN at Lafayette Blue Spring over the study period. The plot also illustrates
some of the differences between nutrient and saline trends. While saline and rock-matrix analyte
plots give evidence of clear increases, trend lines for nutrient plots were sometimes less well
defined and potentially not as strong. Figure 26 shows a significant upward trend for total
phosphorus (MK test, p < 0.05) though the p-value of 0.0443 does not indicate such a strong
increasing trend; data for the first and second half of the time sequence were not significantly
different (WT p-value = 0.3633). The potential causes of nutrient and other trends will be
discussed later.

FLORIDA GEOLOGICAL SURVEY

Poe Spring Time Sequence C (1998-2003)

7/1/1997 4/1/1998 1/1/199910/1/19997/1/2000 4/1/2001 1/1/200210/1/20027/1/2003
Date
MK p-value <0.0001 SS -0.0090
WT p-value <0.0001 ni= 22 n22 23

Figure 25. Decreasing nitrates at Poe Spring. The horizontal line represents
The FDEP's SRA limit for total nitrate (0.45 mg/L). Levels exceeded recommended
SRA limit prior to late 1999 but since then were significantly lower.

12

- 08

0

04

0 0

m m SRA Value= 0 45 mg/L

m m * .
F--^

BULLETIN NO. 69

Poe Spring Time Sequence C (1998-2003)

5/1/1997

1/30/1999 10/30/2000 7/31/2002

MKp-value <0.0001 SS = 0.0007
WTp-value 0.0001 n=i22 n2 23

Lafayette Blue Spring Time Sequence A (1991-2003)

U v

U-^

1/1/1995

6/18/1997

12/5/1999

5/22/2002

11/7/2004

MK p-value = 0.0443 SS = 0.0012
WT p-value = 0.3633 nB= 6 nc= 41

Figure 26. Increasing nutrient analytes at Poe and Lafayette Blue Springs.
Poe (top) and Lafayette Blue (bottom) illustrate two increasing nutrients in
the SRWMD: one for phosphorus and the other for TKN. Poe shows a clear
increase in phosphorus while TKN at Lafayette Blue illustrates one of the many
increasing TKN trends in SRWMD springs. Beginning and ending sampling
dates for these springs are not the same.

0.14

0.12-

--
0.10

I-
0.08 -

0.06 -

0.04 -

. .**

5/1/2004

0.4-

0.3-

0.2 -
I-

0.1 -

0.0-

- .-

FLORIDA GEOLOGICAL SURVEY

St. Johns River Water Management District

The springs located in the St. Johns River Water Management District (SJRWMD) and
used in this report are found in Figure 27. Spring names and abbreviations are found in Table 8.

Figure 27. Location of Springs within the SJRWMD.

BULLETIN NO. 69

Table 8. St. Johns River Water Mana gement District Spring Names and Abbreviations
Spring Abbreviation Spring Abbreviation
Alexander Spring Alexander Salt Springs Salt
Apopka Spring Apopka Sanlando Springs Sanlando
Fern Hammock Springs Fern Silver Glen Springs Silver G
Juniper Springs Juniper Starbuck Spring Starbuck
Miami Spring Miami Sweetwater Spring Sweetwater
Palm Spring Palm Volusia Blue Spring Vol Blue
Ponce De Leon Spring PDL Wekiwa Spring Wekiwa
Rock Spring Rock

Calcium, strontium, fluoride, and pH increased in a significant number of springs over
time Sequence A, while phosphate levels decreased. With respect to individual springs, Miami,
Palm, Sanlando, and Wekiwa Springs had at least eight analytes with increasing trends over
Sequence A, while Volusia Blue Spring (Vol Blue) and Sweetwater Spring decreased in at least
eight analytes over the same time sequence. Alexander, Salt, and Silver Glen (Silver G) Springs
each had six or fewer analytes showing any trend (positive or negative). Sequence B had no
districtwide trends. During Sequence C fluoride and pH increased in a large number of springs
while flow decreased at many locations.

Rock-Matrix and Saline Analytes

Increasing trends were associated with the following rock-matrix analytes: strontium,
calcium, pH and fluoride increased over Sequence A. Nine springs had significant increases in
calcium and pH while one spring had a decreasing trend for theses analytes. Both fluoride and
strontium increased in 10 springs. Strontium decreased in one, whereas fluoride decreased in
none. No trends were observed in Sequence B. Within Sequence C, upward trends were
observed for fluoride and pH, while flow decreased. Thus, major changes for the SJRWMD, like
other districts, occurred during 1998 to 2003 (Sequence C).

Figures 28-30 depict increases in two rock-matrix analytes for three springs in Seminole
and Orange Counties. Not depicted are Starbuck, Rock, and Apopka, which showed the same
pattern. Alkalinity and strontium suggest changing chemistries. Both analytes increased in
Palm, Sanlando, and Wekiwa Springs. All plots show trends closely fitting an increasing best-fit
line. Starting at about 116 mg/L for alkalinity, Palm Springs increases to about 126 mg/L.
Sanlando Springs begins at about 130 mg/L and increases to approximately 150 mg/L.
Strontium at Sanlando Springs began around 60 tg/L and ended over 90 with little variation in
the upward trend. Wekiwa Spring started at a higher level (near 100 tg/L) and ended the time
series at about 140 [tg/L. Wekiwa Spring is also unique in showing an apparently quick increase
in concentration between 1993 and 1995. Palm Springs differed from the other two springs in
having strontium concentrations at the start of the study three to four times higher than the other
two springs.

FLORIDA GEOLOGICAL SURVEY

Palm Springs Time Sequence A (1991-2003)

*.-

*

11/8/1992 8/5/1995 5/1/1998
MK p-value= 0.0141 SS= 0 1291 Date
WTp-value= 0.0209 n=-7 nk=21

1/25/2001 10/22/2003

Palm Springs Time Sequence A (1991-2003)

0-

290

250 -

11/8/1992 8/5/1995

5/1/1998 1/25/2001 10/22/2003

MK p-value= 0.0029 SS= 1.2289 Date
WTp-valuc= 0.0081 n,=6 ni=20

Figure 28. Increasing rock analytes at Palm Springs. Alkalinity (top)
and strontium (bottom) increased significantly over Sequence A. Tests
(p < 0.05) included MK for trend, WT on sequences B and C, plus an SS
calculation. Beginning and ending dates for the springs were not the same.

^120
I-

118 -

116

114 -

U *

I I I I I

BULLETIN NO. 69

Sanlando Springs Time Sequence A (1991-2003)

146

141

" 136

11/8/1992 8/5/1995

MK p-value= 0.0010 SS= 0.417
WT p-value = 0.0032 n = 6 ie = 21

5/1/1998 1/25/2001 10/22/2003
Date

Sanlando Springs Time Sequence A (1991-2003)

90

S80 -

I-

70

60 -

11/8/1992 8/5/1995 5/1/1998 1/25/2001 10/22/2003

MK p-value < 0.0001 SS = 0.625
WT p-value = 0.0009 nb= 6 no= 20

Figure 29. Increasing rock analytes at Sanlando Springs. Alkalinity (top)
and strontium (bottom) increased significantly over Sequence A. Tests
(p < 0.05) included MK for trend, WT on sequences B and C, plus an SS
calculation on rate of change. Beginning and ending sampling dates for these
springs were not the same.

. S.

I

*

* *

FLORIDA GEOLOGICAL SURVEY

Wekiwa Spring Sequence A (1991-2003)

120

I-

100 -

90 -

2/12/1990 11/8/1992 8/5/1995 5/1/1998 1/25/2001 10/22/2003

MK p-value < 0.0001 SS = 0.3568 Date
WT p-value <0.0001 nb =21 i =21

Wekiwa Spring Time Sequence A (1991-2003)

180 -

160 -

140

120 -

100

80
2/12/1990 111811992 8/151995 5/1/1998 1/2512001 10/22/2003
MK p-value <0.0001 SS= 0.368 Date
WTp-value<0.0001 nb=17 nc=21

Figure 30. Increasing rock analytes at Wekiwa Spring. Alkalinity (top)
and strontium (bottom) increased significantly over Sequence A. Tests
(p < 0.05) included MK for trend, WT on sequences B and C, plus an SS
calculation. Beginning dates for these springs were not the same.

BULLETIN NO. 69

Nutrient Analytes

For the study period, there were fewer nutrients trends in the SJRWMD than in other
WMDs. For example, both phosphorus and TKN demonstrated few to no changes (no increases
or decreases for phosphorus, no increases and two decreases for TKN). Nitrate showed no clear
trend direction. For example, in the seven springs showing trends for nitrate, three increased and
four decreased. With respect to nutrients, only phosphate showed consistent trends across the
district. Eleven springs decreased in phosphate while no springs increased.

Figure 31 shows two phosphate trends, which also are considered to be Rock-matrix
analytes. Phosphate levels for both Palm and Starbuck Springs fell by nearly half of the initial
concentrations. Phosphate at Palm Springs (top figure) began the time series at approximately
0.15 mg/L and dropped to about 0.09 in 2002. Values from the end of the time series, 2002 to
2003, suggest a rise in concentrations. Starbuck levels began near 0.17 mg/L and fell to about
0.12 mg/L. Similar to Palm Springs, Starbuck appears to record a rise in concentrations near the
end of the time series in 2003.

Southwest Florida Water Management District

Figure 32 shows the locations of the springs in the SWFWMD. Table 9 displays the
corresponding spring abbreviations.

Note that after our analyses, the SWFWMD notified the authors and told us that they now
question the validity of using Boyette Spring data. They now believe it receives a significant
portion of its water from a nearby sinkhole (< 1 kilometer away) and much of the receiving water
is dairy waste (Morrison, 2000). Since the individual spring analyses were already completed,
we decided to keep the spring in the analyses. However, because of the point-source dairy
contamination, Boyette Spring data were removed from districtwide and statewide analyses.
Also, the SWFWMD was the only WMD to analyze for bicarbonate, rather than alkalinity.

The SWFWMD springs had strong trends in rock-matrix, saline and nutrient indicators.
Similar to the SRWMD, rock-matrix and saline indicators rose significantly. Unlike the
SRWMD, nutrient indicators showed different types of trends. Differences in behavior of
nutrients between the SRWMD and SWFWMD suggest regional differences exist between these
two areas. Similarities in rock-matrix and saline trends between the SRWMD and SWFWMD
suggest these trends extend beyond district boundaries. Some springs showed more changing
chemistries than others. Betty Jay, Boyette, and Tarpon Hole Springs had many analytes with
increasing trends. Buckhorn Main and Hidden River No. 2 Spring had a number of decreasing
trends. Those showing no trends among the analytes studied during time Sequence B included
Boat, Bobhill, Rainbow Swamp No. 3, and Wilson Head Springs.

Rock-Matrix and Saline Analytes

Strong increases in both rock-matrix and saline analytes were evident in springs in the
SWFWMD during time Sequence A. Analytes with increasing trends include bicarbonate,

FLORIDA GEOLOGICAL SURVEY

Palm Springs Time Sequence A (1995-2003)

1995 1996 1997 1998 1999 2000 2001 2002 2003
Date
MKp-value <0.0220 SS -0.0013
WT p-value < 0.0081 nb 6 n, 19

Starbuck Spring Time Sequence A (1991-2003)

019 -

0 17 -

S015-
F-

013 -

0 11-
11/8/1992
MK p-value
WT p-value

8/5/1995 5/1/1998 1/25/2001 10/22/2003
0.0096 SS = -0.0013 Date
0.0218 nb= 6 n, = 20

Figure 31. Decreasing phosphate concentrations at Palm and Starbuck
Springs. Palm (top) and Starbuck Springs (bottom) illustrate the most
sharply decreasing nutrient (phosphate) in the district. Both springs show
substantial reductions since the beginning of the time series, with a potential
increase at the end of the series. Note samples were not collected until 1995.

018

016

a 014

0 12
1- 012

010

0 08

. .

m Im

BULLETIN NO. 69

Figure 32. Location of springs within the SWFWMD.

FLORIDA GEOLOGICAL SURVEY

Table 9. Southwest Florida Water Management District Spring Names and Abbreviations
Spring Abbreviation Spring Abbreviation
Betty Jay Spring Betty Jay Hunters Spring Hunters
Boat Spring Boat Lithia Main Spring Lithia Main
Bobhill Spring Bobhill Magnolia Spring Magnolia
Boyette Spring Boyette Pump House Spring Pump House
Bubbling Spring Bubbling Rainbow No. 1 Spring Rainbow No. 1
Buckhorn Main Spring Buckhorn Main Rainbow No. 4 Spring Rainbow No. 4
Catfish Spring Catfish Rainbow No. 6 Spring Rainbow No. 6
Chassahowitzka Chassahowitzka Rainbow Bridge Seep Rainbow Bridge
No. 1 Spring No. 1 Seep
Chassahowitzka Chassahowitzka Rainbow Swamp Rainbow
Main Spring Main Spring No. 3 Swamp No. 3
Hidden River Hidden River Salt Spring Salt
Head Spring Head
Hidden River Hidden River Tarpon Hole Spring Tarpon Hole
No. 2 Spring No. 2
Homosassa Homosassa Trotter Main Spring Trotter Main
No. 1 Spring No. 1
Homosassa Homosassa Weeki Wachee Main Weeki Wachee
No. 2 Spring No. 2 Spring Main
Homosassa Homosassa Wilson Head Spring Wilson Head
No. 3 Spring No. 3

calcium, chloride, potassium, magnesium, sodium, conductivity, sulfate, strontium, and total
dissolved solids. Of these, increases strongly attributable to rock chemistries were bicarbonate
and strontium. Regarding salinity, rises in sodium, chloride, and total dissolved solids were
observed. Analytes in common to both groups included calcium, potassium, magnesium, specific
conductance, and sulfate (which showed strong increases). Similar to other districts, Sequence B
had very few trends. The majority of the influence for these increases occurred during time
Sequence C.

Chloride increased in 18 springs. Figures 33-35 depict the chloride trends from several
springs in the northern SWFWMD along the Gulf Coast.

Springs occur from north to south along the Gulf Coast. Figure 33 includes two springs
from southern Marion County, one of the northernmost counties in SWFWMD. These springs,
Rainbow No. 6 and Bubbling Springs, increased in chloride concentrations, both springs show a
steady increase during the years of Sequence A. Both springs began with 3.0-4.0 mg/L of
chloride and ended the time series with approximately 5.0-6.0 mg/L.

For a couple of springs just to the south in Citrus County, the increase in chloride was
more dramatic (Figure 34). Hunters Spring (top) began the time series with approximately 50
mg/L of chloride. Values rose quickly at one point, more than doubling, and then declined.
Trotter Main (bottom) showed a similar pattern, though with sharper changes. Trotter Main had
values of approximately 50 mg/L near the start, as did Hunters, but then increased to nearly 250

BULLETIN NO. 69

Rainbow No. 6 Spring Time Sequence A (1991-2003)

5.5 -

5.0 -

S4.5 -

4.0 ,

3.5
3.05-

2.5

3.0 *

5.0

2.5 -
1995 1996 1997 1998 1999 2000 2001 2002 2003
MK p-value 0.0002 SS 0.0321 Date
WTp-value 0.0001 nb 16 n = 23

Bubbling Spring Time Sequence A (1 991-2003)

6.5 -

6.0in chloride. Tests (p < 0.05) included for trend, WT on sequences B and

5.0

4.0 -

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Date
MK p-value = 0.0002 SS 0.0417
WT p-value 0.0014 n1b = 13 nl= 23

Figure 33. Increasing saline analytes at Rainbow and Bubbling Springs.
Rainbow No. 6 (top) and Bubbling Springs (bottom) had significant increases
in chloride. Tests (p < 0.05) included MK for trend, WT on sequences B and
C, plus an SS calculation on rate of change. Beginning and ending dates for
these springs are not the same.

FLORIDA GEOLOGICAL SURVEY

Hunter Spring Time Sequence A (1991-2003)

110-

90-

-
70

C-

50 -

30 -

1995 1996 1997

' I I 1 1i
1998 1999 2000 2001
Date

2002 2003

MK p-value = 0.0005 SS = 1.5802
WTp-value =0.0033 nb= 12 n, -20

Trotter Main Spring Time Sequence A (1991-2003)

250 -

200 -

-J
0)150
E

1995 1996 1997
MK p-value 0.0001 SS
WTp-value 0.0009 nib
23

1998 1999 2000 2001
0.1133 Date
15 nc -

Figure 34. Increasing saline analytes for Hunters and Trotter Main
Springs. Hunter (top) and Trotter Main Springs (bottom) had significant
increases in chloride. Tests (p < 0.05) included MK for trend, WT on
sequences B and C, plus an SS calculation on rate of change. Beginning
and ending dates for these springs are not the same.

U
U

U U U
U
U
EU

2002 2003

BULLETIN NO. 69

Weeki Wachee Sequence A (1991-2003)

1995 1996 1997 1998 1999 2000 2001 2002 2003
Date

MK p-value< 0.0001 SS 0.0457
WTp-value <0.0032 nb =15 n,= 23

Bobhill Spring Time Sequence A (1991-2003)

17 -

15
15 -

13 -

11 -

5 -

1995 1996 1997 1998 1999 2000 2001 2002 2003

MKp-value< 0.0001 SS =0.1133
WT p-value < 0.0009 nb = 14 n = 19

Date

Figure 35. Increasing saline analytes for Weeki Wachee and Bobhill Springs.
Weeki Wachee (top) and Bobhill Springs (bottom) had significant increases in
chloride. Tests (p < 0.05) included MK for trend, WT on sequences B and C,
plus an SS calculation on rate of change.

FLORIDA GEOLOGICAL SURVEY

mg/L at one point-a five-fold increase. Figure 35 depicts chloride concentrations at Weeki
Wachee and Bobhill Springs. Weeki Wachee began the time series with only about 6.0 mg/L of
chloride and ended with a concentration of about 8.0 mg/L. Bobhill Spring began about 5.0 mg/L
and ended with about a 9.0 mg/L chloride concentration.

Flow

Flow data were available for only three gaging stations within the SWFWMD. While
there were inadequate data to make statistical conclusions for the SWFWMD, data from the
three' stations suggested possible declines similar to the SRWMD. Homosassa No. 1 flow levels
declined for the years 1996-2003 (Figure 36). Longer-term trends are depicted in Figure 37,
which further illustrates declines in flow. Since the 1960s, average yearly flow for Rainbow
Springs has declined (dark gray line indicates a timeline for Sequence A). WT tests between the
first and second half of the data series show a difference between the two data series. However,
for data representing time Sequence A, results do not indicate a significant difference. This
suggests that trends on the scale of this study (i.e., 13 years) may sometimes be missed in spite of
being part of a larger change (e.g. 40 years of data).

Long-term flow in Weeki Wachee Springs flow data (Figure 37, bottom) has an equally
interesting pattern. Although no regression is displayed, flow data going back to 1904 displays a
rise until the 1960s, followed by a decline until the present. Gray lines illustrate the time line for
Sequence A and that for post-1960. The range in flow during this time appears to be two-fold
(100 to 250 cfs). Such a pattern may reveal that short-period trends may be part of longer-term
cycles for groundwater; implications of this will be addressed later.

Nutrient Analytes

For Sequence A, nitrate increased strongly (19 springs with upward trends, only one
down), while ammonia, phosphate, phosphorus, TKN, and total organic carbon showed little
indication of trends. Since TKN, phosphorus, and total organic carbon decreased somewhat
(though not significantly) it seems to indicate that nitrate-nitrogen alone showed the strongest
increase for SWFWMD. All other analytes showed little change or even evidence of a slight
decline. Also unlike patterns seen in the rock and saline indicators, nitrate increased during both
sequences B and C. This is in contrast with the rock analytes which showed strongest activity
during Sequence C.

Figure 38 and 39 illustrate nutrient trends and their variability for SWFWMD. Hunter and
Magnolia Springs (Figure 38), illustrate clear increases in nitrate over Sequence A. Nitrate
increases occurred regardless of initial concentrations at the beginning of the time series. For
example, Hunter Spring (top) had a consistent increase from a low initial value (about 0.25
mg/L) to just under the SRA threshold of 0.45 mg/L. Hunter remained under the SRA value for
the time period. Magnolia Spring showed a rate of increase similar to Hunter (SS = 0.0046 and
0.0042, respectively). However, Magnolia began the time series with a higher starting value
(about 0.35 mg/L). The trend for Magnolia crossed over the SRA threshold (Figure 38, bottom,
gray line marks SRA value). Similarly Weeki Wachee (Figure 39, top), began the time series

BULLETIN NO. 69

with a value near 0.45 mg/L and increased to 0.8 mg/L by the end of the time sequence. All three
springs had similar rates of change yet differed in their initial concentrations of nitrate.
Homosassa No. 1 Spring Flow

120-

100-

o
I 80-

60-

1/2/1996 1/2/1998 1/2/2000 1/2/2002 1/2/2004
Date
LRp-value <0.0001
n 2879 WTp-value <0.0001

Figure 36. Decreasing flow at Homosassa No. 1 Spring. Flow rates
declined significantly from 1996 to 2003. (One cfs = 0.028 cms)

Overall, TKN showed little activity (one trend up, four down for Sequence A). Figure 39
(bottom) shows a trend in TKN for Boyette Spring. It started with relatively low initial values
and was followed with a rapid increase in 1998. Values went from near 0.5 mg/L to over 3.0
mg/L in short period of time.

Field Analytes

Only two Field analytes demonstrated decreasing trends over Sequence A, pH and
temperature. The analyte pH decreased in 10 springs, while temperature decreased in eight. The
trends for pH largely occurred in Sequence C.

Wells

The wells used for this study are a subdivision of FDEP's Background Network. The
subdivision is referred to as the Temporal Variability (TV) Network. Although independent of
springs, it was believed that evaluating trends in wells would shed insight as to the chemical
behavior of Florida's groundwater. The TV Network subdivides wells into whether they are
confined or unconfined. Because of the small number of confined and unconfined wells per
WMD, for districtwide and statewide analyses, the wells were also combined into one pool (All).

FLORIDA GEOLOGICAL SURVEY

Decreasing trends in water levels and pH were often observed. Because of the drought, the
lowering of water levels was predictable. However, the decrease in pH was unexpected.
Plausible reasons for the declines will be discussed later.

Rainbow Springs Average Flow (1965-2003)

900

800

| 700

600

500

250

S200

0

150

100

3/30/1905

8/15/1932

1/1/1960

U

5/19/1987

Figure 37. Long-term flow trends at two SWFWMD springs. Rainbow (top) and Weeki
Wachee Springs (bottom) show historic changes. For Rainbow, points represent average
flow per year. Although no regression line on the graph, Weeki Wachee data since 1904
(bottom) showed a rise until about 1960 followed by a subsequent fall. Dark gray lines repre-
sent time line for Sequence A. (One cfs = 0.028 cms)

1960 1970 1980 1990 2000
Year

Weeki Wachee Flow (1904-2003)

;i.
,. ~. .' .

J U .
I I *
-- n

I I* I -

I 1 I ____ *
.," U - "

1991-2003
Present
study time
line

1960s to Present

1991-2003
Present study
time line

BULLETIN NO. 69

Hunter Spring Time Sequence A (1991-2003)

0.40-

0.35-

0 0.30-

0.25 -

0.20
2/12/1990
MK p-value <
WT p-value

11/8/1992 8/5/1995 5/1/1998 1/25/2001 10/22/2003
0.0001 SS 0.0046 Date
0.0001 nb 12 n, 20

Magnolia Spring Time Sequence A (1991-2003)

1995 1996 1997 1998 1999 2000 2001 2002 2003
MK p-value < 0.0001 SS 0.0042 Date
WT p-value 0.0001 nb 13 n, 23

Figure 38. Increasing nitrates at Hunters and Magnolia Springs. Hunters
(top) and Magnolia Springs (bottom) illustrate the most actively increasing
nutrient trend in the district (nitrate). Hunters' increase remained below the
the SRA (0.45 mg/L). Magnolia Spring's increase in nitrate began below
the SRA and ends above it; thus it crosses a recommended limit. Beginning
and ending dates for these springs were not the same.

S .
U *M

U 0.

FLORIDA GEOLOGICAL SURVEY

Weeki Wachee Time Sequence A (1991-2003)

8/5/1995 5/1/1998
MK p-value < 0.0001 SS = 0.0055 Date
WTp-value 0.0111 ni 15 n2 =23

1/25/2001

Boyette Spring Time Sequence A (1991-2003)

11/8/1992

8/5/1995
Date

5/1/1998

MK p-value = 0.0005 SS = 0.0100
WT p-value = 0.0002 nb = 27 n, = 22

Figure 39. Increasing nutrient analytes at Weeki Wachee and Boyette
Springs. Weeki Wachee (top) shows a clear increase in nitrate in Sequence
A. Most of the time series for Sequence A included values above 0.45 mg/L.
Boyette (bottom) illustrates an unusual trend in TKN. TKN values sometimes
rose suddenly over a very short period of time in 1998. SWFWMD staff indicat-
ed the source of the TKN was probably from a nearby dairy waste lagoon.
Beginning and ending dates for these springs were not the same.

-J

0"
z

0.5 -

0.4 -

^ U.
*

S0
SRAValue = 045mg/L

10/22/2003

2/12/1990

BULLETIN NO. 69

Northwest Florida Water Management District

Northwest Florida wells (Figure 40) showed a lowering of water levels for Sequence A
(six of eight wells were down, with no increasing trends). Temperature increased in five wells,
while the analyte sodium and sulfate increased in four wells. No wells demonstrated downward
trends for temperature, sodium, and sulfate. The analyte pH decreased in four wells.

Figure 40. Location of wells within the NWFWMD.

Changes in sequences B and C reflected those in Sequence A. For Sequence B water
level fell (six of eight wells had decreasing levels, while none increased). Several wells also
showed increases in sodium (increased in four wells, decreased in none) and conductivity
(increased in five wells, decreased in none).

Unlike springs, where the main influences on the chemistries occurred during Sequence
C, the only notable analyte in well data demonstrating a change was pH. The analyte decreased
in six of eight wells studied (and increased in none).

Water Levels and pH

Figures 41and 42 illustrate several of these trends. The association of water level and pH
suggest a relationship between the two analytes and will be discussed later. A drop in water
levels occurred in both unconfined and confined wells. Confined aquifer Well 312 (Figure 42)
showed a 5 m (15 ft) decline over the period of record.

FLORIDA GEOLOGICAL SURVEY

Well 91 Time Sequence A: pH and WL

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Date
pH
MK p-value 0.0010 n1b 80 n, 66
WT p-value 0.0028 SS -0.0016

Water Level
MK p-value< 0.0001 b = 81 n =- 54
WTp-value 0.0001 SS = -0.0161

Well 129 Time Sequence A: pH and WL

62

58

54

50

- 12

- 10

-8

-6

-4

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Date
pH
MK p-value 0.0158 n1b 53 n, 36 Water Level
WTp-value 0.0001 SS -0.0012 MKp-value <0.0001 n1b 53 n, 24
WTp-value 0.0001 SS -0.0466

Figure 41. Decreasing pH and water levels in NWFWMD wells (#91 and
#129). Tests (p < 0.05) included MK for trend, WT on sequences B and
C, plus an SS calculation on rate of change.

% %-" "

A o

An AA

46`

A .
AA A ``^^ A

.' ^^% 2, ",
? A~^ ^ "
.',,,., ,-

70 -

65-

u 60 -
E

55 -

50 -

BULLETIN NO. 69

Well 131 Time Sequence A: pH and WL

S. WL

S a a
A "m m

U
a N

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

H Date Water Level
MK p-value <0.0001 nb 53 n, 66 MK p-value <0.0001 nb 52 n, 55
WTp-value 0.0004 SS -0.0030 WTp-value <0.0001 SS -0.0187

Well 312 Time Sequence A: WL

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Date Water Level
MKp-value <0.0001 nb 81 ne 35
WT p-value <0.0001 SS -0.0893

Figure 42. Decreasing pH and water levels in NWFWMD wells (#131
and #312). Tests (p < 0.05) included MK for trend, WT on sequences B
and C, plus an SS calculation on rate of change. Water level is in feet
above mean seal level. (One m = 0.3048 ft)

I-- WL

SA A
t A A
A AA

A A
'^ d

FLORIDA GEOLOGICAL SURVEY

Suwannee River Water Management District

The locations of the SRWMD TV wells are displayed in Figure 43. Decreasing water
level trends were observed in the SRWMD. Temperature rose in Sequence C (seven increased
and none decreased). Trends in Sequence A suggested the same pattern as in northwest Florida:
declines in water level and pH.

Figure 43. Location of wells within the SRWMD.

Water Levels and pH

Over the period of record, water level and pH trends looked similar to other districts. As
two examples (Figure 44) of unconfined wells (Wells 1943 and 2465), a drop in water level
appeared to be accompanied by a decrease in pH. Note that Well 2465 had rapidly declining
water levels but a relatively slower change in pH. The water level decreased approximately 5 m
(15 feet) by the end of the study. Confined groundwater showed similar patterns. Figure 45
shows water level and pH falling simultaneously for wells 2585 and 2675. In Well 2585, the
water level drop is a minimum of 3 m (15 ft); some points in the early time series have
substantially higher water level values [18 m (60 ft)] and suggest the difference was even greater.
By far the most extreme water level difference was recorded was Well 2675. From a high point
of 27 m (90 ft) msl in 1994, water levels fell to approximately 9 m (30 ft) by 2003. This is
nearly 18 m (60 ft) difference is due to its location near the Alapaha River. Local karst features
create differences in water levels in response to rainfall. Like other wells in the district, Well
2675 experienced a decline in pH.

BULLETIN NO. 69

Well 1943: pH and WL

a *a a.
*. ; *a I

a > a *
-a a a

I ,. -

-, .'' A~ a

1011/1990

12/31/1993

4/1/1997

7/1/2000

WL

20

1011/2003

pH %h .,tlr I e.% l
MK p-value = 0.0405 n, = 44 n, = 45 MK p-value = 0.0499 nb = 43 n, = 39
WT p-value = 0.3388 SS= -0.0011 WT p-value = 0.0051 SS = -0.0114

Well 2465 Time Sequence A: pH and WL

--- pH
---- WL SO

8.0

45

75 F** -

7"35
.* -

6,5 "

6.0 30
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
pH Date Water Level
MK p-value = 0.0013 n, =68 n, = 64 MK p-value <0.0001 n = 56 n = 59
WT p-value = 0.0001 SS = -0.0009 WT p-value < 0.0001 SS = -0.0944

Figure 44. Decreasing pH and water levels in SRWMD wells (#1943
and #2465). Both wells are unconfined. Tests (p < 0.05) included MK for
trend, WT on sequences B and C, plus an SS calculation on rate of change.
The beginning sampling dates for wells are not the same. (One m = 0.3048 ft.)

FLORIDA GEOLOGICAL SURVEY

Well 2585 Time Sequence A: pH and WL

1.,M

6,6

6.1
2/12/1990 6/27/1991 11/8/1992 3/23/

pt nb= 72 nc = 35
MK p-value 0. 0491
WT v-value = 0.0011 SS = -0.0009

1994 8/5/1995 12/17/1996 5/1/1998
Date
Water Level n, = 64 n, = 34
MK p-value < 0.0001
WT ,-value = 0.0142 SS = -0.1066

Well 2675 Time Sequence A: pH and WL

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
DI ODate Water Level
MK p-value < 0.0001 n, = 73 n, = 63 MK p-value = 0.0043 nri= 39 n, = 31
WT p-valuc =0.0002 SS -0.0016 WT p-value <0.0001 SS -0.0656

Figure 45. Decreasing pH and water levels in SRWMD wells (#2585
and #2675). Both wells are confined. Tests (p < 0.05) included MK for
trend, WT on Sequences B and C, plus SS calculations. Note beginning
sampling dates for wells are not the same. (One m = 0.3048 ft.)

St. Johns River Water Management District

Figure 46 displays the location of TV Network wells in the SJRWMD. The trends for the

wells shown in Figure 46 in the SJRWMD were slightly different from both the NWFWMD and

* **."* .

- ** _

-- pH

W -
_- *-- _- "

+

.

.

%
m
- */ _* .

MILLISECOND	CLASS.METHOD	MESSAGE
0	sobekcm_page_globals.constructor
0	sobekcm_page_globals.constructor	Application State validated or built
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.constructor	Navigation Object created from URI query string
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.display_item	Retrieving item or group information
0	sobekcm_page_globals.get_entire_collection_hierarchy	Retrieving hierarchy information
0	sobekcm_assistant.get_entire_collection_hierarchy
0	cached_data_manager.retrieve_item_aggregation
0	cached_data_manager.retrieve_item_aggregation	Found item aggregation on local cache
0	item_aggregation_builder.get_item_aggregation	Found 'all' item aggregation in cache
0	system.web.ui.page.page_load (ufdc.page_load)
0	sobekcm_page_globals.constructor.on_page_load
0	html_echo_mainwriter.add_style_references	Adding style references to HTML
0	html_echo_mainwriter.add_text_to_page	Reading the text from the file and echoing back to the output stream
97	html_echo_mainwriter.add_text_to_page	Finished reading and writing the file

MILLISECOND

CLASS.METHOD

MESSAGE

sobekcm_page_globals.constructor

Application State validated or built

sobekcm_database.verify_item_lookup_object

sobekcm_page_globals.constructor

Navigation Object created from URI query string

sobekcm_database.verify_item_lookup_object

sobekcm_page_globals.display_item

Retrieving item or group information

sobekcm_page_globals.get_entire_collection_hierarchy

Retrieving hierarchy information

sobekcm_assistant.get_entire_collection_hierarchy

cached_data_manager.retrieve_item_aggregation

Found item aggregation on local cache

item_aggregation_builder.get_item_aggregation

Found 'all' item aggregation in cache

system.web.ui.page.page_load (ufdc.page_load)

sobekcm_page_globals.constructor.on_page_load

html_echo_mainwriter.add_style_references

Adding style references to HTML

html_echo_mainwriter.add_text_to_page

Reading the text from the file and echoing back to the output stream

html_echo_mainwriter.add_text_to_page

Finished reading and writing the file

	Front Cover
	Title Page
	Preface
	Table of Contents
	Executive summary
	Introduction
	Florida's springs
	Overview of the hydrogeology of...
	Quality of groundwater and spring...
	Front Matter
	Spring selection process
	Well selection process
	Methods
	Analytes and indicators
	Data
	Information goals and data analysis...
	Results
	Discussion
	References
	Appendices