Measurement and prediction of herbicide transport into shallow groundwater

MISSING IMAGE

Material Information

Title:
Measurement and prediction of herbicide transport into shallow groundwater
Physical Description:
xv, 214 leaves : ill. ; 28 cm.
Language:
English
Creator:
Smith, Matthew Clay, 1957-
Publication Date:

Subjects

Subjects / Keywords:
Herbicides -- Environmental aspects   ( lcsh )
Groundwater -- Pollution   ( lcsh )
Atrazine -- Environmental aspects   ( lcsh )
Agricultural pollution -- Environmental aspects   ( lcsh )
Agricultural Engineering thesis Ph. D
Dissertations, Academic -- Agricultural Engineering -- UF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1988.
Bibliography:
Includes bibliographical references.
Additional Physical Form:
Also available online.
Statement of Responsibility:
by Matthew Clay Smith.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 025018511
oclc - 20233935
System ID:
AA00025751:00001

Full Text










MEASUREMT AND PREDICTION OF HERBICIDE
TRANSPORT INTO SHALOW GMNDWATER






By

NATIHEW CIAY SMITH


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENT
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY




UNIVERSITY OF FLODRIDA


1988


F















DEDICATION



To my mother,

Corrine L. Smith



She was taken from her family by illness during this research. Her

wisdom, courage, and strength continue to inspire me. Without that inner

strength, I would never have made it this far.

I love you Mom.















ACKNWEDGMENIS

I would like to express my sincere appreciation to the following:



Dr. A. B. (Del) Bottcher, chairman of my advisory ccmittee, for

his friendship, patience, and guidance. He provided valuable advice and

philosophy during the highs and lows encountered in my Ph.D. program.

Dr. K. L. Campbell, cochairman of my advisory committee, for his

guidance and support. His calm and reasoned approach to problem solving

is a model which I need to emulate.

Dr. E. D. Threadgill, for serving on my advisory committee, for

making available the field site on which this research was conducted,

and for being the best engineer/scientist/administrator that I could

ever hope to be associated with.

Dr. W. C. Huber, for serving on my advisory coanittee, and being

one of the very best instructors that I have encountered during many

years of college. If all instructors were that good, my GPAs would have

been higher.

Dr. P. S. C. Rao, for serving on my advisory committee, for showing

a sincere interest in my research, and for teaching me several very

valuable lessons during the qualifying exam.

Ms. M. W. Smith, for being a wonderful, loving, and supportive

wife. She assumed a greatly disproportionate share of child care and

household duties so that I could devote time to this research.


iii









Nicholas and Sarah Smith, my children, for their love and for

providing meaning and purpose to my life and education.

Dr. G. W. Isaacs, for financial support in the form of a research

assistantship.

Dr. D. L. Thomas, for his friendship and support. He shielded me

from many distractions at work while I mcopleted this document.

Dr. W. B. Wheeler and Ms. S. J. Scherer, at the UF Pesticide

Residue Laboratory, for allowing me to use their equipment and guiding

me through the intricacies of pesticide residue analysis.

Mr. L. A. Asmussen, Dr. R. A. Leonard, Dr. W. G. Knisel, Ms. L. R.

Marti, and others at the USDA Southeast Watershed Research laboratory for

their advice, encouragement, and support.

USDA, Southern Region Pesticide Impact Assessment Program, for

financial support in the form of a grant.

Many others, too numerous to list here, in both Gainesville and

Tifton who contributed in many ways to the success of the project.









TABLE OF COENETS

Page

ACKNOWLEDGEMENTS.................................................. iii

LIST OF TABLES .................................................... vii

LIST OF FIGURES .................................................. viii

ABSTRACT' ........................................................... xiv

CHAPTERS

1 INTRODUCTION ................................................ 1

2 OBJECTIVES.................................................. 4

3 REVIEW OF THE LITERATURE.................................... 5

3.1 Evidence of Pesticide Residues in Groundwater......... 5

3.2 Factors Which Influence Pesticide Transport to

to Groundwater........................................ 7

3.3 Predicting Pesticide Transport....................... 15

3.4 Field Studies of Pesticide Transport................. 27

3.5 Summary .............................................. 36

4 EXPDERIMENTAL METHODS ....................................... 39

4.1 Field Site Description............................... 39

4.2 Site Instrumentation................................. 42

4.3 Chemical Applications................................ 49

4.4 Sample Collection and Storage........................ 52

4.5 Sample Analysis...................................... 55

5 MODELING THE EDPERIMENTAL SITE............................. 66

5.1 Selection of Ccommon Input Parameter Values........... 67

5.2 Parameters Unique to PRZM............................ 73

5.3 Parameters Unique to GLEAMS.......................... 77









6 RESULTS AND DISCUSSION..................................... 79

6.1 Data Collection ...................................... 79

6.2 Sample Analysis ...................................... 83

6.3 Chemical Applications................................ 85

6.4 Chemicals in the Unsaturated Zone.................... 91

6.5 Chemicals in the Saturated Zone...................... 107

6.6 Model Results and Comparisons....................... 142

7 SUMMARY AND CONCLUSICS ................................... 157

8 RECMMENDATICNS FOR IMF3OVEME3S AND FURTHER STUDY........ 160

REFERENCES........................................................ 162

APPENDICES

A. MO41IORING WELL STATISTICS................................ 170

B. HOW TO GET COMPLETE DATA SET ............................. 173

C. WATER BAIANCE PROGRAMS.................................... 176

D. SURFACE PLOTS OF ATRAZINE CONCENTRATION IN GROUNDWATER.... 200

BIOGRAPHICAL SKETCH............................................... 213









LIST OF TABLES


Page

5.1. Soil properties used in simulations........................ 69

5.2. Chemical properties used in simulations ..................... 70

5.3. Chemical applications summary.............................. 73

6.1. Chronological summary of field site activities..............80

6.2. Chemical application results............................... 88

6.3. Measured application rates and soil surface concentrations

of atrazine and alachlor................................... 91

6.4. Mean concentrations of atrazine (mg/kg) in soil samples..... 96

6.5. Mean concentrations of alachlor (mg/kg) in soil samples..... 97

6.6. Mean concentration of bromide (mg/L) at each sampling

following first application............................... 105

6.7. Mean concentration of bromide (mg/L) at each sampling

following second application............................... 106

6.8. Total mass of chemicals in the saturated zone............... 133

6.9. Simulated mass flux of chemicals........................... 142

A.1. Monitoring well statistics................................ 171

B.1. Example listing of the water sample data set............... 174

B.2. Example listing of the soil sample data set................ 175

C.i. Sample output from program ANALYZE........................ 191

C.2. Sample output from program FLUX........................... 198








IST OF FIGURES


4.1. Contour maps of soil surface and restricting layer

showing locations and ID labels of monitoring wells..........41

4.2. Cross-section of soil profile through application area

showing locations of monitoring wells and soil solution

samplers.................................................... 45

6.1. Uniformity of atrazine application.......................... 86

6.2. Uniformity of alachlor application.......................... 87

6.3. Uniformity of bromide application on 4/27/87.................89

6.4. Comparison of alachlor and atrazine concentrations in

application samples......................................... 90

6.5. Correlation between atrazine concentrations in the top 5 cm

of soil and application solution concentrations.............. 93

6.6. Correlation between atrazine concentrations in the top 5 cm

of soil and application rate................................ 93

6.7. Correlation between alachlor concentrations in the top 5 ancm

of soil and application solution concentrations.............. 94

6.8. Correlation between alachlor concentrations in the top 5 ancm

of soil and application rate................................ 94

6.9. Bromide and atrazine concentrations in solution sampler

09N-2 as a function of total water applied since

application................................................. 98

6.10. Bromide concentration for six sampling locations following

the first application at a 61 cmn depth..................... 101


viii









6.11. Bromide concentration for six sampling locations following

the first application at a 122 on depth ..................... 101

6.12. Bromide concentration for six sampling locations following

the first application at a 183 an depth.................... 102

6.13. Average bromide concentration for the three sampling depths

following the first application............................ 102

6.14. Bromide concentration for six sampling locations following

the second application at a 61 cm depth..................... 103

6.15. Bromide concentration for six sampling locations following

the second application at a 122 cmn depth..................... 103

6.16. Bramide concentration for six sampling locations following

the second application at a 183 cm depth.................... 104

6.17. Average bromide concentration for the three sampling depths

following the second application............................ 104

6.18. Bromide concentration in the groundwater on 12/15/86......... 108


concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration


the

the

the

the

the

the

the

the

the

the


groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater


12/22/86.........108

12/29/86.........109

01/05/87.........109

1/12/87. .........110

1/19/87 .........110

5/05/87 .........111

5/08/87. ........111

5/13/87.........112

5/18/87.........112

5/25/87.........113


6.29. Bromide concentration in the groundwater on 6/01/87......... 113


6.19.

6.20.

6.21.

6.22.

6.23.

6.24.

6.25.

6.26.

6.27.

6.28.


Bromide

Bromide

Bromide

Bromide

Bromide

Bromide

Bromide

Bromide

Bromide

Bromide








6.30. Atrazine concentration in the groundwater on 2/23/87.........115

6.31. Atrazine concentration in the groundwater on 3/02/87.........115

6.32. Atrazine concentration in the groundwater on 3/09/87.........116

6.33. Atrazine concentration in the groundwater on 3/16/87......... 116

6.34. Atrazine concentration in the groundwater on 3/23/87......... 117

6.35. Atrazine concentration in the groundwater on 3/31/87......... 117

6.36. Contour plot of water table elevation on 5/08/87 showing

direction of flow ............................................ 118

6.37. Concentration of atrazine and water table elevation in

well 09-11.................................................. 120

6.38. Nitrate concentration in groundwater on 5/01/87............. 122

6.39. Nitrate concentration in groundwater on 5/03/87............. 122

6.40. Nitrate concentration in groundwater on 5/05/87..............123

6.41. Nitrate concentration in groundwater on 5/08/87.............. 123

6.42. Nitrate concentration in groundwater on 5/13/87............. 124

6.43. Nitrate concentration in groundwater on 5/18/87............. 124

6.44. Nitrate concentration in groundwater on 5/25/87..............125

6.45. Nitrate concentration in groundwater on 6/01/87..............125

6.46. Chloride concentration in groundwater on 5/01/87............. 127

6.47. Chloride concentration in groundwater on 5/03/87............. 127

6.48. Chloride concentration in groundwater on 5/05/87............ 128

6.49. Chloride concentration in groundwater on 5/08/87............. 128

6.50. Chloride concentration in groundwater on 5/13/87............. 129

6.51. Chloride concentration in groundwater on 5/18/87............ 129

6.52. Chloride concentration in groundwater on 5/25/87............. 130

6.53. Chloride concentration in groundwater on 6/01/87.............130









Total mass of atrazine stored in the saturated zone......... 133

Subareas used in water balance............................. 135

Comparison of percolation volumes predicted by GLEAMS and

PRZM ........................................................ 144


6.57. Catparison of measured and PRZM predicted atrazine


concentrations in the soil

6.58. Comparison of measured and

concentrations in the soil

6.59. Comparison of measured and

concentrations in the soil

6.60. OCaparison of measured and

concentrations in the soil

6.61. Comparison of measured and

concentrations in the soil

6.62. Comparison of measured and

concentrations in the soil

6.63. Ccnparison of measured and

concentrations in the soil

6.64. Ocparison of measured and

concentrations in the soil

6.65. COoparlson of measured and

concentrations in the soil

6.66. Comparison of measured and

concentrations in the soil

6.67. Comparison of measured and

concentrations in the soil


on 11/18/86 ...................... 145

PRZM predicted atrazine

on 11/24/86......................145

PRZM predicted atrazine

on 12/22/86......................146

PRZM predicted atrazine

on 2/09/87.......................146

PRZM predicted atrazine

on 3/16/87.......................147

PRZM predicted atrazine

on 5/25/87.......................147

PRZM predicted alachlor

on 11/18/86...................... 148

PRZM predicted alachlor

on 11/24/86......................148

PRZM predicted aladchlor

on 12/22/86......................149

PRZM predicted alachlor

on 2/09/87....................... 149

PRZM predicted alachlor

on 3/16/87.......................150


6.54.

6.55.

6.56.









6.68. Oumparison of measured and IRZM predicted alachlor

ccentraticns in the soil on 5/25/87 ................. ...... 150

6.69. Measured and RZM predicted bromide concentrations in the

soil solution at a 61 ncm depth following the first

application................................................ 152

6.70. Measured and FPRZM predicted bromide concentrations in the

soil solution at a 122 ncm depth following the first

application................................................ 152

6.71. Measured and PRZM predicted bromide concentrations in the

soil solution at a 183 cm depth following the first

application................................................ 153

6.72. Measured and PRZM predicted bromide concentrations in the

soil solution at a 61 cman depth following the second

application................................................ 153

6.73. Measured and PRZM predicted bromide concentrations in the

soil solution at a 122 cm depth following the second

application................................................ 154

6.74. Measured and FRZM predicted bromide concentrations in the

soil solution at a 183 cm depth following the second

application................................................ 154

6.75. Measured and PRZM predicted concentrations of atrazine in

the soil solution at a 61 cm depth ..........................155

C. 1. Listing of program to calculate water and chemical

fluxes and storages........................................ .177

C.2. Program to calculate mass balance between sampling periods.. 194

D.I. Atrazine concentration in the groundwater on 1/19/87.........201


xii









D. 2.

D.3.

D.4.

D.5.

D.6.

D.7.

D.8.

D.9.

D.10.

D. 11.

D.12.

D.13.

D. 14.

D.15.

D. 16.

D.17.

D.18.

D.19.

D.20.

D.21.

D.22.

D.23.


Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine

Atrazine


concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration

concentration


in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the

in the


xiii


groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater

groundwater


1/26/87.........201

2/02/87.........202

2/09/87.........202

2/16/87.........203

2/23/87.........203

3/02/87.........204

3/09/87.........204

3/16/87.........205

3/23/87.........205

3/31/87.........206

4/06/87.........206

4/13/87.........207

4/20/87.........207

4/30/87.........208

5/01/87.........208

5/03/87.........209

5/05/87.........209

5/08/87.........210

5/13/87.........210

5/18/87.........211

5/25/87.........211

6/01/87.........212















Abstract of Dissertation Presented to the Graduate Sdchool
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

MEASUREMET AND PREDICTION OF HERBICIDE
TRANSPORT INTO SHALLOW GRNOWATER

By

Matthew Clay Smith

December 1988

Chairman: Dr. A. B. Bottcher
Major Department: Agricultural Engineering

A field study was conducted to observe the movement of the

herbicides atrazine and alachlor within the soil profile and a shallow

water table aquifer following a surface application. Surface-applied

bromide was used as a nonadsorbed tracer of water movement. The movement

of nitrate from a fertilizer application to the site was also monitored.

Measurements of chemical concentration were made using soil core samples,

samples of soil water fram the unsaturated zone, and water samples from

monitoring wells below the water table.

Atrazine was observed to move rapidly with both saturated and

unsaturated flows. Concentrations of atrazine exceeded 350 pq/L in soil

water samples at a depth of 61 can. Samples of shallow groundwater

contained atrazine residues as high as 90 jig/L. Measurable

concentrations of alachlor did not move below a depth of 45 ancm in the

soil.


xiv








Bromide and nitrate concentrations in soil water demonstrated large

variability between sampling locations. Nitrate concentrations in the

groundwater exceeded 40 mg/L after fertilization of the field.

Two root/vadose zone pesticide transport models, GLEAMS and

PRZM, were used to simulate the conditions on the site. Comparisons were

made between model simulation results and observed data. The

uncalibrated models predicted concentrations that were generally within

an order of magnitude, and often were within a factor of 2 to 3, of

observed values. Differences between the predictions of the models

appear to be due to the relative detail by which each model describes

soil properties as a function of depth.














CHAPTER 1
INTR3DUCHON


Agricultural production systems have undergone many changes since

the 1940s. The capability of post-war industry to synthesize almost

unlimited varieties of chemicals in great quantities has provided

farmers with an arsenal with which to battle insects, weeds, and

disease. These chemicals can also provide plants with required

nutrients and regulate their growth. The result of using these chemical

tools along with improved machines for planting, cultivating, and

harvesting has been a dramatic increase in the productivity of the

American farmer.

However, scme serious problems associated with the use of agricul-

tural chemicals have been identified in recent years. Many of these

chemicals are toxic to non-target organisms such as birds, fish, and the

person who is applying them. Rachel Carson (1962) brought to the

attention of the public the potential harmful effects of using these

chemicals in large quantities over large land areas. Since the

publication of her book 'Silent Spring' there has been much research

conducted to evaluate the fate of agricultural chemicals in the

environment. IXDuring the 1960s and 1970s, great effort was initiated to

determine the quantities of soil, fertilizers and pesticides which were

entering our lakes and rivers. Through these efforts methods have been










developed (though only a few have been implemented) to minimize or

eliminate agricultural impacts on surface waters.

The discovery of abandoned waste sites such as the famous Love

Canal and the discovery of the pesticide aldicarb in the shallow

groundwater on long Island in New York generated concern for the

protection of groundwater. These concerns led to analyzing samples of

groundwater for toxic wastes and pesticides.

As more groundwater samples were analyzed, it became apparent that

agricultural chemicals had somehow entered groundwater aquifers in

several areas of the country. These results were publicized, and

citizens throughout the nation expressed alarm that they were drinking

potentially toxic chemicals. The federal and state governmnnts

responded to these concerns by intensifying monitoring efforts and

reviewing data on the many agricultural chemicals to determine potential

for leaching to groundwater axnd the extent of the current problem.

Several federal agencies, along with university and private industry

scientists, began developing computer models to help explain why these

chemicals were moving beyond their target locations and to screen

chemicals for their relative mobilities in soil. A limiting factor in

the development and utilization of these models was a lack of the

detailed data needed to test and validate model predictions.

To date, data on the presence of agricultural chemicals in

groundwater have been limited primarily to sampling of municipal and

domestic wells, providing only a large scale picture of contamination.

There have been only a few intensive monitoring studies completed which

provide sufficient data for model parameter estimation. Currently, there









3

are increased efforts at specific sites to provide more detailed

monitoring of chemical movement through the soil profile and within

groundwater aquifers.

The study described here is an effort to develop data for use in

the development and testing of models which describe the fate and

transport of pesticides and nutrients used in agriculture.















CHEAPER 2
OBJECTIVES

The overall objective of this research is to develop data on the

movement of two widely used herbicides through the crop root zone,

through the vadose zone (unsaturated zone between the crop root zone and

the top of the unconfined water table aquifer), and within the shallow

water table aquifer. These data will then be compared with the

predictions of two root/vadose zone pesticide transport models.

The specific objectives are to perform the following:

1. Instrument a field site for monitoring the movement of water and

chemicals through the soil profile and a shallow, water table

aquifer.

2. Observe the fate and transport of surface-applied herbicides

(atrazine and alachlor) over time.

3. Use surface-applied bromide as a nonadsorbed tracer of water

movement.

4. Observe transport of the nitrate ccuponent of fertilizers

applied to field.

5. Omrpile the observations of chemical movement into a data base

for use in testing pesticide transport models.

6. Simulate the field conditions present during the monitored

period with two pesticide transport models and compare

to observed results.












CHAPTER 3
REVIEW OF THE LTERATURE

3.1 Evidence of Pesticide Residues in Groundwater

The U. S. Environmental Protection Agency estimates that at least 19

pesticides have been detected in groundwater in 24 states as a result of

agricultural practices (U.S. EPA, 1987a). The total number of

pesticides detected in groundwater is greater than nineteen. These

additional detections are the result of manufacturing, storage, and

loading activities which are not included in the above figures. The

number of reported cases of pesticides in groundwater is increasing.

Cohen et al. (1986) suggest that the increase is due to improvements in

the quality and quantity of studies instead of an increase in the

problem.

The insecticide/nematicide aldicarb was detected in the shallow

groundwater on Long Island, New York in 1979 (Wartenburg, 1988).

Aldicarb was used as an insecticide on potatoes. Aldicarb was detected

in 20 of 31 wells tested in 1978 and the presence was confirmed through a

repeat sampling of the wells in June, 1979. According to the U.S. EPA

(1987a), aldicarb residues have been detected in almost 2000 wells on

Long Island. Concentrations of aldicarb exceeded the state health

guideline of 7 pg/L (ppb) in nearly 50% of those wells. Concentrations

as high as 515 Ig/L were recorded in the Long Island wells as reported by

Ritter (1986). Aldicarb residues have been found in groundwater in 15

states (Cohen et al., 1986) with typical concentrations reported in the

range of 0.3 3 pg/L.










In the state of Florida, residues of the nematicide EDB have been

detected in over 10% of the public and private drinking water wells which

have been sampled (U.S. EPA, 1987a). Approximately 1,200 wells require

treatment or have been closed as sources of drinking water. Marti et al.

(1984) detected EDB in the groundwater in southwest Georgia. EDB

residues have been detected in groundwater in at least 6 other states.

In California, residues of approximately 57 pesticides have been

detected in groundwater (U.S. EPA, 1987a). Most of these detections

were related to factors other than routine use on agricultural fields.

The nematicide DBCP has been detected at levels exceeding 1 Vg/L in

approximately 60% of the 2500 drinking water wells tested.

Residues of the widely used herbicide atrazine are commonly

identified in groundwater. Ritter (1986) reports that approximately

36,000,000 kg of atrazine were used in the United States in 1982.

Atrazine has been detected in PA, IA, NE, WI, MN, and MD (U.S. EPA,

1987; COhen et al., 1986; Ritter, 1986). According to Berteau and Spath

(1986), atrazine has also been detected in concentrations of up to 2 iq/L

in California groundwater. Pionke et al. (1988) reported atrazine

concentrations in groundwater at levels up to 1.1 gq/L in an agricultural

watershed in Pennsylvania. The wells sampled by Pionke et al. (1988)

were all located in unconfined aquifers with depths to the water table

ranging from 2 to 21 m. Atrazine concentrations of 10 J/L have been

reported in a karst aquifer in Iowa (Libra et al., 1986; Kelley et al.,

1986).

Residues of the herbicide alachlor have been found in groundwater in

the states listed for atrazine, excluding WI. According to Ritter










(1986), alachlor arnd atrazine together accounted for 25% of all

pesticides sold in the U. S. in 1982. Ritter (1986) reported that in

1982, approximately 38,600,000 kg of alachlor were applied to cropland in

the United States. Kslley et al. (1986) reported aladclor concentrations

in groundwater as high as 16 gg/L in Iowa.

There are many individual cases of pesticide residue detection in

groundwater that could be cited. The reader is referred to the

proceedings of the Agricultural Impacts on Groundwater Conferences

(National Water Well Assoc., 1986, 1988) for a variety of reports

concerning the detection of pesticide residues in groundwater. A book

edited by Garner et al. (1986) is also a good source of information

related to this subject.



3.2 Factors Which Influence Pesticide Transport To Groundwater

Donigian and Rao (1986a) list five processes which affect the fate

and movement of chemicals within the soil. These processes are

transport, sorption, transformation/degradation, volatilization, and

plant uptake. Other authors, e.g., Cheng and Koskinen (1986), use

slightly varied groupings which include the processes named above. The

interactions of these processes over time and space determine the fate of

chemicals in the soil. The spatial variability of soil properties and

other factors is an important consideration when interpreting field data

and model predictions (Donigian and Rao, 1986a). This section will

describe each of the processes and same of the factors which influence or

regulate them.










3.2.1 Transport

The chemicals can be transported in at least three different phases:

adsorbed to solid materials, in solution, and as a vapor. Surface runoff

can transport chemicals in both the adsorbed and solution phases.

Erosion and transport of soil particles during rainfall and irrigation

can result in significant transport of highly-adsorbed, low-solubility

pesticides (Waudcope, 1978). For most pesticides, the majority of

transport in runoff is in the solution phase (Wauchqpe, 1978; Rao and

Davidson, 1980). Chemicals in the solution phase can be transported

within the soil profile by saturated and unsaturated water flows.

Water percolating through the soil profile can transport chemicals

in the solution phase beneath the root zone and possibly into

groundwater. The most common methods used to describe the transport of

water and solutes in porous media are based on the assumptions that

Darcian flow conditions exist, and that solute movement is controlled by

advective and dispersive processes (Freeze and Cherry, 1979; Jury, 1986a;

Wagenet, 1986; Wang and Anderson, 1982). Darcian flow conditions imply

that the flow equations are representative of some volume of soil. All

soil properties such as pore size distribution and volumetric water

content are assumed to be uniform throughout the volume of soil (Jury,

1986b).

The advective component of transport is usually based upon Richards'

equation and describes the average rate of water flow. The average rate

of solute flow is the product of the solute concentration and the average

pore-water velocity. Additional transport of solute can occur due to

mechanical mixing of water in adjoining pores during advective transport









9

and molecular diffusion of solute from pores with high concentrations to

adjoinirng pores with lower concentrations (Freeze and Cherry, 1979).

Molecular diffusion can be significant when average pore-water

velocities are low; otherwise, the processes involved in mechanical

dispersion usually dominate. Differences in pore-water velocities in the

direction of bulk flow cause a spreading out, and consequently a lowering

of peak concentrations, of the solute plume. Some of the solute will

arrive at a reference point earlier, and some will arrive later, than

would be predicted based upon the average linear flow velocity. This is

referred to as longitudinal dispersion. The tortuosity and

interconnection of pores in the soil will also cause the solute to move

perpendicularly to the bulk flow direction. This is referred to as

transverse dispersion. Freeze and Cherry (1979) and Rao et al. (1988)

provide detailed discussions of the dispersive-diffusive processes and

their mathematical formulations.

Transport predictions based upon the methods described above have

not caqpared well with field observations, e.g. Jury et al. (1986).

Numerous modifications to the general theory have been proposed in order

to help explain and reduce the differences between field observations and

theoretical predictions. Rao et al. (1980) proposed separating pore-

water in soils into two regions: inter-aggregate and intra-aggregate.

Convective-dispersive transport was assumed to be limited to the water in

the inter-aggregate region. The intra-aggregate region was assumed to

act as a source/sink of solute. Diffusion between the regions could

remove or add solute to the water in the inter-aggregate region. Jury

(1982) proposed a method of predicting the transport times of solutes to








10

specified depths which is not based on physical processes. Rather, the

transfer function theory uses measured values to develop a probability

density function of travel times and uses it to predict solute travel

times to depths greater than those that were measured. Gish (1987)

combined the ocnvective-diffusive transport theory with a stochastic

representation of water velocities to describe field-scale (average)

concentrations of bromide at various depths in the soil profile.

Camparisons between the average measured bromide concentrations and

predicted values showed good agreement.

Volatilization from soil and plant surfaces transports the chemicals

into the atmosphere and reduces the amount remaining for transport with

runoff or percolation. Diffusion of chemicals in the vapor phase can

also transport the chemical within the soil profile. Most transport

models do not consider vapor diffusion.

Many of the factors which are assumed to influence the transport of

chemicals through porous media are summarized by Jury (1986b). Some of

the soil properties influencing transport are water content, bulk

density, permeability, clay content, organic matter content, and water

retention (field capacity). Helling and Gish (1986) present results of a

simple modeling exercise to demonstrate the effect of several soil

properties on pesticide transport. Some of the environmental factors

which have been shown to influence transport are precipitation,

evapotranspiration, and temperature (Jury, 1986b).

3.2.2 Sorption

The partitioning of solutes between the liquid and solid phases

(dissolved and adsorbed) is a major factor determining the mass of










solute available for advective-dispersive transport through the soil

profile. If the chemical of interest is strongly adsorbed, then only a

small fraction will exist in the solution phase at a given time and be

available for transport. Non-adsorbed chemicals, however, will exist

entirely in the solution phase and are available for transport with

percolating water.

Many pesticides are nonpolar and adsorption occurs primarily on

organic matter surfaces. Pesticides and chemicals which are polar will

adsorb primarily to clay surfaces (Jury, 1986c). Since the majority of

pesticides are nonpolar, the adsorption onto soils has been primarily

related to the organic matter or organic carbon content of the soil. A

detailed discussion of the factors influencing pesticide adsorption onto

soils is given by Jury (1986c). The relationship between concentrations

of a pesticide in the dissolved and sorbed phases is often represented by

the Freundlich equation:

S=KCn 3.1

where S = adsorbed concentration (pq/g of soil), C = solution

concentration (Ag/mL), and K and n are empirical constants for the soil-

pesticide system. The exponent n in equation 3.1 is often assumed to be

equal to 1.0 which results in a linear relationship of the form:

S = KC 3.2
where Kd = partition coefficient (miL/g of soil).

For most uses of equation 3.2 in transport modeling, the

partitioning between the solution and adsorbed phases is assumed to be

instantaneous and reversible. Errors associated with the assumptions of








12

linearity, instantaneos equilibrium, and reversibility are discussed by

Rao and Davidscn (1980).

The partition coefficient, Kd, is unique to a given pesticide-soil

combination. However, Rao and Davidson (1980) report that investigators

have shown that when Kd is normalized for the organic carbon content of

the soil, the resulting value is independent of the soil type and can be

considered a property of the pesticide. The normalized partition

coefficient is represented as Koc and is defined by:

Kc = Kd 100 / %OC 3.3

where %OC is the percent organic carbon content of the soil and Kd is the

measured partition coefficient. Methods for measurement of the

partition coefficient are reviewed by Rao and Davidson (1980).



3.2.3 Volatilization

Volatilization of pesticides from the soil and plant surfaces

reduces the amount of the pesticide available to be leached or

transported with runoff. Volatilization also determines the quantity of

the pesticide which exists in the vapor phase within the soil and thus

the amount available for diffusive vapor transport. Thus, as noted by

Donigian and Rao (1986a), volatilization affects both the fate and

transport of a pesticide.

Same of the factors which influence volatilization are summarized by

Jury and Valentine (1986). They are Henry's constant KH, chemical

concentration, adsorption site density, temperature, water content, wind

speed, and water evaporation. Henry's constant, KH is the ratio of

saturated vapor density to solubility and is an index of the partitioning










between the vapor and solution phases. A larger KH implies increased

volatilization. Increasing concentrations of the pesticide will increase

the volatilization as long as the vapor density is not saturated.

Adsorption of the pesticide will reduce volatilization. Volatilization

increases as temperature increases. As the soil water content decreases,

the rate of volatilization increases. Increasing wind speed can increase

the volatilization of pesticides, particularly those with low KH values.

Evaporation of water from the soil surface can transport pesticides from

within the soil profile to the surface where temperature and wind

effects can increase the volatilization. Each of these factors is

discussed in more detail by Jury (1986d).



3.2.4 Transformation/degradation

Transformation of a pesticide is the change in structure or

composition of the original compound and degradation is the breakdown of

the compound into smaller fragments with eventual inorganic endproducts

such as H20 and CO2 (Cheng and Koskinen, 1986). The transformation and

degradation processes represent a loss of the original (or parent)

compound, thus reducing the amount remaining in the soil for transport by

surface runoff and percolation.

In general, transformation and degradation processes occur at faster

rates on plant surfaces and in the top few centimeters of soil than in

the deeper soil zones (Donigian and Rao, 1986a). Thus, a chemical that

is foliar or surface-applied will likely degrade more rapidly than if it

were incorporated. For non-persistent pesticides (half-lives less than

15-20 days), the timing of rainfall events or irrigation is important in









14
determining the fraction of the applied mass that will be available for

transport. Events occurring shortly after application will likely result

in the largest concentrations in runoff and percolation water (Wauchope,

1979; Donigian and Rao, 1986a). The timing of rainfall and irrigation

events is less critical for persistent (half-lives in excess of 100 days)

pesticides since they will reside in the soil for longer periods

(Donigian and Rao, 1986a).

The major processes involved in transformation and degradation

include biotransformation, chemical hydrolysis, photolysis, and

oxidation-reduction (Donigian and Rao, 1986a). Factors influencing

transformation and degradation of pesticides, as summarized by Jury and

Valentine (1986), include: microbial populations, chemical concentration,

temperature, oxygen, pH, soil water content, and light. Detailed

descriptions of the many factors influencing these processes are

presented by Rao and Davidson (1980), Valentine and Schnoor (1986), and

Valentine (1986).



3.2.5 Plant processes

Plant processes relating to pesticides are very complex. The

uptake, translocation, accumulation, and transformation of pesticides by

plants affect the availability of pesticides for transport processes and

the potential exposure to pesticide residues by the consumers of the

vegetation, fruits, etc. (Donigian and Rao, 1986a). Plant processes can

serve as both a sink and a source of pesticide residues available for

transport.









15
Pesticides applied to plant foliage will likely be transformed or

degraded more rapidly than if the pesticide were applied to the soil or

incorporated. Pesticides on the foliage may also be absorbed into the

plant. Pesticides in the solution phase may be taken up by plant roots

and translocated to various parts of the plant. These processes serve as

a loss mechanism reducing the availability of the pesticide for

transport.

Pesticide residues on plant foliage can be dislodged by rainfall and

irrigation and washed onto the soil surface, resulting in an increased

mass of pesticide available for transport in runoff and percolation

(Smith et al., 1981; Donigian and Rao, 1986a). Plant residues which are

left on the soil surface or incorporated can also serve as a source of

pesticide which may be available for transport.



3.3 Predicting Pesticide Transport

There are many methods that can be used to predict the mobility of a

given pesticide. These range fram simple indices to complex research

models. Each method has value when used for the purpose for which it was

designed and with recognition of the assumptions and limitations

associated with it.



3.3.1 Indices of contamination potential

Rao et al. (1985) compared several methods for computing indices of

the contamination potential of pesticides. Most of the indices are based

on one or more chemical properties such as Koc and solubility. Three

indices also included distance to groundwater and recharge rate. The








16
attenuation factor (AF) proposed by Rao et al. (1985) can also be used to

estimate the mass of pesticide which will leach below the root zone or

out of the vadose zone and into groundwater. The AF indr1ex is the only

one of the irxndices compared by Rao et al. (1985) which incorporated mass

loadings.

Dean et al. (1984) developed a methodology in which soil

characteristics, crop type, chemical properties, and management practices

(e.g. tillage type) can be combined to find a cumulative frequency

distribution of the percentage of applied pesticide mass leadching below

the crop root zone. These frequency distributions were derived from

hundreds of 25-year simulations of pesticide leaching using the FpRZM

model (Carsel et al., 1984). The methodology was applied to four crop

types (corn, soybean, cotton, and wheat) and 19 representative growing

regions. The results are somewhat unique in that the user is presented

with a statistical probability of leaching based on a 25-year

meteorologic record.

Another index which is currently receiving much attention is the

DRASFTIC index developed by Aller et al. (1985). This irxndex does not

consider chemical properties; instead, it is an index of the

vulnerability of groundwater at a given location to contamination from

surface-applied chemicals. The DRASTIC index assigns numerical scores or

weights to seven factors which could influence pollution potential. The

factors are depth to water, net recharge, aquifer media, soil media,

topography, impact of vadose zone, and conductivity of the aquifer. The

final DRASTIC score is used to describe an area as having high, medium,

or low susceptibility to groundwater pollution.










ERASTIC has been used to map every county in the United States

(Alexander and Little, 1986) as part of the first stage of a national

survey of pesticide contamination of drinking water (U.S. EPA, 1987b).

In the first county-level assessment the Southeast Coastal Plain, which

covers all of Florida and southern Georgia, was shown to be the

groundwater region of highest vulnerability.

Indices like those mentioned above are beneficial to regulatory

agencies as a method to screen great numbers of compounds and locate

vulnerable areas. This information can aid in allocating funding for

further studies concentrating on the chemicals of interest in regions

which may be most vulnerable. These studies may involve detailed

modeling of pesticide transport to groundwater, modeling of surface and

groundwater hydrology, and collection of data from test locations.



3.3.2 Transport models

There are many models of pesticide transport reported in the

literature. These range from steady-state screening models, e.g. PESTAN

(Enfield et al., 1982), to very complex process-oriented research models,

e.g. LEACHMP (Wagenet and Hutson, 1986). Donigian and Rao (1986a), and

Shoemaker and Magette (1987) review a number of the models which are

available. Most models which are capable of simulating pesticide fate

and transport do not explicitly represent agricultural management and

cropping practice effects on runoff and leaching. Modifications of

management practices will be one method of reducing pesticide leaching to

groundwater. The models used for regulatory purposes should reflect the

trade-offs between surface water and groundwater quality. There is scme










evidence that no-till farming will increase pesticide transport to

groundwater while reducing surface runoff and thereby reduce pollution

of surface waters (Dick et al., 1986). Thus, solutions to one problem

may exacerbate another problem. The models discussed below are capable

of simulating, to sane degree, agricultural management practice effects

on pesticide fate and transport.

According to Donigian and Rao (1986a), SESOIL (The Seasonal Soil

Caqpartment model) is designed for long-term simulation of pesticide fate

in the soil environment. It was developed for the EPA and is used as a

screening model. SESOIL can simulate many ccnponents of the hydrologic

cycle including precipitation, evapotranspiration (ET) and surface

runoff. The model considers transport within the unsaturated zone

extending from the soil surface to the top of the saturated zone. The

hydrologic responses are determined using physically based equations in

which uncertainty has been included. The water balance used in the

model is a statistical representation of the hydrologic cCnponents over a

"season." A season is the time step of the model, e.g., month or year.

Erosion is simulated using the Universal Soil Loss Equation (Wischmeier

and Smith, 1978). Hetrick and Travis (1988) coupled SESOIL with EROS (a

submodel of the CREAMS (Knisel, 1980) watershed model) to predict surface

runoff and sediment yield from small watersheds. The pesticide ccanponent

of the model considers most of the processes and factors influencing

pesticide transport which were described in previous sections.

SESOIL requires calibration and represents long-term averages. Some

of the results of use and testing of SESOIL reported by Donigian and Rao

(1986a) indicate that it should not be used for short-term predictions.










The combined SESOI/EROS model of Hetrick and Travis (1988) adequately

predicted long-term (several months) average runoff and sediment yields

when tested against data from three small watersheds. However, monthly

predictions were not good, especially when most of the runoff occurred

due to large single storm events.

MUSE (Method Of Underground Solute Evaluation) was developed at

Cornell University as a management model and a training tool for students

and professionals (Steenhuis et al., 1987). MNOSE is an interactive,

menu-driven program which can run on a IBM-PC or compatible computer.

The model can read daily historical weather files or generate synthetic

rainfall, air, and soil temperature patterns. Surface runoff is computed

using the SCS curve number method. Erosion is not considered. Solute

movement in the vadose zone considers advective and dispersive flux, as

well as degradation and adsorption. MOUSE will also simulate the

movement of a pesticide in a two-dimensional unconfined aquifer.

Extensive graphics illustrate movement as it is being simulated.

PRZM (Pesticide Root Zone Model) is a field-scale hydrology and

transport model developed by the EPA (Carsel et al., 1984). PRZM is a

continuous model capable of simulating water and chemical fluxes over

many years of historical daily weather records. Runoff is predicted

based on the SCS curve number equation, and erosion is simulated using a

modification of the Universal Soil Loss Equation for daily time steps.

The model can simulate the entire vadose zone (soil surface to

groundwater). The vadose zone can be characterized by several layers

with varying properties. For calculations, the vadose zone is divided

into many cmpartments of equal depth. The model simulates crop growth








20

(leaf area and rooting depth). Water in the root zone can be removed by

percolation, evaporation, or transpiration by the plant. Percolation is

calculated based on the water-holding capacity of the soil. When the

water content in a layer exceeds field capacity (1/10 1/3 bar tension)

the excess water drains into the next lower cnpartment. There is an

option in PRZM that allows the draining of the profile to ocur over a

longer period than one day.

Pesticide processes represented include advective and dispersive

flux, sorption, degradation in soil and on plant foliage, and plant

uptake. Volatilization and transport in the vapor phase are not

considered. Application of pesticides can be partitioned between the

soil surface and plant foliage. Applications to the soil can be

incorporated by tillage. Different degradation rates can be specified

for soil and foliar pesticide residues. The degradation rate within the

soil can also be varied by soil layer.

PRZM will simulate multiple applications of one pesticide each year

for many years of continuous climatic record. Thus it can predict

temporal variations in leaching and runoff. Effects of agricultural

management and cropping practices can be simulated. PRZM allows the user

to request time series results for many variables in the model. This

feature is useful for observing processes at intermediate locations

within the vadose zone over time.

GLEAMS (Groundwater Loading Effects of Agricultural Management

Systems) is also a field-scale hydrology and transport model (Leonard et

al., 1987). It is based upon the extensively documented and applied

CREAMS model (Knisel, 1980). CREAMS is a nonpoint source model for









21
predicting sediment, nutrient, and pesticide losses with surface runoff

from agricultural management systems. GIEAMS builds upon the foundations

in CREAMS by adding components to simulate movement of water and

chemicals within the crop root zone. Like CREAMS, GLEAMS is a

continuous, daily simulation model.

GLEAMS predicts runoff using the SCS curve number method. Erosion

is predicted based upon modifications to the Universal Soil Loss

Equation. Surface runoff and eroded sediment, with chemicals in both the

dissolved and sorbed phases, can be routed overland, in channels, and

through impoundents.

GLEAMS divides the crop root zone into seven layers. The vadose

zone between the root zone and the water table is not considered. The

first layer has a thickness of 1 an. This layer is assumed to be the

portion of the soil that determines the mass of pesticide available for

extraction into surface runoff. The thickness of the second layer is

1/6 of the root zone depth minus 1 cm for the first layer. The remaining

5 layers are each 1/6 of the root zone in thickness. This layering

structure is fixed by the program. A strong correlation has been

observed between pesticide concentrations in runoff and the concentration

of pesticides in the top 1 cm of the soil profile (leonard, 1988). Thus

the authors of CRFAMS and GLEAMS believe that a 1 cm active zone at the

soil surface is required to maintain sensitivity of runoff concentrations

to soil concentrations (leonard and Knisel, 1987). Soil profile

properties can vary with depth. The model weights the input values to

establish average properties for each layer in the model. Percolation

through the profile is based on the water-holding capacity of the soil as










in PRZM. Water in excess of field capacity drains to the next lower

layer.

Pesticide transport within the root zone is by advection. No

dispersive flux ccponents are included. Volatilization is not

considered. Pesticide applications can be partitioned between plant

foliage and the soil surface. Different degradation rates can be

specified for chemicals on the foliage and within the soil. Soil-applied

chemicals can be incorporated to a specified depth.

The model will simulate up to ten chemicals simultaneously. This

feature makes possible the observation of the effect of changes in

chemical properties, e.g. partition coefficient, on resulting leaching

losses with one model run. The model can also simulate the formation,

fate, and transport of degradation products of the parent chemicals.



3.3.3 Coupled saturated/unsaturated zone transport models

The root/vadose zone models described above are important tools for

assessing the mobility of pesticides within the unsaturated zone. The

concern expressed by the public and governmental officials relate to

pesticide residues in grourndkwater supplies used for drinking water. Thus

the unsaturated zone models need to be linked to saturated zone transport

models so that the concentrations and transport of pesticides within

aquifers can be calculated.

As described above, MNUSE (Steenhuis et al., 1987) can predict the

transport of pesticides in a vertical, two-dimensional cross section of

a water table aquifer. Jones et al. (1987) coupled the PRZM model to a

two-dimensional saturated zone transport model to predict the movement of









23
aldicarb residues within a shallow water table aquifer. Dean and Carsel

(1988) reported on progress in linking PRZM to a two-dimensional

saturated transport model. The linked model will use PRZM for simulating

the root zone, a one-dimensional transport model for the vadose zone, and

a two-dimensional saturated zone model which can simulate confined,

unconfined, and leaky confined aquifers. The saturated zone model will

also simulate pumping from the aquifer(s). The linked model is expected

to be released for testing during the fall of 1988.



3.3.4 Model testing and uncertainty

Methods (such as described above) to predict the transport of the

chemicals used by agriculture are needed by regulators in order to assess

the impacts on surface water and groundwater quality from the use of

these chemicals. Field testing of the contamination potential of all

chemicals registered for use by agriculture would be infeasible due to

both cost and time constraints. Since decisions regarding chemical use

will be made based upon the results of transport models, it is important

to understand the errors and uncertainty associated with the models.

Each model or method will have errors associated with the assumptions

relating to the description of the system, choice of equations to

represent the system, and input parameters required by the models. There

are some general points that can be made regarding the uncertainties

inherent in deterministic, physically based models such as PWZM and

GIEAMS which were described previously. An excellent discussion of the

uncertainty in transport models is presented by leonard and Knisel

(1988).










Limitations of the conceptual formulation of the system that is

described by a model lead to uncertainty in the results. Most

deterministic models assume that same of the physical properties of the

system are uniform with respect to location. That is, soil properties

such as porosity, water-holding capacity, and organic matter content are

constant across the field, although most models will allow these types of

properties to vary with depth in the soil profile. Furthermore,

variations in pore sizes within the soil, which can cause uneven water

velocity distributions, are usually not considered. Each model will

represent a system (ie. the crop root zone) in varying levels of detail.

The equations which are used to represent the fate and transport

processes have limitations and uncertainties. For example, PRZM and

GLEAMS both use the SCS curve number method for partitioning rainfall

between runoff and infiltration. This method utilizes daily rainfall

records and does not consider the effects of rainfall intensity on the

runoff process. Testing of the CREAMS model (Knisel, 1980), frcm which

GLEAMS was derived, showed that the daily hydrology option represented

average annual runoff volumes well, but did not do as well with daily and

monthly runoff volumes (Smith and Williams, 1980). PRZM and GLEAMS both

utilize modified versions of the USLE (Wischmeier and Smith, 1978) for

prediction of daily soil erosion. The USLE was developed for long-term

(20-yr) average annual erosion rates. Although these modifications have

been tested, the basis of these methods is the USLE which is an empirical

equation derived for long-term average predictions. Many transport

models utilize the linear form of the Freundlich equation (Equation 3.1)

to describe the partitioning of a chemical between the dissolved and









25
adsorbed phases. Rao and Davidson (1980) indicate that the assumption of

linearity can lead to errors on the order of a factor of two to three.

The previous examples illustrate that the mathematical representations of

processes used in models often have inherent inaccuracies and

uncertainties.

The parameters which are required by the models to solve the

equations which describe the system also have inherent uncertainties

associated with them. Jury, (1985), in a review of the literature,

reported coefficients of variation (CV) from field measurements of

several soil physical properties. Soil porosity was found to have a mean

CV of 10%. Bulk density measurement had a CV of 9%. Determinations of

particle size fractions had an average CV of 28%, with a range of 3 to

55%. The water-holding capacity of the soil at matric potentials of 0.1

and 15 bars had a CV of 15 and 25%, respectively. Measurements of

saturated hydraulic conductivity and infiltration rates showed a CV of

124 and 72%, respectively. In a study by Jury et al. (1986), the

partition coefficient, Kd, of napropamide measured on 36 samples from a

1.44-ha field had a CV of 31%. Rao and Davidson (1980) reported that the

normalized partition coefficient, Koc, reported from individual studies

generally had a CV of 40-60%. The variability of the properties reported

above are not necessarily due to the fact the samples were taken over

large areas. Jury (1985) reported that one author had reported that up

to 50% of the variation in a parameter over a field occurred within a 1

m2 area.

Given the uncertainties associated with models and the parameters

required by the models as outlined briefly above, how closely must a










model match field observations in order to be considered validated

(determined to accurately represent the system)? Leonard and Knisel

(1988) indicate that there are no standard criteria for model validation.

Since the spatial distribution of variations in soil properties is

usually unknown, it would be unreasonable to expect any model to exactly

match point measurements from a field study. Hedden (1986) reported that

participants of the Predictive Exposure Assessment Workshop sponsored by

the U. S. EPA in Atlanta, GA, on April 27-29, 1982, agreed on two

criteria for model acceptance. For screening applications of the model

(limited site-specific data and not calibrated to previous data from the

site), the model should be able to replicate measured field data within

an order of magnitude. For site-specific applications (parameters

measured on-site and with the model calibrated to the site), the model

should be able to match the field observations within a factor of two.

The screening level criteria seems to be quite reasonable when all of the

sources of error and uncertainty are considered. The site-specific

criteria, however, may be difficult to meet by even the best models using

carefully measured site-specific parameters.

The fact that models can not be expected to exactly replicate field

measured values does not mean that they have no value. Models that have

been verified (ie. their response to changes in parameters reflect what

would be expected based on knowledge of the system) can be used with some

confidence to evaluate the differences in predicted leaching of different

chemicals or, perhaps, differences in leaching of a single chemical due

to different climatic or management conditions.










3.4 Field Studies of Pesticide Transport

Field-scale studies of pesticide runoff and leaching have, until

recently, been very limited. Studies of sufficient detail to be used for

the validation and testing of transport models have been limited due to

three factors, according to Donigian and Rao (1986b):

1. Most field studies have focused on either chemical leaching or

surface runoff from watersheds; few studies have examined both.

2. Since sampling and analysis costs associated with pesticide

leaching are high, most studies are of short duration.

3. Most studies were designed for purposes other than developing

data sets for model testing and therefore are incomplete with

regard to measurement of all the required model input parameters.

In this section, four field studies are reviewed. Methods of

sampling and monitoring will be emphasized. Some of the studies were

conducted specifically for model validation and testing purposes. A

review of the Big Spring Basin study is included due to the unique

conditions in the study area.



3.4.1 Aldicarb in Florida citrus Qroves

A three-year field study of the movement and degradation of

aldicarb in both the unsaturated zone and a shallow water table was

conducted at two locations in Florida (Hornsby et al., 1983; Jones et

al., 1987). The two locations are referred to as the Oviedo site and the

Lake Hamilton site. At both locations, aldicarb was applied to bedded

citrus trees located on coarse-textured soils. Scme of the soils at the

Oviedo site had a thick organic layer overlying the coarse textured









28
subsoil. The Oviedo site is in the flatwoods area and is poorly drained.

The Lake Hamilton site was located on a sand ridge. Drainage of the

upper soil layers was very rapid. The Oviedo site encrmpassed 3.6 ha of

treated area and the other site encmpassed approximately 1.7 ha.

Aldicarb was banded along each side of the trees and incorporated to

a depth of 5 cm. At the Oviedo site, soil samples were collected in 30

cm increments to a depth of 150 cm. At the lake Hamilton site, soil

samples were collected in 30 cm increments to a depth of 60 cm, and then

in 60 am increments to a depth of 300 cman. Soil samples were collected

using a bucket auger and were transferred into plastic bags for storage

at -20C until analyzed. The soil samples were collected monthly for six

months following the first application of aldicarb in 1983. Urndisturbed

soil cores and bulk soil samples were also collected from untreated areas

at each site for laboratory determination of soil and chemical

properties.

At each site, two clusters of observation wells were installed

within the treated area, and additional well clusters were located

upgradient and downgradient of the treated area. The wells were screened

at depths of 2, 3, 5, and 6 m at the Oviedo site, and at depths ranging

from 5 to 15 m, depending on location, at the Lake Hamilton site. Water

samples were usually collected with a peristaltic pump. At least 5 well

volumes were pumped prior to collecting a sample for analysis. When

flowrates into the well were too slow to permit pumping of 5 well

volumes, the wells were evacuated and the water entering the wells was

pumped for an additional 5 minutes prior to collection of the sample.

The groundwater temperature, pH, and conductivity were measured during










sampling. Water samples were collected monthly throughout the study.

The storage conditions prior to analysis were not specified.

Results from the first 6 months of the study were reported by

Hornsby et al. (1983). Results are reported in terms of total toxic

residues (TTR), which is the sum of residues of aldicarb, aldicarb

sulfoxide, and aldicarb sulfone. Fifteen days after application, TTR

were not detected below 30 ncm at the Oviedo site, but were detected in

all replicates at the Lake Hamilton site to a depth of 120 cm. After 120

days, TMR were detected at the deepest sampling depths (150 cm at Oviedo,

and 300 can at Lake Hamilton) at both sites. No TmR were observed at the

Oviedo site in any of the observation wells during the first six months.

Two wells within the treated area at the Lake Hamilton site showed

measurable levels of TMR. In one well, TmR levels as high as 1.2 mg/L

were measured approximately 130 days after application. Data relating to

rainfall or irrigation during the study were not presented.

Jones et al. (1983) used the data franom the aldicarb study above to

evaluate three pesticide transport models. One of the models tested was

PRZM (Carsel et al., 1984). Agreement between PRZM predicted movement of

aldicarb and the observed data was good. A sensitivity analysis was

performed to demonstrate the response of FPRZM to variations in several

soil and pesticide properties. Jones et al. (1987) presented results

from the Lake Hamilton site in which water samples from the shallow water

table were collected franom a network of 174 wells over a period of three

years. These data were used to evaluate predictions from a linkage of

the PRZM model to a one- or two-dimensional saturated transport model.

The linked model was developed to allow predictions of the extent of









30
lateral transport of TIR which have entered a shallow water table. FPRZM

was used to predict daily mass loadings of aldicarb to the top of the

water table. Results of the saturated zone modeling were in reasonably

good agreement with the observed concentrations in the wells and the rate

of lateral movement.



3.4.2 COmparison of tillagce effects in Maryland

Paired watersheds were established in 1984 on the Eastern Shore of

Maryland to observe the impacts of tillage practices on agricultural

chemical leaching and runoff. The experimental design and early results

have been reported by Brinsfield et al. (1987, 1988). One of the

objectives of the experiments was to generate a data set for testing and

validation of pesticide transport models. One watershed was used to grow

corn using conventional tillage (CT) practices, and on the other

watershed corn was grown using no-till (NT) methods. The CT watershed is

approximately 6 ha in size, and the NT watershed covers approximately 8.9

ha. The soils in both watersheds are silty, well-drained, and nearly

level.

Soil samples were collected at 5 locations in each watershed 4 times

per year. The samples were collected using a hand auger in 15 cm

increments to a depth of 120 m.

Monitoring wells made of 6.35 cm diameter PVC were installed to

depths of 3-4 m on 30 m centers in each field. The wells were arranged

to divide each field into four quadrants. The wells were installed so

that only the top of the water table would be sampled. A bentonite-soil

mixture was used to seal the top 0.5 m of each well. Well samples were










collected monthly. The wells were pumped for 5 minutes prior to

collecting a sample for analysis.

To monitor surface runoff, H-flumes with automatic samplers and

recorders were installed in the outlets of the watersheds. All samples

were frozen for transport and storage prior to analysis. Details of the

analytical procedures used for pesticide residue analysis are presented

by Brinsfield et al. (1987).

Gravity-fed lysimeters were installed horizontally through the walls

of a reinforced pit to collect samples of percolating water. The

lysimeters were installed at a depth of 1 m below the soil surface.

Three lysimeters were installed in each pit. These lysimeters collect

samples of saturated flow percolating through the root zone. Samples

were collected from these lysimeters after each rainfall event.

Herbicide treatments were the same for both watersheds. Atrazine,

metolachlor, and simazine were applied pre-emergence at the rate of 1.68

kg AI (active ingredient)/ha. Cyanazine was applied at 2.24 kg AI/ha,

and carbofuran at 1.12 kg AI/ha, both pre-emergence. Dicamba was

applied post-emergence at a rate of 0.55 kg AI/ha.

leachate collected in the lysimeters during January, 1985, had

atrazine concentrations as high as 2 Vg/L. After planting in 1985, no

leachate samples were collected in the CT watershed until September.

Atrazine and metolachlor concentrations in those samples ranged from 1-2

gg/L. Simazine was detected in only one sample, and cyanazine was not

detected. In the NT watershed, samples collected in July had

concentrations of atrazine at 8-10 &g/L, simazine at 7-10 jg/L, cyanazine

at less than 1 jig/L, and metolachlor at less than 2 jg/L. Leachate from









32
the NT watershed in the fall of 1985 had similar concentrations to those

collected from the CT watershed.

Groundwater samples collected 15 days after application in 1984

showed levels of atrazine, simazine, and cyanazine exceeding 1 Jg/L. The

concentration of atrazine in one well reached 7 A/L on this date.

During late summer and fall, atrazine was the only pesticide detected in

the wells. Pesticide concentrations in the groundwater were similar for

the two watersheds, although the concentrations were somewhat higher

during the growing season in the NT watershed. In April, 1986, (prior to

application) atrazine concentrations of 4 mg/L were observed in the water

table on both watersheds. This indicates that atrazine may be persistent

in the lower soil zones and in the water table.



3.4.3 Aldicarb movement in the Dougherty Plain of Georgia

The U. S. Environental Protection Agency and the U. S. Geological

Survey began a study in 1982 to observe the movement of aldicarb and its

residues in the unsaturated and saturated zones at a field site in Lee

County, in southwestern Georgia. Aldicarb is used as an insecticide and

nematicide on peanuts grown on the site. One of the primary objectives

was to develop an extensive database of soil, land use, geologic, and

pesticide data for use in the validation and testing of the PRZM (Carsel

et al., 1984) model. Different aspects of this study have been reported

by a number of investigators (e.g. Cooper, 1986; Rao et al., 1986; Hook,

1987; Hedden, 1986; Smith and Carsel, 1986).

The study site covers an area of approximately 4.5 ha. Four soils

are mapped in the field. All have a fine-loamy texture. Below a depth








33

of approximately 1 m there are layers of clay separating zones of coarser

sand and gravel. The conductivity of the clay layers is very low, which

suggests that perched water tables may form on these layers and cause

lateral transport of leached pesticides. There is a shallow water table

at the site which can vary in position by as much a 6 m during a year.

Soil samples were collected from the site to characterize soil

properties and pesticide sorption and decay characteristics. Geologic

information was collected from holes bored to the top of the Ocala

Limestone. The residuum was approximately 13 m thick. The Ocala

Limestone is a part of the Floridan aquifer system. Four wells were

cased into the Floridan aquifer for measurement of water levels.

A weather station was installed at the site to collect meteorological

data.

Statistical procedures were used to determine that 20 monitoring

sites would accurately reflect the fate and transport of aldicarb at this

site. The locations of the monitoring sites were randomly selected and

distributed between the three major soil series based upon the relative

area in each series.

Each monitoring site was equipped with an impressive array of

instruments for monitoring of conditions anid collecting samples. Five

tensiometers were installed to a depth of 150 ancm for monitoring soil

water content. The tensicmeters were monitored and serviced three times

per week. Five thermistors were installed to a depth of 115 cm for

measurement of soil temperatures. The thermistors were monitored three

times per week during the middle of the afternoon in order to estimate

maximum soil temperatures. Three soil solution samplers were installed









34
at depths of 1.5, 2.1, and 2.7 m. The samplers were made using stainless

steel for the body, a high-flow ceramic cup, and teflon tubing for sample

collection and applying the vacuum. A silica flour slurry was poured

around each sampler to insure good contact with the surrounding soil.

The vacuum and sample lines were buried to avoid interference during

field operations. Vacuum was applied to the samplers for a period of 24

hours prior to collection of the sample.

At 15 of the monitoring sites, permanent stainless steel monitoring

wells were drilled to a depth of approximately 4.6 m. A stainless steel

screen 0.6 m in length was placed at the bottom of each well. A gravel

pack was placed around the screen and a cement grout was placed from the

gravel to the top of the finished well. The wells were finished so that

the top of the well was approximately 0.6 m below the soil surface so as

not to interfere with cropping operations. After field operations are

complete, the wells are extended above the soil surface. Water samples

are collected using a dedicated teflon tube (one per well) and a

peristaltic pump. The 250 mL glass sample containers were connected in

line between the well and the pump to prevent cross contamination. The

wells were typically pumped for 5-7 minutes prior to collecting a sample.

Soil samples were collected using a hand auger in 15 cm incremnts

to a depth of 120 ancm. After each sample was collected, gravel and plant

material was removed and then the sample was sealed in a metal container.

The references which were obtained for this review did not report

any results of pesticide monitoring from the field site. The detailed

monitoring network and experimental design used at this site should

produce a valuable data set for model validation and testing.












3.4.4 Big Sprind Basin in Iowa.

The Big Spring Basin is a 267 km2 groundwater basin located in

northeastern Iowa. The basin is entirely agricultural with

approximately 60% of the land area in row crops. The basin has been

extensively monitored and characterized, and results from these studies

have been presented by a number of investigators (Libra et al., 1986;

Kelley et al., 1986; Hallberg, 1986; Libra et al., 1987). Over 85% of

the basin's groundwater is discharged through Big Spring. Thus,

investigators have an easily accessible point for monitoring average

groundwater quality in the basin as affected by agricultural activities.

Big Spring is a karst spring with an estimated 10% of flow contributed by

directed recharge through sinkholes. By using hydrograph separation

techniques, the investigators can determine the water quality effects of

both normal infiltration and direct recharge through the sinkholes.

Four years of monitoring have shown flow-weighted mean

concentrations of NO3-N ranging from 7-11 mg/L. Combined losses of N03-N

in both surface water and groundwater amount to 33-55% of the average

annual fertilizer nitrogen applied in the basin. Historic records

indicate that the magnitude of N03-N concentrations have increased by

200-300% over the last 20 years. This corresponds to a 200-300% increase

in the application of nitrogen fertilizers during this period.

The herbicide atrazine is the only pesticide that is always present

in the discharge from Big Spring. Flow-weighted mean concentrations of

atrazine are less than 1 g/L, but the average concentrations have

steadily increased during four years of monitoring. Total losses of










atrazine in the groundwater amount to less than 0.1% of the annual

application. Pesticide concentrations in surface waters are generally an

order of magnitude higher than observed in the discharge from Big Spring.

Peak concentrations of pesticides occur following recharge events in the

spring after crops have been planted.



3.5 Sumr

There is a growing body of evidence that pesticide residues are

being leached to groundwater in some areas of the country. The U. S. EPA

recognizes that at least 19 pesticides have been found in groundwater in

24 states as a result of agricultural practices (U. S. EPA, 1987a). Many

of the pesticide detections in groundwater have been from the use of

nematicides, which in general need to be very soluble in order to be

effective. The other major class of compounds found in groundwater from

agricultural practices is herbicides. The herbicides atrazine and

alachlor are used over large areas in many regions of the country, and

have been frequently detected in groundwater. Nitrates from

agriculturally applied fertilizers have also been found in groundwater.

The health risks associated with the low concentrations (typically less

than 10 pg/L) of pesticides in drinking water are still unknown. There

are documented health risks, however, to infants drinking water with high

levels of nitrates. The agricultural production systems in this country

will continue to require large irputs of chemical fertilizers and

pesticides. Scientists and governmental regulators must identify ways to

protect groundwater supplies from contamination while allowing farmnners to

use the chemical inputs required to maintain yields. Simulation models











will play an important role in evaluating the contamination potential

from applications of pesticides and fertilizers and assess the benefits

of modifications to current agricultural management practices

(application methods, timing, chemical formulations, tillage practices,

etc.).

iThere are many factors which influence the mobility and persistence

of pesticides in the environment. Soil physical properties, chemical

properties, crop characteristics, environmental factors, and the

interactions among them determine the fate of agriculturally applied

chemicals. Many of the processes involved in the transport and

degradation of pesticides in the environment are poorly understood, and

the mathematical equations used to represent these processes are

simplifications based on experimental observations. The soil properties

and environmental factors which strongly influence the fate of a chemical

can vary significantly within small areas.

Methods to predict the fate of agriculturally applied chemicals

range from simple indices to complex mathematical research models.

Models which can represent the effects of agricultural management

practices on the transport of chemicals are being used by governmental

regulators as part of the chemical registration process. Most of the

management models developed to date are designed to represent the

processes on the soil and plant surfaces, within the crop root zone and

possibly the unsaturated zone between the root zone and the water table.

New model developments will incorporate groundwater transport models to

assess how far and at what rate a chemical that leaches into groundwater

will move. There are many sources of uncertainty and error associated










with model predictions. Thus models should not be expected to provide

exact predictions of transport and fate within a field. There have been

a limited number of studies however, which provide data of sufficient

detail to assess how well the models do represent average transport

within a field. In order to gain confidence in the use of models for

pesticide fate and transport predictions, many new studies in various

regions throughout the country are needed. To be of real value in model

testing and validation, these studies should last for several years. For

models which require calibration, a year or more of data may be needed to

calibrate the models and then several additional years of data from that

site will be needed to test the predictions of the calibrated models.














CHAPTER 4
EXPERIMENTAL METHODS

4.1 Field Site Description

The study site is located near Tifton, Georgia, in the coastal plain

physiographic region of the southeastern U.S. The geology of this area

has been described by Asmussen et al. (1986). The site is part of a farm

rented by the University of Georgia Coastal Plain Experiment Station

which is referred to as Gopher Ridge in honor of the many gopher

tortoises which make their home on the edges of the fields and in the

woods nearby.

The study site covers 0.7 ha in the northwest corner of a 2.3-ha

field. The field was developed for irrigated agricultural research.

Irrigation sprinkler risers are located on a 12 by 12 m grid. Supply

lines for irrigation water are buried approximately 1 m below the soil

surface. The field was removed from active crop research in 1982. At

that time the field was deep tilled in both north-south and east-west

directions to remove possible residual tillage effects. A permanent

bahia grass cover was established and fertilized and irrigated as needed.

No pesticides were applied after the grass cover was established. The

field was maintained in this condition until the spring of 1986 when

instrumentation for this research was installed. The site is bounded on

the north and west by trees. Just beyond the tree line on the western

edge of the field, the surface elevation drops rapidly into a seepage









40
area where subsurface flows reemerge as surface water that flows to the

Little River.

The soil on the study site is classified as a Lakeland sand (Typic

Quartzipsamrnents, thermic, coated). The soil profile depth on this site

ranges from 1.9 to 4.4 meters, Underlying the soil is a restricting

layer consisting of tight clays. This restricting layer is the top of

the Hawthorne formation which forms the confining layer over much of the

aquifer system (Floridan) of southern Georgia and Florida. The presence

of this restricting layer causes percolating rainfall or irrigation to

saturate the soil above the layer and form a transient water table.

Previous work by Asmussen et al. (1986) had shown, through the use

of ground penetrating radar (GPR), that the restricting layer below the

study site would cause saturated flow above it to converge into channels.

The presence of these channels made the site look promising for saturated

zone monitoring because samples from wells located in these channels

would be very likely to show the presence of any chemicals which were

moving with the saturated flow. The maps drawn from the (GPR)

descriptions also indicated that the slope of the restricting layer near

the study area boundaries was generally away from the study area. This

would minimize influences of saturated flows coming into the site from

other areas.

Figure 4.1 shows the elevations of both the soil surface and

restricting layer for the 0.7-ha study site. The elevations given are

relative to a local bench mark which was assigned an elevation of 30.48

m. The soil surface has a relatively uniform slope of approximately 4%

towards the west. The restricting layer shows a more complicated









Soil Surface


Elevation


3 up- ', -1, I I
0
/_, ,, o/ ,/ i l
07-09 07-10 07-1 0 7-1 07-1 /
ff 0 0 / 0 0, /
0 D 0. 0010 /(D ) G III
17 o / I j /
I8-07 0-09 05- 0 OB-11/ 08-11 06-13 0 /
I I
_oi 0 0,' 0 9-
S37 2 ( p D (/
*/ <5ii/ 'i
U/
Z 09-0o9 09 10 01, 920,3 91 / /
Q j10-8 1 10 10-111 10 12 1?-13 10414
g 0 O P 0
0I \
57 -1-07 "- \ "- /
0 a 0 G I
70 90 110 130 150 170 190
Distance (m)


Restricting


Layer Elevation


(m)


90 110 130 150 170 190


Distance


(m)


Figure 4.1. Contour maps of soil surface and restricting layer
showing locations and ID labels of monitoring wells.


(m)


-3


'17
C3

-37


57









42

topography. Slopes of the restricting layer range from 1 to 15 percent,

with the general direction of the slope also being towards the west. The

fact that the soil is a coarse sand also was considered to be

advantageous for monitoring pesticide movement. The high hydraulic

conductivities and low clay and organic matter contents should result in

increased leaching and movement of applied chemicals as compared with

tighter soils. Another advantage of this site was the availability of

irrigation which could be used to supplement the natural rainfall if

required to maintain a dynamic flow regime.

This study was not intended to represent typical agricultural

chemical usage and practices in the coastal plain of Georgia. The

primary objective was to observe chemical movement within the unsaturated

soil and in a shallow water table aquifer (shallow groundwater). The

properties of the soil and restricting layer described above were suited

for this objective.



4.2 Site Instrumentation

The movement of the applied chemicals was monitored using a

combination of soil samples, soil-water (soil solution) samples from the

unsaturated zone, and saturated zone samples. These samples were

collected using augers, soil solution samplers, and monitoring wells,

respectively. Soil sample collection required no permanent

instrumentation, and the collection procedure will be described in a

later section.

Samples of the water from unsaturated soil were collected using

soil solution samplers. An excellent review of soil solution samplers is









43
presented by Litaor (1988). Smith and Carsel (1988) presented a design

for a stainless steel solution sampler that do not react with pesticides

during sampling. Samplers constructed of stainless steel were not

economically feasible in this research.

The samplers used in this study were constructed of 4.2 cman OD,

schedule 40 PVC and a 1 bar, high flow ceramic cup, 3.99 can diameter by

19.05 cm long attached to the PVC pipe with epoxy. A 0.64 cm

polypropylene tube was extended from the inside bottanom of the ceramic cup

to a bulkhead fitting on the top of the sampler. A second fitting was

placed on the side of sampler, near the top, for attaching a vacuum line.

The samplers were constructed in two lengths, 1.08 and 2.09 m. The

polypropylene tubing used in the samplers was not tested for adsorption

of atrazine and alachlor. This was not done since the solution samplers

were to be used for collection of samples to be analyzed for inorganic

chemicals such as bromide and nitrate. The volumes of samples collected

from the solution samples were in general too small (less than 50 mL) to

permit reliable quantification of pesticide residues.

The samplers were placed such that the center of the ceramic cup

was located at depths of 61, 122, and 183 cm. The samplers were

installed by augering a 5 cm hole to a depth that was approximately 10

ancm less than the desired sampling depth. A piece of thin wall aluminum

tubing with ID slightly smaller than the ceramic cup was then driven 20

cm beyond the augered hole, removing a core of soil. The sampler was

pushed into the hole until seated on the bottan. A slurry of water and

soil taken from a borrow pit adjacent to the site was then poured into

the amnnulus between the PVC and soil. The top 12 cm of the hole was










widened to approximately 10 an in diameter and was filled with a

bentonite slurry to form a seal around the sampler and prevent direct

flow from the surface down the side of the PVC pipe to the ceramic cup.

Figure 4.1 shows the location of the application area, which is the

strip to which the herbicides and bromide tracer ware applied. The

samplers and three monitoring wells (08-09, 09-09, and 10-09) are within

this area. The samplers were installed in groups of three at a distance

of 3.05 m on either side of each well along a north-south axis. The

samplers were assigned identification labels that indicated their

position with respect to the wells and the depth to which they were

installed, eg. the group of samplers located 3.05 m south of well 09-09

were assigned labels of 09S-2, 09S-4, and 09S-6 for depths of 61, 122,

and 183 an (2, 4, and 6 ft), respectively. A cross-section of the

application area showing the locations of solution samplers and

monitoring wells is shown in Figure 4.2.

A small instrument trailer was located approximately 4 m east of the

application area. A vacuum pump located in the trailer was used to

apply a vacuum to all samplers simultaneously through a manifold. The

pump was capable of creating a vacuum of approximately 28 cm of Hg. A

0.64 an diameter polypropylene sample tube was connected to the sample

fitting on each sampler and routed back to the trailer where it was

connected to a 50 mL polycarbonate sample container. All sample tubing

was enclosed in a protective PVC conduit which was located 1.5 m from the

samplers to avoid interference with chemical application in the vicinity

of the samplers. Between the samplers and the PVC conduit, the stiff

polypropylene vacuum and sample lines were fastened together and held











<- 36.6 M

-> 3 r, 1<2- M ., >< 12.2 --- > 3
0. n Inn
0.6 m
I-

1-t~
0.6 C

0.6 - SOLUTION
0.6 M T SAMPLERS %3
_,k_ .CR -. ,2 M

'-'^ ^ CELL ^^^ '^^\

__ RESTRICTING LAYER


Figure 4.2. Cross-section of soil profile through application area
showing locations of monitoring wells and soil solution
samplers.



about 0.6 m above the soil surface to minimize interference around the

samplers. When it was time to collect the samples from the samplers,

the vacuum was switched to the sample collection side and the vacuum side

was opened to atmospheric pressure. Any water that had accumulated in

the samplers would then be pulled into the sample containers.

Monitoring wells for collection of samples frcm the saturated zone

and measurement of the water table elevation were installed during May

of 1986. The wells were constructed of 6.3 cm diameter, schedule 40 PVC

and were slotted using a veneer blade on a radial arm saw. Three rows of

slots were cut along the circumference of the well. Each slot was

approximately one-sixth as long as the pipe circumference. The slots

were spaced approximately 2.5 cm apart over the bottComn 1.22 m of the

well. Although several manufacturers offer slotted PVC well screens,










these were beyond the budgetary constraints of this project. Scmne

existing wells, installed by the USDA-ARS Southeast Watershed Research

Laboratory in conjunction with the GFPR mapping by Asmussen et al. (1986)

were also used. Conversations with the technicians involved with the

installation revealed that the wells were constructed of the same

materials as described above but were perforated with tiny drill holes

instead of being slotted. They were unable to recall the diameter of the

holes or the extent of the perforation above the bottom of the well.

Appendix A lists statistics associated with each well and indicates which

wells were installed by USDA.

The wells were installed by hand augering a 10 an diameter hole

down to the top of the restricting layer. The restricting layer was

identified by a sudden change from yellcowish-white sand to red clay

mixed with small rocks. The change was very abnrupt and the effort

required to auger through the clay was significantly greater than to

auger through the coarse sand. A PVC end cap was slipped onto the end

of the well and held in place by friction during installation of the

well. Fine gravel was placed around the wells in the zone of the slots

in an attempt to minimize sand migration into the wells through the

relatively coarse slots. Soil from a nearby borrow pit was used to fill

the hole around the well to within 12 cm of the surface. Ten centimeters

of a water and bentonite slurry was then added to form a seal around the

well. The final 2 cm was filled with soil. The USDA augered down to

the restricting layer with a hollow stemn auger on a small drill rig.

The well was slipped into the center of the auger and the auger was










withdrawn. No details on backfill procedures used by the USDA are

available.

The wells were continuous from the restricting layer to a height of

approximately 50 an above the soil surface. No solvent weld joints or

connections were used. Each well was fitted with a 0.64 ncm diameter

polypropylene tube which extended from approximately 2.5 an above the

bottom of the well through a #2 rubber stopper placed in a vented PVC

cap. The polypropylene tubes were not tested for adsorption of atrazine

and alachlor. The caps were vented by milling small channels into

opposite sides of the inside of the cap. This was done to prevent a

vacuum from forming in the well during sampling.

The wells were installed on a 12 by 12 m grid corresponding to the

locations of existing sprinkler risers. The wells were offset by

approximately 50 cm to the west of the sprinkler risers to avoid hitting

supply lines beneath the soil surface. In the application area the

wells and solution samplers were offset from the sprinkler risers by

about 3.05 m so that the risers and supply lines would not interfere

with the downward movement of chemicals within the vicinity of the wells

and samplers. The monitoring wells were assigned ID labels to indicate

their position relative to the sprinkler riser in the SE corner of the

2.7 ha field on which the experimental site is located. Figure 4.1 shows

the locations of the wells and the associated ID labels. Well 08-15 is

located at the edge of the tree line bordering this side of the study

site. One well is located in the woods approximately 30 m west northwest

of well 08-15. This area is a seepage zone where subsurface flows











reemerge as surface water and flow to the Little River. This well is

referred to as "I1'".

Additional instrumentation installed at the site consists of a

weighing rain gage which was located near well 09-12, a water table

recorder which was located near well 08-12, tensicneters for measuring

the water content of the unsaturated soil, and a "deep" well located

between wells 08-13 and 09-13.

The rain gage is a US Weather Bureau standard weighing bucket rain

gage with a seven day chart and 30.5 ancm capacity. The rain gage was

calibrated using standard calibration weights after installation.

The water table recorder was installed to provide a continuous

record of the water table elevation to supplement the weekly

measurements taken on all wells. The recorder was equipped with a seven

day clock and a drum type chart recorder.

Eight tensicmeters were installed around each set of solution

samplers. The tensicmeters were water filled and connected with very

small tubing to a mercury manometer board. The tensicmeters were

installed at depths of 30, 60, 90, 110, 122, 140, and 183 cmn. There were

two tensiameters located at the 60 ancm depth. The tubing connecting the

tensiumeters to the manometer boards was routed above the soil surface

with the vacuum and sample lines as noted above.

The "deep" well was installed by drilling a 10 ancm diameter hole

inside of a 11.4 ancm diameter PVC casing. The casing was pushed down as

the auger advanced. The casing was installed in this manner using 1.5 m

sections to a depth of approximately 12.2 m below soil surface. The

last 3 m of the well was located in a saturated formation that liquified









49
during drilling. The observation well was made of a 1.5 m section of

6.3 cm diameter commercial PVC well screening solvent-welded to two 6.1 m

sections of schedule 40 PVC. This was jetted down inside of the casing

until it bottomed cut at the bottom of the previously augered hole.

After placement of the well, a mixture of bentonite and water was poured

into the area between the casing and well up to the top of the casing.

The purpose of this well was to observe the piezcmetric head difference

across the restricting layer.



4.3 Chemical Applications

The first application of chemicals to the experimental site

occurred on 11/12/86. On the day preceding this first application, the

entire field site was rotary mowed. The grass within the application

area was cut to a height of approximately 5 cm. All dead grass and

clippings were raked up and removed from the area prior to application.

The application area consists of a strip 36.6 m long by 9.14 m wide as

shown in Figure 4.1.

Atrazine and alachlor were applied simultaneously on November 12.

Atrazine was applied in the form of AAtrex (Ciba-Geigy Corp.) which is

an emulsifiable concentrate containing 0.479 kg AI (active ingredient)/L

(4 lb/gal). Alachlor was applied in the form of Lasso (Monsanto

Agricultural Products Co.) which is also an emulsifiable concentrate

containing 0.479 kg AI/L. The herbicides were applied using a self-

propelled born sprayer with provisions for injecting chemicals directly

into the water stream (Sumner et al., 1987). The boom length on the

sprayer is 9.14 m.









50

The intended application rate of the herbicides was 4.5 kg AI/ha (4

lb/ac). One liter of herbicide solution was prepared by mixing 313 mL of

each of the chemicals with 374 mL of deionized water. The sprayer was

operated at a velocity of 3.05 m/min moving from south to north along the

edge of the application area. The chemical solution was injected into

the water stream of the sprayer using an injection pump calibrated to

deliver 83 ml/min. The water pressure at the inlet of the sprayer was

207 kPa. The pressure at the spray nozzles was maintained at 138 kPa.

Despite the calibration of the injection pump and measurement of sprayer

ground speed, the chemical solution ran out as the sprayer reached well

10-09 which is 6.1 m short of the intended end of the application area.

The application of the herbicides was considered complete at this point,

and the remaining portion of the application area was left untreated.

Fourteen 0.35 1 plastic cups were randomly placed within the

application area. The cups were supported upright in holders at a

height of approximately 35 cm above the soil surface. The cups were

used to collect samples of the application solution for analysis of

chemical concentrations, determination of the depth of water applied to

the area during application, and uniformity of chemical and water

application. The use of these cups and holders has been tested and

found to accurately reflect water application amounts (Stansell et al.,

1982).

Immediately after completing the application of the herbicides, the

sprinkler irrigation system was started and 5.1 cm of water was applied

over the entire study site. This irrigation was intended to wash the









51
herbicides off of the grass foliage and move them into the soil profile

to minimize volatilization losses from the soil and plant surfaces.

Bromide in the form of potassium brmanide (KBr) solution was applied

to the application area using the same sprayer on 11/17/86. There was a

light drizzle of rain during application. Five hundred grams of KBr was

dissolved in 1050 mL of deionized water and injected into the sprayer as

described above. This was equivalent to an application rate of 10 kg/ha

of bromide. The sprayer again was operated at a nozzle pressure of 138

kPa, and a ground speed of 3.05 m/min. Fourteen collectors in the same

locations as for the herbicide application were used to collect

application samples.

No further chemical applications were made prior to fertilization

of the grass on 4/16/87. The fertilizer was applied at a rate of 560

kg/ha of 5-10-15 using a tractor-nounted broadcast spreader. Fifteen

percent of the nitrogen in the fertilizer was in the form of nitrate

nitrogen (NO3-N). This is equivalent to applying 18.6 kg/ha of nitrate

(NO3).

In an effort to further characterize bromide movement within the

soil profile and determine flow velocities within the grmrdwater, an

additional bromide application and a separate chloride application were

made. On 4/27/87, KBr was applied to the application area using the

sprayer utilized during the previous application. The application

solution was made by dissolving 1.2 kg of KBr in 2.3 1 of deionized

water. The sprayer made two passes across the application area during

which it applied approximately 1.6 1 of the KBr solution. This was

equivalent to an application rate of 17 kg/ha of bromide. Fifteen









52
plastic cups were used to catch application solution samples in the same

manner as previously described. Immediately following application the

field was irrigated. Irrigations continued daily for a week in order to

raise the water table and provide plenty of infiltrating water to

transport the bromide through the soil.

Prior to the beginning of irrigations on April 27, most of the wells

on the site were dry. The water table began to rise on May 1 at which

time 10 1 of solution containing 500 g of potassium chloride (KCI) was

poured directly into well 07-09 which is near the top of the site. It

was anticipated that the high concentration of chloride in the solution

could be followed downslope with the saturated flow and allow for

determination of water table flow velocities.



4.4 Sample Collection and Storage

Samples were collected weekly (Monday) throughout the study. There

were occasional periods of more frequent sampling immediatelyy after

initial applications and during the last week of April through the

middle of May).

The flint glass containers used for well sample collection and

storage throughout this study did not have teflon lined caps. Therefore,

a small piece of parafilm (a common laboratory film used for sealing

beakers and containers to prevent evaporation and contamination) was

stretched over the mouth of the bottle prior to screwing on the caps.

Samples of the application solutions of the herbicides and bromide

were transferred to glass containers and placed in a cooler with ice









53
packs immediately after the sprayer had passed completely by the sample

collector.

Soil samples were collected in one of two ways: taking 2.5 ncm

diameter by 5 cm long cores using a soil sampling probe, or using a 5 cmn

diameter stainless steel bucket auger. The soil sampling probe was

generally used for collection of samples from the top 5 cm of the soil

profile. When this method was used, 5 to 8 samples would be collected

franom a small area and composited into a plastic sample bag. This would

represent one final sample for the soil surface at that location. For

deeper samples the auger was used. The auger was used to remove soil to

within approximately 5 cm of the desired sample depth. The auger was

then cleaned to remove traces of soil, and a sample was collected which

represented soil from a depth of 5 cm above to 5 cm below the target

depth. The sample was carefully poured from the top of bucket into a

plastic sample bag and sealed. The number of locations within the

application area sampled depended on the number of depths to be sampled

and ranged from 5 to 14. After collection, the soil samples were placed

into coolers with ice packs for transport to Gainesville. Soil samples

franom outside of the application area were collected several times during

the study to use as blanks and spikes during soil residue analysis.

Vacuum was applied to the solution samplers immediately upon

arrival at the study site (7:00 8:00 am). The vacuum would typically

be left on for 8 to 10 hours prior to collecting the soil solution

samples. The vacuum applied to the samplers was approximately 28 cm of

Hg. Near the end of the day the vacuum would be released and the samples

would be drawn directly into individual containers. The containers are










made of polycarbonate with a capacity of 50 mL. The containers had

polycarbonate screw caps for sealing. Immediately after all soil

solution samples were collected, the sample containers were placed in a

cooler for transport.

The elevation of the water table was measured using a well depth

indicator with a stainless steel probe. The depth indicator would sound

an audible alarm when the probe contacted water in the wells. The probe

was attached to a flat cable with markings in feet and inches, with the

smallest division being 1/4 in. Water depths were recorded early in the

morning after arrival.

After the water table elevations in all wells were measured, the

wells were pumped out to insure that the water to be sampled was

representative of the surrounding water in the saturated zone at that

time. When possible, the well was pumped dry. When the depth of water

in the well exceeded approximately 0.4 to 0.5 m, the pumping rate was

less than the rate of flow into the well. When this occurred, the well

was pumped long enough to insure that 2-3 well volumes had been pulled

from the well. A small, 12v battery powered, peristaltic pump was used

to pump the wells and collect samples.

Samples were collected after all wells had been pumped as described

above. A rubber stopper was attached to the intake side of the pump

through a 1 m length of 1.25 cm OD Tygon tubing. The stopper also had a

piece of Tygon tubing approximately 10 cm in length connected to it for

attachment to the sampling tube in each well. The stopper was placed in

a sample bottle and the short piece of tygon tubing was attached to the

well sample tube. Since the pump was downstream from the sample con-










trainer, the only components common to each sample collection were the

rubber stopper and the short piece of Tygon tubing. When a sample was

collected, the initial water from the well was swirled around the bottle,

and the bottle was then inverted to remove this rinse water. This served

to rinse the Tygon tubing, stopper, and bottle. Thus, cross

contamination between wells was minimized. The sample bottles used for

the collection and storage of well samples had a capacity of 250 mL.

Well samples were placed in coolers with ice packs for storage until they

could be refrigerated.

All soil samples were frozen as soon as possible after collection

and maintained at -18C until preparation for analysis. All water

samples were refrigerated after collection and maintained at 4C until

prepared for analysis.



4.5 Sample Analysis

4.5.1 Inorganic tracer analysis

The concentrations of the three inorganic chemicals of interest

which were applied during this study were quantified using ion

chromatography. A DIONEX QIC ion chrcmatograph (IC) was used for all

analyses.

A DICNEX AS4A anion separation column was used with an AG4A guard

column in place to remove organic contaminants prior to reaching the

separation column. The IC was equipped with a conductivity detector

connected to an external integrator/recorder. The IC was also equipped

with an anion micromembrane suppressor to reduce the background









56
conductivity of the eluant and thus allow for greater sensitivity to the

conductivity changes caused by the presence of anions in the sample.

There was no automatic sampling device with this instrument, so

each sample was injected by hand. Approximately 1 mIL of sample was

loaded onto a sample loop prior to injection onto the column. The

sample loop retained approximately 0.25 mL of the sample with the rest

being discarded to a waste line. Between injections, the sample loop

was flushed with 1 mL of deionized (DI) water. Prior to the guard

column, the sample was passed through two filters rated at 30 and 5

microns, respectively, to remove solids from the sample. These filters

were an integral part of the IC and the filter elements were changed

whenever the system pressure was observed to be increasing above nominal

levels.

The eluant used in these analyses was 2.2 millimolar (mM) sodium

carbonate (0.933 g Na2CO3 in 4 1 DI H20) and 0.75 mM sodium bicarbonate

(0.25 g NaHCD3 in the same 4 1 of DI water). This solution was made

fresh as needed. All references herein to deionized (DI) water refer to

water that had been passed through carbon filters and ion exchange resins

to produce DI water with a resistivity of at least 1 megohm. The

regenerant solution was 0.025 N sulfuric acid (2.8 mL of concentrated

H2SO4 in 4 1 of DI water). The regenerant solution was passed through

the micromembrane suppressor to reduce background conductivity.

In normal operation the eluant flowrate was approximately 2.0-2.5

mL/min, and regenerant flowrate was approximately 4 mL/min. The

pressure at the discharge of the pump at the stated eluant flowrate was

approximately 8,274 kPa (1200 psi).








57
A single injection could reveal the presence and concentrations (if

standards were prepared) of the following anions (in order of increasing

retention times): fluoride (F-), chloride (Cl-), bromide (Br), nitrate

(NO3-), phoshate (P043"), and sulfate (SO42-).

In ion chrImatograpy, the retention times of the various anions are

dependent upon total ionic strength of the sample, ionic strength of the

individual ions, eluant flowrate, and temperature of the sample and

eluant. Thus, the retention times could, and did, vary. When the

integrator was set to calculate the areas under the peaks, it was found

that the automatic selection of the beginning and ending points of peaks

was not consistent. A small change in retention time could

significantly alter the reported area for a given ion even when using

duplicate injections. Better results were achieved by setting the

integrator to report maximum peak height. This value was not sensitive

to variations in the starting and stopping times of a peak as long the

baseline was relatively stable.

Initial injections of water table samples from the experimental
site revealed the presence Cl- and S042- in all samples, and relatively

high concentrations of N03- in samples from well IOW. Standards were

prepared using DI H20 and oven-dried quantities of certified or primary

standard grades of potassium chloride, potassium bromide, potassium

nitrate, and potassium sulfate. The standards initially included Cl-

and S042- because of their presence in the samples. They were not used

to quantify the concentrations of either species during initial sample

analysis. When C- was introduced into the water table as a tracer late

in the experiment, sample concentrations of Cl- were quantified. It was









58
observed that the response of the chromatogram to Cl- was approximately

twice the response to the other anions of interest. Thus, standards were

prepared with Cl- at half of the concentrations of the other species so

that the responses of all species would be similar and on scale

simultaneously. Standards were prepared with concentrations of the main

species of interest ranging frcum 0.01 to 100.0 mg/L. A typical standard

would contain 0.5 mg/L Cl- and 1.0 mg/L of Br, N03-, and SO42-. It

should be noted here that all references to nitrate (N03-) indicate

concentrations of N03- and not N03-N (nitrate-nitrogen).

Samples were taken from the refrigerator and allowed to cme to

room temperature prior to injection. No sample preparation was

performed. The analysis protocol typically followed the pattern of 3

standards, 10 samples, 3 standards, etc. The standards used were

selected based upon the concentrations of the analytes being observed in

the samples. Thus, standards were selected to bracket the observed

values as closely as possible. Occasionally a sample would be reinjected

to observe the repeatability of the analysis procedures. Samples from

wells located away from the experimental site, including the deep well

which supplies irrigation water to the farm, were regularly analyzed as

field blanks (no Br- would be expected; however, C-, N03-, and S042-

could be present).

A piecewise linear fit to each sequence of injected standards was

used to generate a function of the form: concentration = f(peak height).

The piecewise fit was selected instead of linear regression because for

some species (particularly high concentrations of Cl-) a linear

regression resulted in a negative y-axis intercept. This implies that










the species can generate a noticeable positive response and yet be

calculated to have a negative concentration. It is expected that a

positive intercept would be a normal result, i.e. there is sane threshold

concentration of the species required before the detector indicates a

response. The piecewise fit insured that any positive response would

result in a positive calculated concentration. Conversations with an

organic chemist at the Pesticide Residue Laboratory of the USDA-ARS

Southeast Watershed Research Laboratory indicated that the piecewise fit

would be an acceptable method of generating a standard curve (Marti,

1987).

The standard curves generated on either end of the analysis of 10

samples were used to compute the concentrations of the species in the

samples. The resulting concentrations were then linearly weighted based

on sample position with respect to either set of standards, i.e. the

first sample following the standards would be weighted towards the curve

generated by the preceding standards and only slightly effected by the

standard curve generated 9 samples later. The sample midway between

standards would be evenly weighted between the two standard curves. In

general, there was little drift in the instrument and the standard

curves did not vary much over a period of hours.



4.5.2 Herbicide residue extraction

Analysis of samples for residues of atrazine and alachlor required

different extraction procedures for soil and water samples. Once the

extractions were completed, all samples were analyzed on a ccnmmon gas

chrcnatograph with identical operating conditions.










There are many published methods for preparing samples for

determination of atrazine and alachlor residues (e.g., Rdhde et al.,

1981; Voznakova and Tatar, 1983). In general, the procedures are rather

detailed and costly in terms of time and supplies. Dr. Willis Wheeler,

Director of the Pesticide Residue Laboratory in the Food Science and

Human Nutrition Department at the University of Florida, Gainesville,

indicated that a simpler method of sample preparation might be developed

which would be satisfactory in terms of the recovery of the pesticide of

interest while being more economical in both the time required for sample

preparation and expense (Wheeler, 1987). The more steps involved in the

extraction and preparation of a sample, the more likely it is that

something can go wrong and the residues lost. With these factors in

mind, methods for extraction of the two herbicides from soil and water

samples were developed which were in keeping with the budgetary

constraints of the project and the time constraints on the analyst.

Extraction of the herbicide residues from water samples was done

using a liquid-liquid extraction procedure. One hundred fifty

milliliters of the sample water were placed in a quart mason jar along

with 150 mL of pesticide-grade hexane. The jar was sealed with a

standard mason jar lid which had been lined with a self-adhesive teflon

film. The jar was agitated on a shaker table for thirty minutes after

which the contents of the jar were poured into a labeled 500 mL

separatory funnel. The sample water was drained through the separatory

funnel back into the jar, leaving the hexane in the funnel. Another 150

mL of hexane was added to the jar and it was again agitated for 30

minutes. The contents of the jar were again added to the funnel. The









61

water was then drained and discarded. The hexane was filtered into a 500

mL boiling flask through sodium sulfate to remove any remaining water.

Next, the boiling flask was placed on a rotary evaporator and the

reduced in volume to near dryness. A few mL of fresh hexane was then

added to the flask and swirled to wash the sides of the flask. This

hexane was placed in a 10 mL volumetric flask. A few mL of hexane was

again added to the boiling flask and the swirling repeated. After two

rinses of the flask, the hexane in the volumetric flask was brought to a

final volume of 10 mL. The finished sample was stored under

refrigeration in test tubes with teflon lined screw closures. The

initial sample volume of 150 mL and the final extracted volume of 10 mL

indicate that this procedure concentrated the residues by a factor of 15.

Blanks taken from remote wells as well as DI water were extracted as a

check on the procedure. Spiked samples of well water and DI water were

extracted to determine the recovery levels of the herbicides using the

procedure described above.

Soil samples were removed from the freezer and allowed to air dry

for one day prior to extraction. This was done to allow excess moisture

to evaporate since the acetone used in the extraction process can absorb

water which will not be removed by filtration through sodium sulfate as

was done for the hexane extractions. Acetone was chosen as the solvent

for soil extractions because it was observed that the water in the soil

samples, or sane other factor, tended to repel hexane. It did not

appear that the hexane was adequately dispersing the soil particles.

A fifty gram portion of the air-dried soil sample was weighed and

placed into the mason jar. One hundred mL of pesticide grade acetone










was then added to the soil sample. The sample was agitated for 30

minutes and the acetone was carefully decanted into a separate labeled

glass container. Another 100 mL portion of acetone was added to the soil

and the agitation was repeated for another 30 minutes. The acetone was

poured into the container with the previous portion. Several rinses of

acetone were poured through the extracted soil to remove any acetone

left from the extractions. The acetone rinses were added to the acetone

frcm the extractions and the entire volume was vacuum filtered through a

47 nmm diameter filter with a 0.45 micron pore size. The filtrate was

then reduced to near dryness on the rotary evaporator and brought to a

final volume of 10 mL in acetone. Soil samples taken from outside the

study area were extracted as field blanks and were spiked with known

quantities of the herbicides and extracted as a check on the extraction

and analysis procedures.

A subsample of approximately 10 g of each soil sample was taken to

determine the soil-water content so that the concentrations of the

herbicides in the soil could be reported on dry weight basis. Soil

samples for soil-water content determinations were dried for 24 hours in

an oven at 100C.



4.5.3. Herbicide residue analysis

All herbicide residue samples were analyzed on a Hewlett Packard

model 5840A gas chromatograph (GC). The GC was equipped with a 35

position autosampler and a model 5840 GC terminal (integrator). The GC

had both electron capture (EC) and nitrogen-phosphorus (NP) detectors.

IDuiring the early phases of the study (methods development), the electron











capture detector was used. This detector had a high sensitivity to

alachlor and a lower sensitivity to atrazine. However, when field

samples were analyzed, as opposed to DI water spiked with pesticides, it

was observed that the EC detector was sensitive to many naturally-

occurring cmpourds (the EC detector is sensitive to chlorine molecules)

which had been extracted along with the pesticides. Thus, the output was

so crowded with peaks that the quantities of atrazine and alachlor

present could no longer be reliably determined. The NP detector is very

sensitive to molecules of nitrogen and phosphorus and insensitive to

almost anything else. When the samples were analyzed on this detector,

the chromatograms were much cleaner and atrazine and alachlor were often

the only visible peaks. The NP detector is somwhere between 5 and 10

times more sensitive to atrazine than it is to alachlor due to the

greater number of nitrogen atoms in the structure of atrazine. The NP

detector was operated with detector gas flowrates of air at 50 mL/min and

hydrogen at 3 mL/min. The carrier gas was helium with a flowrate of 30

mL/min.

The column used in the GC was 1.83 m long with a 4 mm ID. The

column was packed with 3% OV-17 (50% phenyl, methyl silicone) on 100/120

mesh Gas Chrom Q. The operating conditions were: injector port

temperature set at 2400C, column temperature at 200C, and detector

temperature at 300C.

The sample protocol used with the auto sampler was usually of the

form: 4 standards, 1 solvent blank, 10 samples, 4 standards, 1 solvent

blank, 10 samples, 4 standards, 1 solvent blank. The solvent blank was

used to help prevent carryover from the last (and highest) standard into










the following sample. Carryover was normally not a problem, but the

solvent blank would often show some trace of the herbicides.

Working standards of the two herbicides were prepared using

reference standards provided by the EPA in Research Triangle Park, North

Carolina. The initial standard was made by dissolving 10 mg of the

reference into 10 mL of solvent yielding a concentration of 1 mg/mL

(1000 mg/L). This standard was then used as the basis for subsequent

working standards. The standards were prepared with the concentration

of alachlor being 5 times greater than the concentration of atrazine so

that the detector response to the chemicals would be nearly equal.

Working standards of atrazine ranged from 0.01-500.0 mg/L with the

concentrations of aladhlor generally being 5 times higher as noted above

(very high atrazine standards did not include alachlor). The 0.01 mg/L

standard of atrazine represented the smallest concentration which was

detectable without pushing the limits of machine and detector. The

autosampler injected a volume of 4.8 AL. So an injection of 4.8 AL of

0.01 mg/L atrazine represented a lowest detectable mass of 10 nanograms

of atrazine.

Whenever new standards were prepared, they were compared with the

previous standards to assure similar response and that no mistakes had

been made during the dilutions. Two sets of standards were made. One

set was made in acetone for use in analyzing the soil sample extracts.

The other set was made in hexane for use in analyzing the water sample

extracts. Comparisons between the two sets of standards revealed no

differences in response, but recommended procedures (Wheeler, 1987)

dictate that the standards be made up in the same solvent as the samples.











The integrator available with the GC did not work properly. In

order to quantify the amount of herbicides present in the samples, all

peak heights were hand measured (a peak height mode, as used with the IC

was not available on this integrator).

A linear regression was performed from the standards to give a

relationship of the form: concentration = f(peak height). There were no

problems with negative intercepts in these regressions as discussed in

the section on ion chromatography. Sample concentrations were computed

with the regression equations calculated from standards on either side of

the samples. In order to account for drift in the detector response, the

final concentrations were calculated by linearly weighting the

concentrations computed from each set of standards as discussed

previously. The NP detector had a tendency to drift and this correction

for drift was usually needed.















CHAPTER 5
MOELIG THE EXPERDfIMETAL SITE



Two models were selected for the purpose of simulating the movement

of the chemicals applied to the experimental site. The models selected

for use were the Pesticide Root Zone Model, PRZM (Carsel et al., 1984),

version 2, and Grourxdwater Loading Effects of Agricultural Management

Systems, GLEAMS (leonard et al., 1987), version 1.8.53. These models

were selected because they were designed specifically to simulate

pesticide transport through the root zone of a crop and reflect

influences of agricultural management practices such as tillage.

Reference to the manual for GLEAMS includes the documentation for the

CREAMS model (Knisel, 1980) from which GLEAMS was derived as well as the

supplementary GLEAMS user manual which is provided with the model code

and describes the differences in input data sets between CREAMS and

GLEAMS.

This chapter will discuss the selection of input values used in the

models. Only the parameters relevant to the results will be discussed.

There are parameters in both models for which values were selected, but

which had no bearing on the results to be presented. For example the

models utilize the Universal Soil Loss Equation, USLE, (Wischmeier and

Smith, 1978) for prediction of soil erosion, however, in the simulations

surface runoff was never predicted and therefore the parameters chosen

for use with the USLE did not affect the reported results. There are

66










many parameters such as rainfall, pesticide properties, and soil

characteristics which are common to both models. Thus, the selection of

common parameter values will be discussed first, followed by selection of

values for parameters which are unique to each model.



5.1 Selection of Common Input Parameter Values

5.1.1 Rainfall and irrigation

Both models use daily rainfall data. The formats of the rainfall

files for the models are different and PRZM reads in values in

centimeters while GLEAMS reads in values in hundredths of an inch.

The data used to create the precipitation files came from a

combination of two rain gages. As stated previously, one weighing rain

gage is centrally located within the experimental site. This rain gage

was installed in June of 1986. A second rain gage is located

approximately 300 m north-northeast of the experimental site. This gage

is a tipping bucket type rain gage that is connected to a data logger.

All rainfall data used in the simulations came from this off-site gage

since the data fram it was already on a computer and available. The

rainfall amounts recorded in both gages agreed well. The gage located on

the site was used to supplement the rainfall record with amounts of

irrigation applied to the site. This rainfall plus irrigation file was

then manipulated into the units and formats required by each model, and

checked to insure that the two formats contained identical information.











5.1.2 Soil properties

Both models use the Soil Conservation Service (SCS) curve number

method (USDA-SCS, 1972) to partition rainfall between runoff and

infiltration. The runoff curve number was selected from Table 9 in the

PRZM manual and was chosen based on the assumption that the bahia grass

cover could be assumed to be similar to a pasture in good condition

without contouring (Lakeland sand is classified as being in hydrologic

soil group A).

Physical descriptions of the soil profile were obtained from two

sources. Hook (1985) measured properties of the soil in another section

of the farm on which the experimental site is located. The data

collected by Hook (1985) were measured within the top 120 cm of the soil

profile. The soils on the farm are uniformly classified as either a

Lakeland sand or Bonifay sand. The primary differentiation between the

soils is the depth to the restricting layer with the Lakeland soil being

deeper than the Bonifay. Soil physical properties such as water

retention and hydraulic conductivities are considered to be the same

between these soils on this farm. The second source of data used to

determine soil properties is a soil characterization report for Florida

soils prepared by the Soil Science Department at the University of

Florida (Carlisle et al., 1978). This report contains characterization

data for a Lakeland sand.

The properties that were selected to represent the soil profile of

the experimental site are summarized in Table 5.1. The average depth to

the restricting layer for the three wells within the application area is

2.62 m and this was chosen to represent the profile depth.










Table 5.1 Soil properties used in simulations.1

Organic Bulk Hydraulic Water Content (%)
Depth Carbon Density Conductivity Effective Field Wiltng
(cm) (%) (g/cm3) (cm/hr) Saturation Capacity2 Point

0- 13 0.55 1.45 13.3 33.0 10.0 2.1

13- 20 0.28 1.60 13.3 33.0 10.0 2.1

20- 51 0.08 1.58 22.9 33.3 10.0 2.1

51-102 0.04 1.59 36.5 33.8 9.0 2.1

102-262 0.03 1.59 45.0 34.0 9.0 2.1

Data frcm Hook (1985) and Carlisle, et al., (1978).
2Field gravimetric water content following 48 hours of drainage under
plastic following 16 hours of flooding.
3Water content at 15 bars tension.



5.1.3 Pesticide chemical properties and applications

The pesticide chemical properties required by the models are

solubility, partition coefficient, degradation rates on foliage and in

the soil, and a coefficient of plant uptake. The models may require

these parameters in slightly different forms, but the basic information

required is the same. In addition to the model manuals, the manual for

LEACH (Dean et al., 1984) was a good source of estimates of the required

parameters. No exhaustive search of possible values of parameters for

each chemical was undertaken. At this point in the study, the models

were run using values that a user could obtain frcm the manuals. This

may result in less than optimum performance of the models, but probably

reflects the way in which a typical user would apply them.

Reported properties of pesticides are often widely variable. For

properties such as partition coefficient and degradation rates, the










variability is understandable and expected. However, even water

solubility was found to vary by a factor of two. The solubility of

atrazine is not available from the PRZM manual. The CFEAMS manual lists

a value of 33 mg/L. However, the Agrichemical Handbook (1983) gives a

value of 70 mg/L at 20C. The solubility of alachlor is given in the

PRZM manual as 220 mg/L at 20-25C. The CEEAMS manual reports a value of

242 mg/L at an unspecified temperature. The Farm Chemical Harxndbook

(1984) does not report a value for alachlor and the Agrichemical Handbook

(1983) gives the solubility of alachlor in water as 148 mg/L at room

temperature. The final values selected for all chemical properties are

shown in Table 5.2.

The solubilities of brmnide, nitrate, and chloride listed in Table

5.2 are considerably less than the actual solubilities. However,

solubilities of ionic species are seldom listed for reference. The

values in Table 5.2, while less than the actual solubility values, are

sufficiently high such that solubility was not a limiting factor in the

mobility of the chemicals.



Table 5.2 Chemical properties used in simulations.
Property Atrazine Alachlor Bromide Nitrate Chloride

Solubility 33 148 800 800 800
(m/L)

Partition coef., 163 268 0 0 0
Koc (an3/g)

Half-life 78 18 999 999 999
(days)

Plant Uptake Coef. 1.0 1.0 1.0 1.0 1.0
0.65 0.52
0.0 0.0 0.0 0.0 0.0











The user manuals for the models contain tables drawn from many

sources which provide values of the partition coefficient for many

pesticides. The IEAL H manual also contains a number of tables. The

tables in CREAMS and IEACH are more extensive than those in FPRZM.

Different tables within the same manual my give differing values. Table

values often list coefficients of variation (CV) on the order of 50-130

percent. PRZM presents equations by which the organic carbon partition

coefficient, Koc, can be calculated if solubility or the octanol-water

partition coefficient, Kow, is known.

The partition coefficients of alachlor and atrazine were not

measured on this soil and thus values were chosen from the sources

described above. A Koc value for atrazine of 163 cm3/g with a CV of 49%

was found in the IEACH manual. No listing of a Koc was found for

alachlor. PRZM did list a value for log(Kow) which was 2.78. PRZM

presents a relationship between log(KFw) and Koc which is

log Koc = 1.00 (log Kow) 0.21 5.1

Using this equation a Koc of 371 was calculated for alachlor. IEACH

lists a Kw value for alachlor as 434, using this and equation 5.1, a Koc

value of 268 was calculated. Since alachlor has been reported as one of

the pesticides ocamnmly found in groundwater, it was decided to choose

the lower Koc value for use in the model as this would tend to cause

higher predicted leaching losses.

The inorganic tracers bromide, chloride, and nitrate were assumed to

act as non-adsorbed pesticides. Thus they were assigned a Koc value of

zero.









72

Mhe degradation rate constant of atrazine in soil was given in the

LEACH manual as ranging fromn 0.0149 to 0.0063 days-1, corresponding to a

half-life of between 46 and 110 days respectively. The midpoint value

of 78 days was chosen for use in these simulations. A degradation rate

constant of 0.0384 days-1, which corresponds to a half-life of 18 days,

was given for alachlor in both the CREAMS and IEACH manuals.

The coefficient of plant uptake of pesticides with transpiration was

determined using the relationship given in the PRZM manual in which the

uptake factor is a function of KOw and is given as

UPlKF = 0.784 exp [(log Kow 1.78)2/2.44] 5.2

where UPMKF = plant uptake efficiency factor.

Using the Kow for alachlor of 434 as discussed above, the uptake factor

was calculated to be 0.52. Back calculating a Kow for atrazine, based

upon the chosen Koc of 163 using equation 5.1, yields a KOw of 264.

Using this value in equation 5.2 gives a plant uptake factor of 0.65.

The models were also run using uptake factors for both atrazine and

alachlor of 0.0 and 1.0 in order to assess the sensitivity to these

factors. GLEAMS currently recmmends that in the absence of well defined

uptake coefficients, a value of 1.0 should be used. Uptake factors for

the inorganic tracers were assigned values of both 0.0 and 1.0 to observe

the sensitivity of these non-adsorbed chemicals to plant uptake.

The dates and amounts of the chemicals applied to the soil surface

of the experimental site were entered into the models in a direct manner,

ie. there were no calculations or transformations required to convert the

actual values into model parameters. The application rates used in the

simulations correspond to the intended application rates as opposed to









73

the measured rates. The dates and application rates used as model inputs

are shown in Table 5.3. The applications were assumed to reach the soil

surface with no chemical residues remaining on plant surfaces. The

amount applied was further assumed to be uniformly incorporated into the

top 1 cm of the soil.



5.1.4 Length of simulations

Both models simulated the two year time period from 1/1/86 through

12/31/87. This was done to allow the models to overcome any effects of

initial conditions prior to beginning simulation of chemical

applications. Continuing simulations for six months beyond the end of

field data collection would show if any significant leaching would likely

have occurred after sampling was terminated.


Table 5.3 Chemical applications summary.


Date Chemical Applied Application Rate
(kg/ha)
11/12/86 Atrazine 4.9
Alachlor 4.9

11/17/86 Brcnide 10.0

4/16/87 Nitrate 18.6

4/27/87 Braomide 17.0


5.2 Parameters Unique to PRZM

5.2.1 Evapotranspiration prediction.

PRZM uses either measured daily pan evaporation or daily mean

temperature to determine evaporation from the soil and plant surfaces,

and transpiration by the crop. In these simulations, daily pan











evaporation values recorded at a weather station located on the

University of Georgia Coastal Plain Experiment Station campus were used.

This weather station is located approximately 8 km from the research

site. PRZM requires a pan coefficient which is a multiplier used to

adjust pan readings to represent total daily potential evapotranspiration

(PET). A pan factor of 0.75 was selected front a figure provided in the

manual showing pan factors as related to geographic location.

Two additional parameters required for evapotranspiration (ET)

prediction are the depth in the soil profile fram which evaporation can

occur, and the depth of rainfall and irrigation that can be stored on the

plant canopy (interception storage). A figure in the manual indicates

that for southern Georgia, the depth of soil contributing to evaporation

is approximately 25 cm. Data pertaining to the interception storage

capacity of bahia grass was not included in the manual. Values presented

in the manual for various crops ranged fram 0.0-0.3 an. Assuming that

the storage capacity of bahia grass is relatively small, a value of 0.05

cm was assumed.



5.2.2 Crop related parameters

Characteristics of the crop(s) grown during the simulation period

are defined by parameters related to rooting depth, maximum canopy

coverage, and dates for emergence, maturation, and harvest. A rooting

depth for bahia grass was difficult to obtain. Most references to bahia

grass simply stated that it is a deep or very deep rooted plant. Based

on such generalizations and without more specific information, an active

rooting depth (depth containing 90% of root mass) of 91 cm (3 ft) was











selected. Based on observations at the research site, a maximum areal

coverage of the bahia grass foliage was estimated as 98 percent.

Dates of crop emergence, maturation, and harvest are used in the

model to define the development and growth of roots. Roots are assumed

to begin growing at crop emergence and reach the maximum rooting depth at

plant maturation. The roots stay at maximum depth until harvest when

they are set back to zero. These parameters are somewhat difficult to

define in relation to a perennial crop such as bahia grass. In view of

the use made of the selected dates, some artificial dates for emergence

and harvest could be selected. Since the emergence date really defines

the beginning of plant transpiration as a component of the water balance,

an emergence date of March 1 was selected to correspond to the time when

the grass begins to "green up" in the spring. Since the majority of the

roots already exist when the grass begins active growth in the spring, a

maturation date of April 1 was selected to reflect water extraction from

the full maximum rooting depth shortly after growth for the new year

begins. The harvest date was chosen to be December 1 which reflects the

approxiate date when the grass turns brown in the winter and active

growth ceases.



5.2.3 Soil related parameters

PRZM will allow simulation of a larger section of the profile than

just the root zone. A soil core depth of 2.62 m was chosen to represent

the average profile depth beneath the application area. The model will

allow the user to select the number of opartments used to represent the

profile. Too many compartments will increase simulation times and too











few will lead to increased numerical errors in the solution of the

equations used within the model. The manual suggests that no fewer than

30 compartments be used. Thirty-five ccpartments were selected because

this gives a compartment depth that is one-half the crpartment depth in

GLEAMS for a root zone depth of 91 cman (see section 5.3.3).

PRZM4 and GLEAMS both normnnally operate on the assumption that any

soil water in excess of field capacity will drain into the next lower

compartment. This drainage will continue until any excess water drains

from the bottom of the simulated profile. This drainage is assumed to

occur within one day. In the PRZM model, this type of drainage is called

'free drainage'. PRZM has a feature to allow selection of a drainage

rate parameter that serves to slow down the percolation process. This is

an empirical constant and requires calibration. A figure is provided in

the manual whereby the user can make a first cut selection of the

drainage parameter based upon the soil type (sand, clay loam, etc.) and

the number of simulation coapartments selected as described above.

Observations at the experimental site clearly show that there is a delay

of 2-4 days from the time excess water is applied to the time that the

water table responds. Therefore a few simulations were made using the

restricted drainage option to compare with the experimental data. The

drainage rate parameter for this case was selected from the graph in the

manual for a sand soil with 35 ccmpartments to be 2.63 day-1.











5.3 Parameters Unique to ES

5.3.1 Evapotranspiration prediction

Potential evapotranspiration (PET) is computed in GLEAMS based on

mean monthly maximum and minimum temperatures and mean monthly solar

radiation. The mean monthly temperatures and solar radiation data were

obtained from the same weather station that was used for the pan

evaporation data input to PRZM (section 5.2.1). GLEAMS also

differentiates between soil evaporation and plant transpiration. A soil

evaporation parameter is required as a model input and the manual gives a

suggested value of 3.3 (no units) for this parameter for sands.



5.3.2 Crop related parameters

Crop related parameters which are different from PRZM are leaf area

index (IAI), and winter cover factor. The IAI data are used to reflect

changes in plant transpiration with stage of growth. In keeping with the

observed behavior of bahia grass discussed in section 5.2.2, the IAI of

bahia grass was assumed to be zero until March 1 of each year and to

return to zero on December 1st. The maximum IAI for bahia grass was

assumed to be similar to the IAI for pasture as presented in the CRFAMS

manual which had a maximum IAI of 3.0. The IAI for the bahia grass was

assumed to reach 2.0 on April 1 and to reach 3.0 May 1 where it remained

until it began to decline again to a value of 2.0 on Nov. 1.

The winter cover factor is used to moderate evaporation from the

soil surface during the winter due to presence of a bare surface or cover

crops. Since the soil surface at the site was not bare during winter,

the recommended value of the winter cover factor of 0.5 was used.













5.3.3 Soil related parameters

GLEAMS was designed to simulate only the root zone of a crop. The

vadose zone below the crop is ignored. Thus, the soil profile

description input to GLEAMS covers only the 91 cm root zone as opposed to

the 2.62 m profile input to PRZM. There are no soil parameters which are

significant to the results presented in this work which differ from those

discussed in section 5.1.2. In order to highlight one of the major

differences between the two models, a short discussion of the

computational layering of the soil profile in GLEAMS will be presented

here. GLEAMS divides the crop root zone into seven camputational layers.

The first layer is defined to be 1 cm in depth and the other 6 layers are

equal in thickness (layer 2 is 1 an less than layers 3-7) and together

account for the rest of the root zone. The 1 cm upper layer is

considered an important feature of CREAMS and GLEAMS because it increases

the sensitivity of erosion and runoff to chemical concentrations near the

soil surface. PRZM divides the entire profile into a specified number of

ccompartments of equal depth. For the case presented here, GLEAMS layers

would be 15 cm thick after the surface layer (91 can / 6 layers). For

FPRZM with a 2.62 m profile and 35 ccmpartments, each compartment will be

7.5 cm thick. Thus, PRZM may be better able to describe effects of

actual soil layering since properties of adjoining layers would not be

averaged over as large a layer as in GLEAMS. However the surface layer

in PRZM will also be 7.5 cm thick and pesticide storage will be averaged

over this whole layer effectively lowering the mass at the surface which

is available for transport with eroded sediment and runoff water.















CHAPrER 6
RESULTS AND DISCUSSION



There are many areas in this study for which results can be

reported. The approach used in this chapter is to report significant

results relative to each area such as sample collection, sample analysis,

simulation results, etc., and then relate the individual results where

appropriate. A summary of activities at the field site is presented in

Table 6.1 for reference.



6.1 Data Collection

The site was visited on at least a weekly basis throughout the

study. At each visit, samples were collected from the monitoring wells

and soil solution samplers. The water table elevation was also recorded

for every monitoring well on the site and for several wells located

adjacent to the study area. Soil samples were collected on an irregular

basis during the study.

The procedure used for collecting samples from the groundwater

worked well in general. Sand intrusion into a few of the wells, however,

was a consistent problem. It was possible in some instances to fill the

250 mL sampling bottle with sand. This problem was overcome by slowly

raising the bottom of the sampling tube in order to stay above the sand.

These wells were pumped out using a 1.6 cm diameter suction line to

remove accumulated sand as needed. The problem could have been avoided

79











Table 6.1 Chronological summary of field site activities.

Date Event

11/12/86 Apply 4.9 kg/ha of atrazine and alachlor, apply 5.1 an

of irrigation after herbicide application.

11/17/86 Apply 500 g of KBr (10 kg/ha Br-), collect soil solution

samples.

11/18/86 Collect soil solution samples. First soil samples taken

frcm top 5 an in application area.


Collect soil solution and soil samples.


Collect soil

Collect soil

irrigation.

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil

Collect soil


solution

solution


samples.

samples,


apply 5.1 cm of


solution and well samples.

solution, well, and soil samples.

solution and well samples.

solution and well samples.

solution and well samples.

solution and well samples.

solution and well samples.

solution and well samples.

solution, well, and soil samples.

solution and well samples.

solution and well samples.

solution and well samples.

solution and well samples.


11/24/86

12/01/86

12/08/86



12/15/86

12/22/86

12/29/86

1/05/87

1/12/87

1/19/87

1/26/87

2/02/87

2/09/87

2/16/87

2/23/87

3/02/87

3/09/87

3/16/87


Collect soil solution, well, and soil samples.











Table 6.1 Contir

Date

3/23/87

3/31/87

4/06/87

4/13/87

4/16/87

4/20/87



4/27/87


ned.


Event

Collect soil solution and well samples.

Collect soil solution and well samples.

Collect soil solution and well samples.

Collect soil solution and well samples.

Apply 560 kg/ha of 5-10-15 to entire study site.

Collect soil solution and well samples, Very few wells

have water.

Collect soil solution and well samples, Apply 2.5 an o,

irrigation, apply approximately 17 kg/ha Br-, apply 2.


5


ancm of irrigation.

Collect soil solution and

irrigation.

Collect soil solution and

irrigation.

Collect soil solution and

irrigation.

Collect soil solution and

solution containing 500 g

an of irrigation.

Collect soil solution and

irrigation.

Collect soil solution and

irrigation.


well samples, apply 4.2 can of


well samples, apply 3 can of



well samples, apply 3 man of


well samples,

KCl into well


pour 10 1 of

07-09, apply 2.7


well samples, apply 3.2 ancm of



well samples, apply 1.9 an of


4/28/87



4/29/87



4/30/87



5/01/87





5/02/87



5/03/87











Table 6.1

Date

5/05/87



5/08/87



5/11/87



5/13/87

5/18/87

5/25/87

6/01/87


continued.


Collect soil

irrigation.

Collect soil

irrigation.

Collect soil

irrigation.

Collect soil

Collect soil

Collect soil

Collect soil


solution ar


Event

rd well


samples, apply 0.9 an of


solution and well samples, apply 1.9 an of



solution and well samples, apply 5.2 an of



solution and well samples.

solution and well samples.

solution, well, and soil samples.

solution and well samples.


by using

wells.


cxmercially available PVC well screening in the monitoring


No problems were observed with the operation of the soil solution

samplers. The volumes of water collected during sampling varied from 0

to greater than 50 mL. Even though the samples were pulled back to the

instrument trailer through as much as 35 m of 0.64 ancm ID polypropelene

tubing, the vacuum applied to the sample lines was sufficient to pull

very small droplets of water along the wall of the tube and into the

containers. No carryover of the tracers between sampling periods was

observed.

The tensicneters were difficult to maintain properly on the weekly

sampling schedule. Even when the tensicmeters were working, their

usefulness was limited by the short length of time during which they


Continued.









83

could be observed. If an irrigation was initiated early in the morning,

a few of the 30 cm tensiameters might have responded by the end of the

day. Weekly observations of soil-water tension were insufficient for the

purpose of calculating the flux of water through the unsaturated zone.



6.2 Sample Analysis

Water samples collected from the monitoring wells were analyzed for

residues of atrazine and alachlor and concentrations of the inorganic

tracers utilized in this research (Br-, NO3-, and Cl-). The volume of

samples from the solution samplers ranged from 0 to more than 50 mL, with

the average sample volume being approximately 10 mL. Water samples from

the soil solution samplers were generally analyzed for the inorganic

tracers only. Most of the solution samples were not extracted for

residues of the herbicides due to the small sample volumes.

Approximately 50 of the soil solution samples were extracted and analyzed

for residues of the herbicides. Soil samples were extracted and analyzed

for residues of atrazine and alachlor. The presence of alachlor and

atrazine residues in the soil samples was confirmed in a composite of

soil sample extracts that was analyzed using gas chrcamiatography/mass

spectrometry (GC/MS).

During extraction of water samples with hexane, an emulsion of

hexane and water sometimes formed. This would usually separate if

allowed to sit undisturbed for a few minutes. If the enulsion did not

separate, herbicide residues contained within the emulsion could have

been discarded. There was an apparent contamination of the glassware

used for water sample collection or extraction. Samples of deionized











water were extracted and they showed concentrations of atrazine of

approximately 1 gg/L. All glassware was routinely washed with a

cumnercial glassware detergent, triple rinsed with deionized water, and

rinsed with acetone. The contamination levels appeared to be consistent

in that water samples, in which atrazine was not expected, showed a

concentration of approximately 1 J/L. The source of the contamination

was not determined. Several attempts were made to isolate and identify

the source of the contamination, however these attempts were

unsuccessful. To account for the contamination, the measured

concentrations were reduced by 1 pg/L. Thus no concentrations of less

than 1 Mg/L are reported even though the sensitivity of the GC would

allow detection down to the range of tenths of a pq/L.

The gas crciduatograph performed well except that the N-P detector

response had a tendency to drift. This was accounted for by running

standards frequently and adjusting calculated concentrations as described

in the methods section.

The recoveries of atrazine and alachlor from spiked water samples

were nearly identical and ranged frcm 68 to 104% with a mean of 79% and a

standard deviation of 12 percent. The standard deviation would probably

be reduced if these analyses had been performed by an experienced

analyst. Experience contributes significantly to the ability to produce

accurate and consistent results.

The recoveries of atrazine and alachlor from spiked soil samples

could not be determined from the data collected. Moisture contents of

the soils used for spiking were not recorded. The spike levels were

based on the moist weight of the soil. After spiking, the soil was air-











dried for 1 day in order to allow the solvent from the herbicide

standards to evaporate prior to taking subsamples for analysis. The loss

of water from the samples during this time resulted in an increased mass

of pesticide per unit weight of wet soil. Since the initial moisture

contents were unknown, it was not possible to calculate the expected

concentrations in the air dried samples. Calculations based on the

spiking levels of the moist soil resulted in apparent recoveries that

exceeded 100 percent. It can be concluded from the data that recoveries

from the soil were high. When the missions in the soil recovery study

were discovered, the facilities used to perform extractions and analyses

were no longer available.

All concentrations (water and soil) reported in the following

sections are the analytically determined concentrations and have not been

adjusted for expected recovery rates.



6.3 Chemical Applications

Applications of the herbicides and bromide tracer were monitored to

determine the concentration of the chemicals in the application water and

the volume of water applied. Concentrations of the chemicals were

multiplied by the volume of water collected and divided by the opening

area of the collectors in order to determine the application rate on a

mass per unit area basis. Table 6.2 summarizes the measured chemical

application data. The intended application rates were presented in Table

5.3.

The application of atrazine and alachlor to the designated

application area was not uniform as can be seen in Figures 6.1 and 6.2.