Analysis of Large-Scale Climate Datasets and Hydrologic Variables in the Tampa Bay Region: Selection of Predictor Climat...

MISSING IMAGE

Material Information

Title:
Analysis of Large-Scale Climate Datasets and Hydrologic Variables in the Tampa Bay Region: Selection of Predictor Climate Indices
Physical Description:
Book
Creator:
Martinez, C. J.
Risko, S. L.
Graham, W. D.
Jones, J. W.
Publisher:
Department of Agricultural and Biologial Engineering, University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Notes

Abstract:
This work was conducted to identify climate patterns related to monthly and seasonal rainfall, streamflow, and demand in the Tampa Bay region and to recommend climate indices that could be used to improve forecasts of these variables. As part of this work a review of online climate analysis tools was conducted to evaluate their suitability in producing results such as those presented here. Lagged linear correlation maps were produced between seasonal mean rainfall, streamflow, and detrended demand with seasonal means for each of the three gridded climate variables: sea surface temperatures, sea level pressures, and 500mb geopotential heights. Lagged correlation maps were produced for regional means/totals and for individual stations in order to examine the variability of results. Lagged composite anomaly maps of the gridded climate variables were then created for extreme seasonal hydrologic events (e.g. 10th and 90th percentiles). Based on the patterns found by lagged correlation and composite analyses, indices of each climate variable were identified for further analysis. Each index was evaluated using lagged Pearson’s product-moment correlation and Spearman’s rank correlation of monthly and seasonal values. The most significant results, in terms of correlation magnitude and persistence, were found with indices of ENSO. The Niño 3 and Niño 3.4 sea surface temperature indices and the station-based and reanalysis-based Southern Oscillation Indices (SOI and eqSOI, respectively) were found to show significant and coherent correlations at lead-times up to nine months. The variability of these relationships during different phases of the Atlantic Multidecadal Oscillation (AMO) were then examined. Significant differences were found between different time periods of the AMO, however no clear pattern between phases was found. It is recommended that one of the identified ENSO indices be employed when developing climate-based forecasts. However, it is also recommended that the strength and pattern of the relationship be verified according to the time-period of historical data used to develop forecasts or train models.
General Note:
Report presented during the Tampa Bay Water project meeting held by Chris Martinez (University of Florida)

Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
All rights reserved by the source institution and holding location.
System ID:
AA00012272:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Analysis of Large-Scale Climate Data sets and Hydrologic Variables in the Tampa Bay Region: Selection of Predictor Climate Indices Christopher J. Martinez1, Susan L. Risko1, Wendy D. Graham1, 2, and James W. Jones1 1 Department of Agricultural and Biologi cal Engineering, University of Florida 2University of Florida Water Institute

PAGE 2

2 Table of Contents Executive Summary ............................................................................................................. ......... 3 Introduction .................................................................................................................. ................. 4 Data .......................................................................................................................... ...................... 4 Hydrologic data ............................................................................................................... ............ 4 Climate data .................................................................................................................. ............... 7 Methods ....................................................................................................................... ................... 9 Online Tools Available ........................................................................................................ ....... 10 Results ....................................................................................................................... ................... 12 Relationships with sea surface temperatures ............................................................................. 12 Rainfall and SSTs ............................................................................................................. ..... 12 Streamflow and SSTs ........................................................................................................... 29 Demand and SSTs ............................................................................................................... .. 46 SST Index selection and evaluation....................................................................................... 63 Relationships with Sea Level Pressures .................................................................................... 70 Rainfall and SLPs ............................................................................................................. ..... 70 Streamflow and SLPs ........................................................................................................... 87 Demand and SLPs ............................................................................................................... 104 SLP Index selection and evaluation..................................................................................... 121 Relationships with 500 mb Geopotential Heights ................................................................... 131 Rainfall and GpHs ............................................................................................................. .. 131 Streamflow and GpHs ......................................................................................................... 1 49 Demand and GpHs............................................................................................................... 166 GpH Index selection and evaluation .................................................................................... 183 Influence of the Atlantic Multidecadal Oscillation ................................................................. 188 Summary and Recommendations ............................................................................................ 195 References .................................................................................................................... .............. 197Appendix A. Pearson’s Correlation of St ation Data with Sea Surface Temperatures Appendix B. Pearson’s Correlation of St ation Data with Sea Level Pressures Appendix C. Pearson’s Correlation of Stat ion Data with 500mb Geopotential Heights Appendix D. Spearman’s Rank Correlation of Selected Indices with Mean Standardized Rainfall, Streamflow, and Total Regional Demand

PAGE 3

3 Executive Summary This work was conducted to identify climate patterns related to monthly and seasonal rainfall, streamflow, and demand in the Tampa Bay regi on and to recommend climate indices that could be used to improve forecasts of these variables. As part of this work a review of online climate analysis tools was conducted to evaluate their suitability in producing results such as those presented here. Lagged linear correlation maps were produced be tween seasonal mean rainfall, streamflow, and detrended demand with seasonal means for each of the three gridded climate variables: sea surface temperatures, sea level pressures, and 500mb geopotential heights. Lagged correlation maps were produced for regional means/totals and fo r individual stations in order to examine the variability of results. Lagged composite anomal y maps of the gridded climate variables were then created for extreme seasona l hydrologic events (e.g. 10th a nd 90th percentiles). Based on the patterns found by lagged correlat ion and composite analyses, indi ces of each cl imate variable were identified for further analysis. Each i ndex was evaluated using lagged Pearson’s productmoment correlation and Spearman’s rank correl ation of monthly and seasonal values. The most significant results, in terms of correla tion magnitude and persistence, were found with indices of ENSO. The Nio 3 and Nio 3.4 sea su rface temperature indices and the station-based and reanalysis-based Southern Os cillation Indices (SOI and eqSO I, respectively) were found to show significant and coherent corre lations at lead-times up to nine months. The variability of these relationships during different phases of the Atlantic Multidecadal Oscillation (AMO) were then examined. Significant differences were f ound between different time periods of the AMO, however no clear pattern between phases was f ound. It is recommended that one of the identified ENSO indices be employed when deve loping climate-based forecasts. However, it is also recommended that the strength and pattern of the relationship be verified according to the time-period of historical data used to develop forecasts or train models.

PAGE 4

4 Introduction The purpose of this document is to identify climate variables that likely influence rainfall, streamflow, and demand in the Tampa Bay re gion and to identify an optimal set of monthly/seasonal climate indices that may be used to forecast hydrologic variables. This document focuses on monthly and seasonal variabil ity rather than decadal, multi-decadal, or longer-term climate change. Resu lts of this work are intended to compliment previous work on decadal/multi-decadal climate variations and West-Central and South Florida rainfall and streamflow (Obeysekera et al., 2007; Asefa and Adams, 2008; Kelly and Gore, 2008). Several studies have analyzed climate linkages with hydrologic variables in the southeast United States (e.g. Ropelewski and Halpert, 1986; Gers hunov and Barnett, 1998; Enfield et al., 2001). However, most of these analyses were conducted over large areas (s tates, regions, or continents) and commonly-used climate indices may be le ss than optimal at smaller spatial scales (Harshburger et al., 2002; Grantz et al., 2005). Fo r example, Grantz et al. (2005) found a custom index of 500 mb geopotential heights over the nor thern Pacific Ocean and western US to be a more significant predictor of streamflow in the Truckee-Carson River system compared to standard indices of the El Ni o-Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO). This study was performed to identify optimal indices of climate variables for use in the Tampa Bay Water tri-county region, an area of approximately 2670 square miles (4296 km). Correlation and composite analyses of gridded climate datasets with hydrologic variables were used to identify an optimal set of climate indices. Climate indices were identified based on relationships found between historical rainfall streamflow, and member government demand with historical sea surface temperatures (SST s), 500 mb geopotential heights (GpHs), and sea level pressures (SLPs). Particular emphasis wa s given to significant lagged relationships and preference was given to climate phenomena known to influence clim ate in the southeastern USA. As part of this work a review of online climat e analysis tools was conducted to evaluate their suitability in producing the results presented he re. While it was ultimately decided to produce a code for the research and documen tation of this particul ar analyses in Matlab due to the large number of maps produced, a su mmary of the online tools currently available and a brief evaluation of these t ools is presented. Data Hydrologic data Historical records of monthly/ seasonal rainfall, streamflow, and demand were used in this analysis. Monthly gauge rainfall records were obtained from the National Climatic Data Center (Figure 1 and Table 1), streamfl ow records were obtained from the United States Geological

PAGE 5

5 Survey National Water Information System and Tampa Bay Water (Figure 1 and Table 2), and monthly demand was obtained fro m Tampa Bay Water (Table 3). For the rainfall and streamflow datasets the mean of standardized anomalies of all ga uges/stations was calculated for analysis. Converting each stat ion into standardized anomalie s equally weights each station (mean = 0, standard deviation = 1). By this approach more observations are included in the calculation of the mean in later years due to th e length of each observed dataset. To evaluate potential differences between st ations/observations correlation an alyses were also conducted for each observation (correlation maps for each stat ion/observation can be found in Appendices A, B, and C). Figure 1. Location of rainfall and streamflow stations used in this study

PAGE 6

6 Table 1. Rainfall data used in this study COOP ID Station Name County Latitude LongitudeData Rangea 80478 Bartow Polk 27.90 81.85 10/1900-8/2008 80645 Bradenton 5 ESE Manatee 27.45 82.50 4/1965-9/2008 81046 Brooksville Chin Hill Hernando 28.62 82.37 10/1900-9/2008 81163 Bushnell 2 E Sumter 28.67 82.08 11/1936-9/2008 81632 Clearwater Pinellas 27.97 82.77 9/1931-3/1977 83153 Fort Green 12 WSW Manatee 27.57 82.13 9/1955-9/2008 83986 Hillsborough River SPHills borough28.15 82.23 9/1943-9/2008 86880 Parrish Manatee 27.62 82.35 1/1958-8/2008 87205 Plant City Hillsbor ough28.02 82.15 2/1903-9/2008 87851 St. Leo Pasco 28.33 82.27 10/1900-9/2008 87886 St. Petersburg Pinellas 27.77 82.63 8/1914-9/2008 88788 Tampa Intl. Airport Hills borough27.97 82.53 2/1950-9/2008 88824 Tarpon Springs SWG Pinellas 28.15 82.75 3/1901-9/2008 89430 Weeki Wachee Hernando 28.52 82.58 10/1969-9/2008a Some years missing or contain missing values Table 2. Streamflow data used in this study USGS ID Station Name Latitude Longitude Data Rangea2301000 North Prong at Keysville (Alafia) 27.88 82.10 10/1950 – 9/2008 2301300 South Prong near Lithia (Alafia) 27.80 82.12 10/1963 – 9/2008 2301500 Alafia River at Lithia 27.87 82.21 10/1932 – 9/2008 Alafia River at Bell Shoals b 10/1974 – 9/2008 2303000 Hillsborough River Near Zephyrhills 28.15 82.23 10/1939 – 9/2008 2303330 Hillsborough River at Morris Bridge 28.10 82.31 10/1972 – 9/2008 S160 Adjusted (Tampa Bypass Canal)c 10/1974 – 9/2002 a Some years missing or contain missing values b Calculated by Tampa Bay Water from Lithia Springs and Lithia Gauge (USGS ID 2301300 and 2301500) c Adjusted flow over S160 structure, withdrawals by City of Tampa removed Table 3. Demand data used in this study Total and Member Government DemandData Range Total Regional Demand 10/1991 – 12/2008 City of Tampa WDPAa 10/1991 – 12/2008 New Port Richey WDPA 10/1991 – 12/2008 Northwest Hillsborough WDPA 10/1991 – 12/2008 Pasco County Delivered 10/1991 – 12/2008 Pasco County Self Supply 6/1998 – 12/2008 Pinellas WDPA 10/1991 – 12/2008 South-Central Hillsborough WDPA 10/1991 – 12/2008 St. Petersburg WDPA 10/1991 – 12/2008 a WDPA = Water Demand Planning Area

PAGE 7

7 Historical demand was detrended in an attempt to remove non-clim atic influences such as the influences of increased dema nd due to population growth, decreases in demand due to implementation of conservation measures, changes in demand due to changes in land use, etc (Figure 2). A linear trend was us ed due to the relative shortnes s of the datasets. Plots of detrended demand anomalies for each of the memb er governments listed in Table 3 are shown in Figure 3. 60 40 20 0 20 40 60 80 100 120 0 50 100 150 200 250 300 35010/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) Total Regional Demand Linear Trend Detrended Anomaly Figure 2. Total regional demand and linearly detrended anomalies. Climate data The 2 x 2 monthly Ex tended Reconstructed SST Version 2 (ERSST.v2) dataset of Smith and Reynolds (2004) was obtained from the International Research In stitute for Climate and Society1 (IRI) data library. Following Ba rnett and Priesendorfer (1987) and Tootle and Piechota (2006) climate variables in the region bounded by 30S 70N and 120E 0 were used in this analysis. A linear global warmi ng trend was removed from the gr idded SSTs for this analysis. The ERSST.v2 dataset begins in January 1854, howe ver it is heavily damped before 1880 due to sparse data. More information is available on the ERSST.v2 dataset at http://www.ncdc.noaa.gov/oa/clim ate/research/s st/ersstv2.php 1 http://iridl.ldeo.columbia.edu/SOURCE S/.NOAA/.NCDC/.ERSST /.version2/.SST/

PAGE 8

8 15 10 5 0 5 10 15 20 25 30 0 10 20 30 40 50 60 70 80 90 10010/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) City of Tampa WDPA Linear Trend Detrended Anomaly 2 1 0 1 2 3 4 0 1 2 3 4 5 610/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) New Port Richey WDPA Linear Trend Detrended Anomaly 5 3 1 1 3 5 7 9 11 13 15 0 5 10 15 20 2510/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) Northwest Hillsborough WDPA Linear Trend Detrended Anomaly 10 5 0 5 10 15 20 25 30 0 5 10 15 20 25 30 3510/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) Pasco County Delivered Linear Trend Detrended Anomaly 2 1 0 1 2 3 4 5 6 2 1 0 1 2 3 410/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) Pasco County Self Supply Linear Trend Detrended Anomaly 20 15 10 5 0 5 10 15 20 25 30 0 10 20 30 40 50 60 70 80 9010/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) Pinellas WDPA Linear Trend Detrended Anomaly 10 5 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 4510/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) South Central Hillsborough WDPA Linear Trend Detrended Anomaly 6 4 2 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 45 5010/1991 10/1992 10/1993 10/1994 10/1995 10/1996 10/1997 10/1998 10/1999 10/2000 10/2001 10/2002 10/2003 10/2004 10/2005 10/2006 10/2007 10/2008 Detrended Anomaly (MGD) Demand (MGD) St. Petersburg WDPA Linear Trend Detrended AnomalyFigure 3. Member government demand and linearly detrended anomalies.

PAGE 9

9 Monthly 2.5 x 2.5 SLPs2 and 500 mb GpHs3 from the NCEP/NCAR reanalysis project (Kalnay et al., 1996) were obtained from IRI data librar y. Since reanalysis data via IRI is limited to 1949-present the resulting correlation and composite analyses are limited to this time period. More information on the reanalys is project can be found at http://www.cdc.noaa.gov/cdc/reanalysis/reanalysis.shtml Gridded SSTs, SLPs, and GpHs were converted in to 3-month seasonal an omalies for correlation and composite analysis. Three-month averages were used to reduce noi se in the analyses. Subsequent evaluation of indices based on the co rrelation and composite analyses are presented in both 3-month and monthly values. Methods Linear correlation and composite analyses were used to identify relati onships between seasonal gridded climate datasets and hyd rologic variables. Correlati on and composite analyses of gridded climate variables have been shown to be effective techniqu es in the selection of climate indices (e.g. Grantz et al., 2005; Oplitz-Stapleton et al., 2007; Sveinsson et al., 2008). Correlations and composites were determined for concurrent seasons (lag 0) and with climate datasets lagging hydrologic obser vations. Lagged correlations between 3-month averaged hydrologic observations and 3-mont h averaged climate variable s were conducted in 3-month increments for a total of 12 m onths (lags of 0, 3, 6, 9, and 12 months) for SSTs and SLPs and in 1-month increments for a total of 4 months for GpHs. The rationale for this difference in lags evaluated is based on the lagged response each variable is expected to have on climate in the southeast United States as well as our own initial analyses. For example, changes in SSTs in the tropical Pacific Ocean do not have a direct effect but rather influence atmospheric pressure and atmospheric flow patterns over the Pacific which may in turn influence the southeast via the jetstream (Horel and Wallace, 1981). Correlations were used to identify linear relatio nships between 3-month gr idded climate datasets and hydrologic observations. Only spatially coherent (a pproximately stationary and persistent) and statistically significant corr elation patterns were considered in climate index selection. Composite analysis, sometimes referred to as supe rposed epoch analysis or just epoch analysis, was conducted to examine differen ces in climate states that coincide to extreme hydrologic conditions. Composite analysis consists of sorting data into categories and examining differences in the means of different categories. The advantage of composite analysis is that it makes no assumption of symmetry and can be us ed to identify nonlinear relationships. The 2 http://iridl.ldeo.columbia.edu/SOU RCES/.NOAA/.NCEP-NCAR/.CDAS-1/.MONT HLY/.Intrinsic/.MSL/.pressure/ 3 http://iridl.ldeo.columbia.edu/ SOURCES/.NOAA/.NC EP-NCAR/.CDAS1/.MONTHLY/.Intrinsic/.PressureLevel/.phi/

PAGE 10

10 drawbacks to composite analysis are that it is based on a limited set of the original data and can be vulnerable to leveraging, resulting from the in fluence of a single large anomaly. For this study the 10th and 90th percentiles of rainfall and streamflow were chosen to identify extreme wet and dry years for each season. For total regional demand the 20th and 80th percentiles were used to increase the composite sample size. Composite maps of gridded climate variables during these extremes were created to identify typical climate patterns that correspond to these extremes and are displayed as departures from climatological means. Climatological means were defined by the length of each hydrologic dataset (no single reference period was us ed). Composite maps are simply the mean conditions (mean anomalies) of climate variables during the wet and dry periods identified. Based on the correlation and composite patterns found, climate indices are identified and the concurrent and lagged co rrelation of hydrologic variables with these indices is presented using plots of lagged Pearson’s product-moment corre lation and Spearman’s rank correlation. Where multiple spatial patterns existed more than one climate index was selected for evaluation from each gridded climate dataset. Relationships between the selected indices and hydrologic variables are presented using both se asonal means and monthly values. The potential influence of the Atlantic Multid ecadal Oscillation (AMO ) on the results found using correlation and composite an alysis is also evaluated. Ra infall and streamflow datasets were stratified by AMO phase. Lagged Pearson’ s product-moment correlation plots of select indices and rainfall and streamfl ow were repeated for each phase of the AMO. Non-parametric tests for differences of rainfall and stream flow categorized by AMO phase was conducted by one-way Kruskal-Wallis tests (analogous to parametric ANOVA). Online Tools Available There are currently three online tools available for the analysis of climate datasets (Table 4). Each has advantages and disadvantages and the c hoice of which to use is dependent on the type of analysis to be conducted. Each tool listed in Tabl e 4 contains an onlin e library of climate datasets (with download capability) and offers va rious statistical and graphical analyses. The Earth Systems Research Laboratory (ESRL) Physical Science Division (PSD) interactive plotting and analysis tools are the easiest to use for first-time users, partially due to the online help and instructions available. The ESRL PSD tools allow for plotting of gridded datasets and allows for user-defined time series to be uploaded or accessed from another web or ftp site. The ESRL PSD site also includes pre-made global a nd US ENSO composites of several variables. The ESRL plotting tools are stab le but appear to be the slowest of the tools in terms of expansion.

PAGE 11

11 Table 4. Online climate exploration tools currently available Tool Source Web Address Advantages Disadvantages Earth Systems Research Laboratory Physical Science Division Interactive Plotting and Analysis Tools National Oceanic and Atmospheric Administration (NOAA) http://www.cdc.noaa.gov/cgibin/PublicData/getpage.pl Accepts custom time-series Good online help (easiest tool for first-time user) Limited number of analysis techniques (relative to the other tools reviewed) Ingrid (a part of the IRI data library) International Research Institute for Climate and Society (IRI) http://iridl.ldeo.columbia.edu/ index.html Multiple datasets available Multiple analysis techniques available Does not accept custom time-series Many analysis techniques require understanding Ingrid code (Online help available) Climate Explorer Royal Dutch Meteorological Institute (KNMI) http://climexp.knmi.nl/ Accepts custom time-series and fields Multiple datasets available Multiple analysis techniques available Limited online help available The IRI data library contains a comprehensive library of climate, hydrologic, and model datasets. Numerous statistical techniques are available, however some require the user to use ‘expert mode’ and enter Ingrid code (the post-script language the library is built upon). The Ingrid script and some step by step examples are well docum ented on the website. The data library and statistical analyses offered by IR I are stable and are periodically updated w ith new techniques and datasets. One disadvantage of the IRI data library is that us er-defined time series or fields currently cannot be used. The IR I data library and statistical t ools are particularly useful for selecting, viewing, and downloading subsets of a dataset and for pr e-processing a dataset prior to download. The Climate Explorer of the Royal Dutch Meteor ological Institute (KNMI) may contain the most comprehensive data library and av ailable statistical methods. The Climate Explorer also appears to be undergoing the most active ex pansion in terms of datasets and statistical methods compared to the other tools. However, there is a possibility that it is less stable than other tools since techniques and data may undergo less quality cont rol before being added. This is only our

PAGE 12

12 impression and is based on some of the ‘corrections ’ and ‘bugs fixed’ listed on the site. This is likely only a concern with recent datasets and techniques added. Climate Explorer also allows users to upload custom time series and fields for an alysis. Climate explorer is relatively easy to use, however help/instructions are sometimes lacking making the details of certain statistical techniques unclear. Results Relationships with sea surface temperatures Rainfall and SSTs Significant positive correlations (p < 10) occur between SSTs in the east and central tropical Pacific and mean standardized rainfall from No vember January (NDJ) through February April (FMA) (Figures 4-7). A low, localized correlation can be seen in the east tropical Pacific in the October December (OND) season (Figure 7). The pattern of the correlations confirm the influence of ENSO on rainfall in the region, with highe r (lower) than normal rainfall occurring during periods of warmer (cooler) than normal SSTs in the eastern and central tropical Pacific. The strongest correlations are observed in the January March (JFM) (F igure 4) and December – February (DJF) (Figure 7) seasons Correlations are st rongest during concurre nt seasons (lag 0), however significant correlations exist at 3a nd 6-month lags. No significant and coherent correlations were found between March May (MAM) and Se ptember – November (SON) (Figures 4-6). Composites of SSTs for years below the 10th and above the 90th percentiles of seasonal mean standardized rainfall support th e correlation patterns found (Figur e 8-19). For several seasons anomalous SSTs are observed at greater lags comp ared to the correlations seen in Figures 4-7. Comparison of the 10th and 90th composites for each season indi cates that the relationship between rainfall and SSTs is approximately lin ear and stationary. Composite patterns may indicate nonlinearity in the re lationship between rainfall and SSTs in the FMA and OND seasons (Figures 9 and 17) where a stronger deviatio n from mean conditions is seen for the 90th percentile of rainfall compared to the 10th percentile. Weak anomalies of the opposite sign are present in April-June (AMJ), May-July (MJJ) and August-October (ASO). However these results may be spurious since they are not pr esent in the composites of seasonal streamflow which will be presented later in this report. The correlation patterns of indivi dual rainfall gauges are relatively consistent (Appendix A.1). However, it is important to note that shorter da tasets tend to exhibit stronger correlations and also tend to show greater correlations in the FMA and OND seasons. The reason for this difference is uncertain, but could be due to data quality in early years or the influence of decadal/inter-decadal climate influences.

PAGE 13

13 Figure 4. Pearson's correlation of January March (left column), February April (center column), and March May (right column) standardized rainfall with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.164 is significant at p = 0.10.

PAGE 14

14 Figure 5. Pearson's correlation of April June (left column), May July (center column), and June August (right column) standardized rainfall with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.164 is significant at p = 0.10.

PAGE 15

15 Figure 6. Pearson's correlation of July September (left column), August October (center column), and September November (right column) standardized rainfall with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.164 is significant at p = 0.10.

PAGE 16

16 Figure 7. Pearson's correlation of October December (left column), November January (center column), and December February (right column) standardized rainfall with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.164 is significant at p = 0.10.

PAGE 17

17 Figure 8. Composite anomalies (C) of concurrent and lagged sea surface temperatures during January March for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 18

18 Figure 9. Composite anomalies (C) of concurrent and lagged sea surface temperatures during February April for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 19

19 Figure 10. Composite anomalies (C) of concurrent and lagged sea surface temperatures during March May for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 20

20 Figure 11. Composite anomalies (C) of concurrent and lagged sea surface temperatures during April June for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 21

21 Figure 12. Composite anomalies (C) of concurrent and lagged sea surface temperatures during May July for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 22

22 Figure 13. Composite anomalies (C) of concurrent and lagged sea surface temperatures during June August for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 23

23 Figure 14. Composite anomalies (C) of concurrent and lagged sea surface temperatures during July September for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 24

24 Figure 15. Composite anomalies (C) of concurrent and lagged sea surface temperatures during August October for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 25

25 Figure 16. Composite anomalies (C) of concurrent and lagged sea surface temperatures during September November for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 26

26 Figure 17. Composite anomalies (C) of concurrent and lagged sea surface temperatures during October December for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 27

27 Figure 18. Composite anomalies (C) of concurrent and lagged sea surface temperatures during November January for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 28

28 Figure 19. Composite anomalies (C) of concurrent and lagged sea surface temperatures during December February for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1900 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 29

29 Streamflow and SSTs Correlations between SSTs and mean standardized streamflow (F igures 20-23) follow similar patterns to those seen with rainfall, supporting th e influence of ENSO in the region. However, unlike rainfall, significant correlations continue into the MAM season due to the lagged response of streamflow to rainfall (Figure 20). In addition, correlations between NDJ and MAM show stronger significance at 9-month la gs compared to rainfall. Composites of SSTs during years of high/low seas onal streamflow years show similar patterns (Figures 24-35) as those seen during high and low ra infall years. Overall, the composite anomalies were greater for years of high s easonal streamflow (greater than the 90th percentile) compared to years of low seasonal streamflow. As also seen in the correlation patterns, anomalous SSTs are observed during the MAM s eason for streamflow (Figure 26) where these were not observed for rainfall. The results of the composites indicate that the relationship between SSTs is most linear in DJF and JFM compared to other seasons. In FMA, OND, and NDJ the anomalies are greater for the upper 90th percentile of streamflow compared to the lower 10th percentile. The correlation patterns are fairly consistent be tween stations (Appendix A.2), as was the case for rainfall. The strength of the correlations of individual streamflow sta tions with SSTs varies depending on the length of the data set (shorter datasets tend to exhibit correlations of greater magnitude), supporting the results found for rainfall.

PAGE 30

30 Figure 20. Pearson's correlation of January March (left column), February April (center column), and March May (right column) standardized streamflow with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.195 is significant at p = 0.10.

PAGE 31

31 Figure 21. Pearson's correlation of April June (left column), May July (center column), and June August (right column) standardized streamflow with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.195 is significant at p = 0.10.

PAGE 32

32 Figure 22. Pearson's correlation of July September (left column), August October (center column), and September November (right column) standardized streamflow with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.195 is significant at p = 0.10.

PAGE 33

33 Figure 23. Pearson's correlation of October December (left column), November January (center column), and December February (right column) standardized streamflow with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.195 is significant at p = 0.10.

PAGE 34

34 Figure 24. Composite anomalies (C) of concurrent and lagged sea surface temperatures during January March for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 35

35 Figure 25. Composite anomalies (C) of concurrent and lagged sea surface temperatures during February April for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 36

36 Figure 26. Composite anomalies (C) of concurrent and lagged sea surface temperatures during March May for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 37

37 Figure 27. Composite anomalies (C) of concurrent and lagged sea surface temperatures during April June for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 38

38 Figure 28. Composite anomalies (C) of concurrent and lagged sea surface temperatures during May July for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 39

39 Figure 29. Composite anomalies (C) of concurrent and lagged sea surface temperatures during June August for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 40

40 Figure 30. Composite anomalies (C) of concurrent and lagged sea surface temperatures during July September for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 41

41 Figure 31. Composite anomalies (C) of concurrent and lagged sea surface temperatures during August October for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 42

42 Figure 32. Composite anomalies (C) of concurrent and lagged sea surface temperatures during September November for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 43

43 Figure 33. Composite anomalies (C) of concurrent and lagged sea surface temperatures during October December for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 44

44 Figure 34. Composite anomalies (C) of concurrent and lagged sea surface temperatures during November January for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 45

45 Figure 35. Composite anomalies (C) of concurrent and lagged sea surface temperatures during December February for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1932 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 46

46 Demand and SSTs The spatial correlation patterns of total regional demand with SSTs (Figures 36-39) are more variable and less stationary than that seen for rainfall and streamfl ow. This is likely due to the relative shortness of the datase t (1991-2008) as well as the use of a linear trend to remove nonclimatic influences on demand. Of particular no te is the negative correla tion found in the central and east Tropical Pacific during th e winter months at lags of 0, 3, 6, and 9 months. This is the inverse of that seen for rainfall and streamfl ow, indicating greater de mand during drier periods (e.g. La Nia conditions). Positive correlations can be seen in the same region during MJJ and JJA at 3 – 12 month lags (Figure 37). However, these results may be suspect given the absence of these patterns in rainfall and streamflow. The winter composites of SSTs for years of high/low (80th/20th percentiles) seasonal demand (OND-MAM) tend to show the relationship with east and central tropical Pacific SSTs more clearly compared to the correlation patterns (F igures 40-42 and 49-51). However, since few years are used to create these composites they ar e subject to leveraging by a single year with a large anomaly. In MAM (Figure 42) a stronger de viation from mean conditions is seen for the 80th percentile of demand compared to the 20th percentile; the reverse of this can be seen in OND (Figure 49). During the period from MJJ through JAS large positive anomalies are seen for high demand years, but there does not appear to be an inverse relationship for low demand years (Figures 44-46). Correlations between SSTs and demand vary si gnificantly between WD PAs (Appendix A.3). Most WDPAs show negative correl ations with east and central tr opical Pacific SSTs, however two show little coherent patterns (City of Tampa and South-Central Hillsborough) and Pasco County self-supply shows positive correlations in these regions. Several WDPAs exhibit correlations (positive and negative) in the cen tral and east Tropical Pacific during warmer months (AMJ – SON). Given the variability f ound between WDPAs, the results of this report should be used with caution and merit further study in isolating climate influences on demand from other influences.

PAGE 47

47 Figure 36. Pearson's correlation of January March (left column), February April (center column), and March May (right column) detrended total regional demand with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 48

48 Figure 37. Pearson's correlation of April June (left column), May July (center column), and June August (right column) detrended total regional demand with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 49

49 Figure 38. Pearson's correlation of July September (left column), August October (center column), and September November (right column) detrended total regional demand with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 50

50 Figure 39. Pearson's correlation of October December (left column), November January (center column), and December February (right column) detrended total regional demand with concurrent and lagged sea surface temperatures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 51

51 Figure 40. Composite anomalies (C) of concurrent and lagged sea surface temperatures during January March for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 52

52 Figure 41. Composite anomalies (C) of concurrent and lagged sea surface temperatures during February April for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 53

53 Figure 42. Composite anomalies (C) of concurrent and lagged sea surface temperatures during March May for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 54

54 Figure 43. Composite anomalies (C) of concurrent and lagged sea surface temperatures during April June for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 55

55 Figure 44. Composite anomalies (C) of concurrent and lagged sea surface temperatures during May July for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 56

56 Figure 45. Composite anomalies (C) of concurrent and lagged sea surface temperatures during June August for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 57

57 Figure 46. Composite anomalies (C) of concurrent and lagged sea surface temperatures during July September for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 58

58 Figure 47. Composite anomalies (C) of concurrent and lagged sea surface temperatures during August October for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 59

59 Figure 48. Composite anomalies (C) of concurrent and lagged sea surface temperatures during September November for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 60

60 Figure 49. Composite anomalies (C) of concurrent and lagged sea surface temperatures during October December for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 61

61 Figure 50. Composite anomalies (C) of concurrent and lagged sea surface temperatures during November January for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 62

62 Figure 51. Composite anomalies (C) of concurrent and lagged sea surface temperatures during December February for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 63

63 SST Index selection and evaluation Based on the results of the correlation and com posite analyses, the most significant patterns occur with eastern and central tropical Pacific SSTs associated with ENSO. Two indices of spatially-averaged SSTs, the Nio 3 (5N-5S, 90-150W) and Nio 3.4 (5N-5S, 120-170W) indices were selected for evaluation. Both indices were derived from the Extended Reconstructed SST Version 3b dataset of Smith et al. (2008) and were obt ained from the Royal Dutch Meteorological Institute (KNMI) ( http://climexp.knmi.nl/ ). In addition, the Multivariate ENSO Index (MEI) was selected for evaluation and was obtained from the Climate Diagnostics Center (CDC, http://www.cdc.noaa.gov/people/ klaus.wolter/MEI/table.html ). The MEI is not strictly an SST index, but is a composite index that combines the ocean and atmosphere (Wolter and Timlin, 1993; Wolter and Timlin, 1998). Th e MEI combines SLP, zonal and meridional surface winds, SST, surface air temperature, and tota l cloudiness fraction of the sky into a single index. The drawback to the MEI is that data is only available from 1950, while other ENSO indices are considerably longer. Due to thei r historical record, other indices have been recommended for use in research (e.g. SOI, Nio 3, Nio 3.4) while the MEI has been recommended for operational use as an indicator of ENSO conditions (J.J. O’Brien, personal communication). Evaluation of seasonal SST indices Contour plots of lagged Pearso n’s correlation of seasonal mean standardized rainfall, mean standardized streamflow, and to tal demand are shown in Figure s 52, 53, and 54 for the Nio 3.4, Nio 3, and MEI indices, respectiv ely. All three show similar relationships, however the Nio 3.4 and Nio 3 appear to show stronger correlations and at earlier lead times compared to the MEI. Slight differences can be seen between the Nio 3 inde x and Nio 3.4 index. The Nio 3 index exhibits stronger correla tion with streamflow in DJF, while the Nio 3.4 index shows slightly stronger correla tion with demand in FMA, MAM, O ND, and NDJ at greater lags. The Spearman’s rank correlations were generally similar, though some seasons exhibited lower values (Appendix D). Evaluation of monthly SST indices Contour plots of lagged Pearso n’s correlation of monthly mean standardized rainfall, mean standardized streamflow, and to tal demand are shown in Figure s 55, 56, and 57 for the Nio 3.4, Nio 3, and MEI indices, respectively. Monthl y results are generally lower and less smooth compared to seasonal. For rainfall and stre amflow the Nio 3.4 and Nio 3 are quite similar, while the MEI shows lower correlation in D ecember. For demand the Nio 3.4 index shows higher correlation at greater lags in January an d February compared to the Nio 3 and MEI. Spearman’s rank correlations were similar and can be seen in Appendix D.

PAGE 64

64 Figure 52. Seasonal lagged correlation of the Nio 3.4 index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 65

65 Figure 53. Seasonal lagged correlation of the Nio 3 index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 66

66 Figure 54. Seasonal lagged correlation of the MEI index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 67

67 Figure 55. Monthly lagged correlation of the Nio 3.4 index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 68

68 Figure 56. Monthly lagged correlation of the Nio 3 index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 69

69 Figure 57. Monthly lagged correlation of the MEI index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 70

70 Relationships with Sea Level Pressures Rainfall and SLPs Correlation patterns between rainfall and SLPs were less stationary than those for SSTs, however significant and coherent pattern s were found between OND and FMA that are related to the Southern Oscillation (Figures 58-61). This is demonstrated by the correla tion in the vicinity of Tahiti (1740’S, 14927’W) and Darwin, Austra lia (1227’S, 13050’E), the two stations typically used to define the Southern Oscillat ion Index (SOI). The correlation patterns also support those that choose to use the equatorial SOI, which is calculated as the standardized anomaly of the difference of ar ea-averaged SLPs in the eastern Pacific (5N-5S, 80W-130W) and over Indonesia (5N-5S, 90E-140E). While th e equatorial SOI may be optimal at certain times of year, it is limited to th e length of the reanal ysis dataset (1949-prese nt). The correlation patterns are well defined up to 9-month lags in JFM and FMA and to 6-month lags in NDJ and DJF. The patterns in OND were less well defined. The Southern Oscillation is generally absent from the composites of SLPs during high/low seasonal rainfall (Figures 62-73), indicating that while a linear relationship may exist between rainfall and SLPs in the tropics as exhibited in the correlation maps, the anomalies are either small in magnitude or noisy and are thus not a prominent feature in the composite maps. The most prominent feature of the composite maps is the Aleutian Low which tends to exhibit lower than normal pressures during El Nio and higher than normal pressures during La Nia. These anomalies are most prevalent in years above the 90th percentile of seasonal rainfall and show relatively low persistence. The correlation patterns associated with individua l rain gauges vary somewhat in magnitude, but follow the same general pattern seen in Figures 58-61 (Appendix B.1). The correlation patterns are well defined up to 9-month lags in JFM and FMA and to 6-month lags in NDJ and DJF. The patterns in OND were less well defined.

PAGE 71

71 Figure 58. Pearson's correlation of January March (left column), February April (center column), and March May (right column) standardized rainfall with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 72

72 Figure 59. Pearson's correlation of April June (left column), May July (center column), and June August (right column) standardized rainfall with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 73

73 Figure 60. Pearson's correlation of July September (left column), August October (center column), and September November (right column) standardized rainfall with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 74

74 Figure 61. Pearson's correlation of October December (left column), November January (center column), and December February (right column) standardized rainfall with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 75

75 Figure 62. Composite anomalies (mb) of concurrent and lagged sea level pressures during January March for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 76

76 Figure 63. Composite anomalies (mb) of concurrent and lagged sea level pressures during February April for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 77

77 Figure 64. Composite anomalies (mb) of concurrent and lagged sea level pressures during March May for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 78

78 Figure 65. Composite anomalies (mb) of concurrent and lagged sea level pressures during April June for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 79

79 Figure 66. Composite anomalies (mb) of concurrent and lagged sea level pressures during May July for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 80

80 Figure 67. Composite anomalies (mb) of concurrent and lagged sea level pressures during June August for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 81

81 Figure 68. Composite anomalies (mb) of concurrent and lagged sea level pressures during July September for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 82

82 Figure 69. Composite anomalies (mb) of concurrent and lagged sea level pressures during August October for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 83

83 Figure 70. Composite anomalies (mb) of concurrent and lagged sea level pressures during September November for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 84

84 Figure 71. Composite anomalies (mb) of concurrent and lagged sea level pressures during October December for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 85

85 Figure 72. Composite anomalies (mb) of concurrent and lagged sea level pressures during November January for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 86

86 Figure 73. Composite anomalies (mb) of concurrent and lagged sea level pressures during December February for years of the 10 percent lowest (left column) and the 10 percent highest rainfall (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 87

87 Streamflow and SLPs Correlation patterns of mean standardized stream flow and SLPs generally follow those found for mean standardized rainfall (Fi gures 74-77), further confirming th e influence of ENSO on the region. However, as with SSTs the relationship is extended to the MAM season (Figures 74). Well defined patterns can be seen up to 9-month lags in JFM, FMA, and MAM and 6-month lags for NDJ and DJF. The patterns observed in the composites for st reamflow and SLPs (F igures 78-89) closely resemble those seen for rainfall and SLPs, with the Southern Oscillatio n generally absent and nonlinear anomalies observed in the Gulf of Alas ka during concurrent winter months. These anomalies are most prominent for high (above the 90th percentile) seasonal st reamflow. As with rainfall, other transient anomalies are exhibited in norther n latitudes. As with individual rain gauges, the correlation patterns associated with individual streamflow stations vary somewhat in magnitude, but follow the same general pattern seen in Figures 74-77 (Appendix B.2). The correlation patterns are well defined up to 9-month lags in JFM MAM and a lesser extent to 6-mont h lags in NDJ and DJF.

PAGE 88

88 Figure 74. Pearson's correlation of January March (left column), February April (center column), and March May (right column) standardized streamflow with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 89

89 Figure 75. Pearson's correlation of April June (left column), May July (center column), and June August (right column) standardized streamflow with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 90

90 Figure 76. Pearson's correlation of July September (left column), August October (center column), and September November (right column) standardized streamflow with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 91

91 Figure 77. Pearson's correlation of October December (left column), November January (center column), and December February (right column) standardized streamflow with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 92

92 Figure 78. Composite anomalies (mb) of concurrent and lagged sea level pressures during January March for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 93

93 Figure 79. Composite anomalies (mb) of concurrent and lagged sea level pressures during February April for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 94

94 Figure 80. Composite anomalies (mb) of concurrent and lagged sea level pressures during March May for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 95

95 Figure 81. Composite anomalies (mb) of concurrent and lagged sea level pressures during April June for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 96

96 Figure 82. Composite anomalies (mb) of concurrent and lagged sea level pressures during May July for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 97

97 Figure 83. Composite anomalies (mb) of concurrent and lagged sea level pressures during June August for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 98

98 Figure 84. Composite anomalies (mb) of concurrent and lagged sea level pressures during July September for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 99

99 Figure 85. Composite anomalies (mb) of concurrent and lagged sea level pressures during August October for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 100

100 Figure 86. Composite anomalies (mb) of concurrent and lagged sea level pressures during September November for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 101

101 Figure 87. Composite anomalies (mb) of concurrent and lagged sea level pressures during October December for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 102

102 Figure 88. Composite anomalies (mb) of concurrent and lagged sea level pressures during November January for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 103

103 Figure 89. Composite anomalies (mb) of concurrent and lagged sea level pressures during December February for years of the 10 percent lowest (left column) and the 10 percent highest streamflow (right column) between 1950 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 104

104 Demand and SLPs Correlation patterns between demand and SLPs ar e generally noisy, though a relationship with the Southern Oscillation can be seen during the wi nter months (Figures 90-93). The sign of the correlations with the Southern Oscillation are inverted compared to rainfall and streamflow. In the composites of SLPs for high/low seasona l demand a clearer patte rn emerges during the winter months in the Gulf of Alaska (Figures 94-105) compared to the correlation maps. The region is approximately the same as that seen fo r rainfall and streamflow but the anomalies are of the opposite sign. However, due to the few years used to constr uct these composites they are subject to leveraging. While the correlation of individual WDPAs with SLPs (Appendix B.3) are generally noisy and vary between WDPAs, similar patt erns found in Figures 90-93 can be seen. As with SSTs, the City of Tampa and South-Cent ral Hillsborough WDPAs show little relations hip with SLPs and the Pasco County self-supply WDPA shows corre lations of opposite sign compared to the remaining WDPAs.

PAGE 105

105 Figure 90. Pearson's correlation of January March (left column), February April (center column), and March May (right column) detrended total regional demand with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 106

106 Figure 91. Pearson's correlation of April June (left column), May July (center column), and June August (right column) detrended total regional demand with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 107

107 Figure 92. Pearson's correlation of July September (left column), August October (center column), and September November (right column) detrended total regional demand with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 108

108 Figure 93. Pearson's correlation of October December (left column), November January (center column), and December February (right column) detrended total regional demand with concurrent and lagged sea level pressures. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 109

109 Figure 94. Composite anomalies (mb) of concurrent and lagged sea level pressures during January March for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 110

110 Figure 95. Composite anomalies (mb) of concurrent and lagged sea level pressures during February April for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 111

111 Figure 96. Composite anomalies (mb) of concurrent and lagged sea level pressures during March May for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 112

112 Figure 97. Composite anomalies (mb) of concurrent and lagged sea level pressures during April June for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 113

113 Figure 98. Composite anomalies (mb) of concurrent and lagged sea level pressures during May July for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 114

114 Figure 99. Composite anomalies (mb) of concurrent and lagged sea level pressures during June August for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 115

115 Figure 100. Composite anomalies (mb) of concurrent and lagged sea level pressures during July September for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 116

116 Figure 101. Composite anomalies (mb) of concurrent and lagged sea level pressures during August October for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 117

117 Figure 102. Composite anomalies (mb) of concurrent and lagged sea level pressures during September November for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 118

118 Figure 103. Composite anomalies (mb) of concurrent and lagged sea level pressures during October December for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 119

119 Figure 104. Composite anomalies (mb) of concurrent and lagged sea level pressures during November January for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 120

120 Figure 105. Composite anomalies (mb) of concurrent and lagged sea level pressures during December February for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 3 month intervals from lag 0 (bottom row) to lag 12 (top row).

PAGE 121

121 SLP Index selection and evaluation The correlation and composite results of SLPs show the most significant patterns with the Southern Oscillation and to a le sser extent in the north Pacific. Two indices of the Southern Oscillation were chosen for evaluation. The fi rst is the ‘conventional’ Southern Oscillation Index (SOI) and is calculated as the gradient between standardized SLP anomalies at Tahiti (1740’S, 14927’W) and at Darwin, Australi a (1227’S, 13050’E) a nd was obtained from KNMI ( http://climexp.knmi.nl/ ). The second is the ‘equatori al’ Southern Oscillation Index (eqSOI) which is calculated as the standardized anomaly of the difference of area-averaged SLPs in the eastern Pacific (5N-5S, 80W-130W) and over Indonesia (5N-5S, 90E-140E) and was obtained from the Earth Syst ems Research Library (ESRL) ( http://www.cpc.ncep.noaa.gov/ data/indices/reqsoi.for ). To investigate the potential relationship with the north Pacific two additi onal indices were evaluated. The first is the North Pacific Index (NP) of Trenberth and Hurrell (1994) which is the area-averaged SLP over the region 30N65N, 160E-140W and was obtained from the Climat e Analysis Section of the National Center for Atmospheric Research ( http://www.cgd.ucar.edu/cas/jhurrell/npindex.html ). The second index is the Northern Oscillati on Index (NOI), which has been ar gued to provide a more direct representation of the influence of ENSO on North America as well as more local forcing mechanisms in the north Pacific (Schwing et al. 2002) and was obtai ned from the Pacific Fisheries Environm ental Laboratory ( http://www.pfeg.noaa.gov/products/PF EL/modeled/indices/NOIx/noix.html ). The NOI is calculated as the gradient of standardized SL P in the north Pacific (35N, 130W) and near Darwin, Australia (10S, 130E). While these c limate indices come from a variety of sources, they are also available from ESRL ( http://www.cdc.noaa.gov/data/climateindices/list/ ) where links are provided to their origin al sources (which were used in each case since they provided the most recently updated index values). Evaluation of seasonal SLP indices Contour plots of lagged Pearso n’s correlation of seasonal mean standardized rainfall, mean standardized streamflow, and to tal demand with the SOI and eqSO I indices are shown in Figures 106 and 107. Correlation patterns of the SOI a nd eqSOI are generally similar (as would be expected) and follow that of the ENSO SST indi ces (Figures 52-54). However, the eqSOI shows somewhat greater correlation over a larger portion of the winter months and exhibits greater lagged correlations compared to the SOI. Over all, the ENSO SST indices exhibit correlation of greater magnitude compared to the SOI and eqSOI. However, stronger correlations of the eqSOI occur earlier in the cooler months (e.g. OND) comp ared to the SST indices. Similar results were found using Spearman’s rank correlation (Appendix D). Similar plots of the NP and NOI indices are shown in Figures 108 and 109. Correlations with the NP index are generally low and showed very little indication of persistence. Correlations

PAGE 122

122 with the NOI are similar to that of the SOI a nd eqSOI, but are slightly lower in magnitude. Similar results were found using Spearma n’s rank correlation (Appendix D). Evaluation of monthly SLP indices Contour plots of lagged Pearso n’s correlation of monthly mean standardized rainfall, mean standardized streamflow, and tota l demand with the SOI, eqSOI, NP, and NOI indices are shown in Figures 110 113. As would be expected, corr elations are generally lo wer and less smooth for monthly values compared to seasonal. Correla tion patterns of monthl y indices are generally similar to those for seasonal indices, with the eqSOI showing the highest values at the largest lags. Similar results were found using Sp earman’s rank correlation (Appendix D).

PAGE 123

123 Figure 106. Seasonal lagged correlation of the SOI index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 124

124 Figure 107. Seasonal lagged correlation of the eqSOI index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 125

125 Figure 108. Seasonal lagged correlation of the NP index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 126

126 Figure 109. Seasonal lagged correlation of the NOI index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 127

127 Figure 110. Monthly lagged correlation of the SOI Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 128

128 Figure 111. Monthly lagged correlation of the eqSOI Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 129

129 Figure 112. Monthly lagged correlation of the NP Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 130

130 Figure 113. Monthly lagged correlation of the NOI Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 131

131 Relationships with 500 mb Geopotential Heights Rainfall and GpHs The correlation between GpHs and mean standardi zed rainfall show less significant persistence compared to SSTs and SLPs (Figures 114-117, note: lags are shown in 1-month increments compared to the 3-month increments used for SS Ts and SLPs). Significant positive correlations are seen during the winter months (NDJ-FMA) over th e tropics. This pattern is highly correlated with the SOI (an example is shown in Figure 11 8) and will not be addressed further since the SLPs related to the SOI have been shown previously in this report to exhi bit stronger correlations at greater lags compared to Gp Hs. Significant negative correlations are also seen in the north Pacific and over the southeast US, both of these are known to be cen ters of action of the PacificNorth American (PNA) pattern (Wallace and Gutzler, 1981). The most prominent features of the com posites of GpHs during years below the 10th and above the 90th percentiles of mean standardized seasonal rainfall are the centers of action of the PNA (Figures 119-130). Thes e centers of action coin cide with the point-wis e PNA index of Wallace and Gutzler (1981) which used 500 mb GpH near Hawaii (20N, 160W), the Aleutian Islands (45N, 165W), northwest Canada (55N, 1 15W), and New Orleans, LA (30N, 85W). Anomalies in the tropics are comparatively weak. These anomalies are present at lags up to 4 months. The correlation patterns of indi vidual gauges are relatively consistent (Appendix C.1) and verify the results shown in Figures 114 -117. In general, gauges with shorter datasets exhibited correlations of greater magnitude.

PAGE 132

132 Figure 114. Pearson's correlation of January March (left column), February April (center column), and March May (right column) standardized rainfall with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 133

133 Figure 115. Pearson's correlation of April June (left column), May July (center column), and June August (right column) standardized rainfall with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 134

134 Figure 116. Pearson's correlation of July September (left column), August October (center column), and September November (right column) standardized rainfall with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 135

135 Figure 117. Pearson's correlation of October December (left column), November January (center column), and December February (right column) standardized rainfall with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 136

136 Figure 118. Clockwise from upper left: Correlation of the July September SOI index with 500mb Geopotential heights in July September (0 month lag), October December (3 month lag of SOI), and January March (6 month lag of SOI).

PAGE 137

137 Figure 119. Composite anomalies (m) of concurrent and lagged geopotential heights during January March for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 138

138 Figure 120. Composite anomalies (m) of concurrent and lagged geopotential heights during February April for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 139

139 Figure 121. Composite anomalies (m) of concurrent and lagged geopotential heights during March May for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 140

140 Figure 122. Composite anomalies (m) of concurrent and lagged geopotential heights during April June for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 141

141 Figure 123. Composite anomalies (m) of concurrent and lagged geopotential heights during May July for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 142

142 Figure 124. Composite anomalies (m) of concurrent and lagged geopotential heights during June August for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 143

143 Figure 125. Composite anomalies (m) of concurrent and lagged geopotential heights during July September for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 144

144 Figure 126. Composite anomalies (m) of concurrent and lagged geopotential heights during August October for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 145

145 Figure 127. Composite anomalies (m) of concurrent and lagged geopotential heights during September November for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 146

146 Figure 128. Composite anomalies (m) of concurrent and lagged geopotential heights during October December for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 147

147 Figure 129. Composite anomalies (m) of concurrent and lagged geopotential heights during November January for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 148

148 Figure 130. Composite anomalies (m) of concurrent and lagged geopotential heights during December February for years of the 10 percent lowest (left column) and the 10 percent highest standardized rainfall (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 149

149 Streamflow and GpHs Correlations between GpHs and mean standardi zed streamflow (Figur es 131-134) follow the patterns found with rainfall with the largest pattern of correla tions occurring along the equator and spatially smaller correlations in the north Paci fic and southeast US (cen ters of action of the PNA). As with SSTs and SLPs, correlations ex tend into the MAM seas on for streamflow, one month longer than that seen for rainfall. As seen for rainfall, the most prominent featur es of the composites of GpHs during years below the 10th and above the 90th percentiles of mean standardized seasonal streamflow are the centers of action of the PNA (Figures 135-146). Anoma lies are larger in magnitude in MAM for the 90th percentile of streamflow (Figure 135) compared to rainfall (Figure 121). The correlation patterns of indi vidual stations are relatively c onsistent (Appendix C.2), however correlations tend to be stronger for shorter da tasets (e.g. Alafia at Bell Shoals, Hillsborough River at Morris Bridge, and the S160 structure) As seen with individual rainfall gauges, streamflow stations with s horter datasets exhibited corre lations of greater magnitude.

PAGE 150

150 Figure 131. Pearson's correlation of January March (left column), February April (center column), and March May (right column) standardized streamflow with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 151

151 Figure 132. Pearson's correlation of April June (left column), May July (center column), and June August (right column) standardized streamflow with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 152

152 Figure 133. Pearson's correlation of July September (left column), August October (center column), and September November (right column) standardized streamflow with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 153

153 Figure 134. Pearson's correlation of October December (left column), November January (center column), and December February (right column) standardized streamflow with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.231 is significant at p = 0.10.

PAGE 154

154 Figure 135. Composite anomalies (m) of concurrent and lagged geopotential heights during January March for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 155

155 Figure 136. Composite anomalies (m) of concurrent and lagged geopotential heights during February April for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 156

156 Figure 137. Composite anomalies (m) of concurrent and lagged geopotential heights during March May for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 157

157 Figure 138. Composite anomalies (m) of concurrent and lagged geopotential heights during April June for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 158

158 Figure 139. Composite anomalies (m) of concurrent and lagged geopotential heights during May July for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 159

159 Figure 140. Composite anomalies (m) of concurrent and lagged geopotential heights during June August for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 160

160 Figure 141. Composite anomalies (m) of concurrent and lagged geopotential heights during July September for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 161

161 Figure 142. Composite anomalies (m) of concurrent and lagged geopotential heights during August October for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 162

162 Figure 143. Composite anomalies (m) of concurrent and lagged geopotential heights during September November for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 163

163 Figure 144. Composite anomalies (m) of concurrent and lagged geopotential heights during October December for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 164

164 Figure 145. Composite anomalies (m) of concurrent and lagged geopotential heights during November January for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 165

165 Figure 146. Composite anomalies (m) of concurrent and lagged geopotential heights during December February for years of the 10 percent lowest (left column) and the 10 percent highest standardized streamflow (right column) between 1950 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 166

166 Demand and GpHs As with SSTs and SLPs, correlations are less consistent for total demand than for mean standardized rainfall and streamflow (Figures 14 7-150). Similar to the relationships with SSTs and SLPs, the correlation of demand with GpH is inverted in sign to that seen for rainfall and streamflow. In addition to the negative correlation along the equator between NDJ and FMA, correlation can be seen with cente rs of action of the PNA (particu larly in the north Pacific and southeast US). Anomalies in northern latitudes are more prominent in compos ites of low/high (20th and 80th percentiles) seasonal demand (Figures 151-162) compared to the correlation maps and approximate centers of action of the PNA. Howe ver, due to the relatively few years used to construct the composites they are prone to leveraging. Correlation patterns of member government WDPA s are variable (Appendi x C.3), but generally follow the patterns seen in Figures 147-150. Th e City of Tampa, South-Central Hillsborough, and St. Petersburg WDPAs show less correlation co mpared to the other WDPAs and the sign of the correlation for Pasco County self-supply is invert ed compared to the others as was also seen for SSTs and SLPs.

PAGE 167

167 Figure 147. Pearson's correlation of January March (left column), February April (center column), and March May (right column) detrended total regional demand with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 168

168 Figure 148. Pearson's correlation of April June (left column), May July (center column), and June August (right column) detrended total regional demand with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 169

169 Figure 149. Pearson's correlation of July September (left column), August October (center column), and September November (right column) detrended total regional demand with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 170

170 Figure 150. Pearson's correlation of October December (left column), November January (center column), and December February (right column) detrended total regional demand with concurrent and lagged geopotential heights. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row). Pearson correlation of 0.400 is significant at p = 0.10.

PAGE 171

171 Figure 151. Composite anomalies (m) of concurrent and lagged geopotential heights during January March for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 172

172 Figure 152. Composite anomalies (m) of concurrent and lagged geopotential heights during February April for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 173

173 Figure 153. Composite anomalies (m) of concurrent and lagged geopotential heights during March May for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 174

174 Figure 154. Composite anomalies (m) of concurrent and lagged geopotential heights during April June for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 175

175 Figure 155. Composite anomalies (m) of concurrent and lagged geopotential heights during May July for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 176

176 Figure 156. Composite anomalies (m) of concurrent and lagged geopotential heights during June August for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 177

177 Figure 157. Composite anomalies (m) of concurrent and lagged geopotential heights during July September for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 178

178 Figure 158. Composite anomalies (m) of concurrent and lagged geopotential heights during August October for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 179

179 Figure 159. Composite anomalies (m) of concurrent and lagged geopotential heights during September November for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 180

180 Figure 160. Composite anomalies (m) of concurrent and lagged geopotential heights during October December for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 181

181 Figure 161. Composite anomalies (m) of concurrent and lagged geopotential heights during November January for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 182

182 Figure 162. Composite anomalies (m) of concurrent and lagged geopotential heights during December February for years of the 10 percent lowest (left column) and the 10 percent highest total regional demand (right column) between 1991 and 2008. Lags are evaluated at 1 month intervals from lag 0 (bottom row) to lag 4 (top row).

PAGE 183

183 GpH Index selection and evaluation The correlation and composite results of GpHs s how significant relationships with the tropics and the centers of action of the PNA pattern. Since the relationship found in the tropics was shown to be highly connected to the SOI (Figur e 118, which also has stronger connections with rainfall, streamflow, and demand at larger lags), indices of the PNA were chosen for investigation since the PNA has been documente d to influence climate in the southeast US during winter and spring months (e.g. Leathe rs et al., 1991; Yin, 1994; and summarized by Martinez et al. 2009). Two indices of the PNA were chosen for eval uation. The first is the index defined by the Climate Prediction Center (CPC) us ing a rotated principal component analysis of 500mb GpH in the northern hemisphere and was obtained from the CPC ( ftp://ftp.cpc.ncep.noaa.gov/w d52dg/data/indices/tele_index.nh ). The second index, also obtained from the CPC ( http://www.cpc.ncep.noaa.gov/products /precip/CWlink/pna/month_pna_index2.shtml ), is the modified point-wise PNA which is calculated from 500mb GpH as: W) 90 W 70 N, 35 N 5 (2 W) 125 W 105 N, 60 N 45 ( W) 180 W 140 N, 50 N (40 W) 180 W 140 N, 25 N 15 ( mpPNA Evaluation of seasonal GpH indices Contour plots of lagged Pearso n’s correlation of seasonal mean standardized rainfall, mean standardized streamflow, and total demand with the PNA and mpPNA indices are shown in Figures 163 and 164. The correlati on patterns are generally sim ilar, exhibiting relatively low correlations and limited lags between OND and FM A. The mpPNA produces correlations that are slightly higher for streamflow, rainfall, a nd demand during the winter months compared to the PNA. Total regional demand shows high positive correlation with both the PNA and mpPNA during the summer months. However, thes e results may be spuri ous since they are not present in the rainfall and streamflow plots and the PNA pattern is known to only influence climate in the southeast during cooler months of the year. Similar results were found using Spearman’s rank correlation (Appendix D). Evaluation of monthly GpH indices As seen for other indices, plots of monthly co rrelations are lower and less smooth but generally follow the same patterns seen for seasonal plot s (Figures 165 and 166). Similar results were found using Spearman’s rank correlation (Appendix D).

PAGE 184

184 Figure 163. Seasonal lagged correlation of the PNA Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 185

185 Figure 164. Seasonal lagged correlation of the mpPNA Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 186

186 Figure 165. Monthly lagged correlation of the PNA Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 187

187 Figure 166. Monthly lagged correlation of the mpPNA Index with mean standardized rainfall (upper left), mean standardized streamflow (upper right), and total regional demand (bottom).

PAGE 188

188 Influence of the Atlantic Multidecadal Oscillation The Atlantic Multidecadal Oscillation has been shown to influence climate in Florida and may modulate the signal of ENSO (Asefa and Adams, 2008; En field et al., 2001; Kelly and Gore, 2008; Obeysekera et al., 2007). McCabe et al. (2004) identified a cold phase of the AMO between 1900-1925 (denoted AMO1), a warm pha se between 1930-1960 (AMO2), a cold phase between 1970-1990 (AMO3), and a warm phase from 1995-present (AMO4) (Figure 167). 0.6 0.4 0.2 0 0.2 0.4 0.6 1870189019101930195019701990Figure 167. Time series of the Atlantic Multidecadal Oscillation (AMO) index showing monthly values and 10 year running mean (obtained from the ESRL PSD: http://www.cdc.noaa.gov /data/timeseries/AMO/ ). Boxplots of water year (October – September) a nd seasonal mean rainfall and streamflow are shown in Figures 168 and 169. Notches in boxplots show 95% confidence intervals and in some cases are larger than the interquartile range. Me dians are significantly di fferent if the confidence intervals do not overlap. Rainfall and streamflow were not found to be significantly different (at p=0.10) between AMO phases using the one-way Kruskal-Wallis test. Due to the significant lagged corre lations of the Nio 3 and SOI i ndices found previously in this work, as well as the available le ngth of these indices, they were c hosen to evaluate differences in response during different phases of the AMO. Figures 170 and 171 show the lagged correlations of the Nio 3 index with mean st andardized rainfall and streamflow during the four time periods corresponding to warm and cold phases of the AMO. Significant diffe rences can be seen between the time periods, however no clear pattern common to each phase can be seen. Similar plots for the SOI can be seen in Figures 172 an d 173. Results are generally consistent between rainfall and streamflow, with the strongest co rrelations occurring in AM O4 and the weakest in AMO2. This correspondence sugge sts that the strength of the lagged relationship with a given index should be verified for each new application de pending on the data period used for analysis.

PAGE 189

189 Figure 168. Boxplots of mean total water year rainfall (top) and JFM, AMJ, JAS, and OND seasonal rainfall (clockwise from center left) by AMO phase. Notches in boxplots show 95% confidence intervals. AMO1: 1900 1925 (cold), AMO2: 1930 1960 (warm), AMO3: 1970 1990 (cold), AMO4: 1995 2008 (warm).

PAGE 190

190 Figure 169. Boxplots of mean standardized water year streamflow (top) and JFM, AMJ, JAS, and OND seasonal streamflow (clockwise from left) by AMO phase. Notches in boxplots show 95% confidence intervals. AMO2: 1932 1960 (warm), AMO3: 1970 1990 (cold), AMO4: 1995 2008 (warm).

PAGE 191

191 Figure 170. Correlation of the Nio 3 index with seasonal rainfall during the AMO1 cold phase 1900 1925 (upper left), AMO2 warm phase 1930 1960 (upper right), AMO3 cold phase 1970 1990 (lower left), and AMO4 warm phase 1995 2008 (lower right).

PAGE 192

192 Figure 171. Correlation of the Nio 3 index with seasonal streamflow during the AMO2 warm phase 1932 1960 (top), AMO3 cold phase 1970 1990 (lower left), and AMO4 warm phase 1995 2008 (lower right).

PAGE 193

193 Figure 172. Correlation of the SOI index with seasonal rainfall during the AMO1 cold phase 1900 1925 (upper left), AMO2 warm phase 1930 1960 (upper right), AMO3 cold phase 1970 1990 (lower left), and AMO4 warm phase 1995 2008 (lower right).

PAGE 194

194 Figure 173. Correlation of the SOI index with seasonal streamflow during the AMO2 warm phase 1932 1960 (top), AMO3 cold phase 1970 1990 (lower left), and AMO4 warm phase 1995 2008 (lower right).

PAGE 195

195 Summary and Recommendations Based on the lagged correlation and composite analyses of SSTs with seasonal rainfall, streamflow and demand, three in dices were identified as having the most relevance: the Nio 3.4, Nio 3, and the MEI Indices. Of these three, th e strongest relationships were found with the Nio indices based on lagged Pearson’s and Sp earman’s correlation. Both indices showed significant correlations between the OND and FMA seasons (MAM for streamflow) at leadtimes up to 9 months. Monthly co rrelations were generally weaker with significant correlations seen between November and March (and into April for streamflow). Analysis of sea level pressures identified four indices for investigation. These indices include the station-based Southern Oscillation Index (SO I), the reanalysis-based equatorial Southern Oscillation Index (eqSOI), the North Pacific I ndex (NP), and the Northern Oscillation Index (NOI). The SOI and eqSOI showed correlations of the greatest magnitude and at greater lags compared to the NP and NOI indices. Overall the eqSOI produced correlati ons that were slightly higher than the SOI. However the SOI may be th e index of choice in ce rtain circumstances due to the period of record available for this inde x compared to the eqSOI (1882-present compared to 1949-present). In general, the SOI and eqSOI produ ced similar patterns as the SST indices since both are indicators of ENSO. Correlations were slightly lower for the SOI and eqSOI compared to the SST indices, with the exception that th e eqSOI showed stronger correlations occurring earlier (e.g. OND) compared to the SST indices. Two indices of the Pacific North American (PNA) pattern were evaluated based on the analyses of 500mb geopotential heights. Only low correlations were found during winter months with little to no significant lagged correlations. The results of this analysis yiel ded significant results for indices of or re lated to ENSO, with significant relationships limited to the cooler mont hs of the year. Based on these results, it is recommended that either the Nio 3 or Nio 3.4 be used for forecasting in the region. However, in certain circumstances the SOI or eqSOI may be used in cases where measured gauge data (e.g. the SOI) are deemed more reliable compared to r econstructed SST datasets or if more accurate results can be derived using th e eqSOI to forecast hydrologic va riables during th e onset of the dry season (due to the stronger correlations exhibited in OND). Compared to the results for rainfall and streamfl ow the results for demand were more variable due to the relative shortness of the dataset and the assumption that non-climatic influences could be removed with a linear trend. However, simila r (but inverted) patterns associated with ENSO were found. Since many other fact ors influence demand a thorough anal ysis of this variable falls outside the scope this work, but it is recommended that further investigation be performed should a more detailed understanding of the infl uence of climate on demand be required.

PAGE 196

196 The influence of the AMO on the pattern of resu lts found here was investigated. Correlation patterns were found to differ between the four AMO periods corresponding to alternating warm and cool phases, however no clear pattern co mmon to each phase was found. No significant differences were found between the four AMO peri ods for annual (water-year) or seasonal total rainfall or mean standardized streamflow. Howe ver, differences may exist for individual stations or gauges as was found by Obeysekera et al. (2007) in south Florida. It is recommended that the resu lts found in this work be veri fied before being employed in operational use. In particular, it is recommended that the st rength of relationship between climate indices, such as Nio 3.4, Nio 3, SOI, or the eqSOI, and a hydr ologic variable (i.e. streamflow, rainfall, or demand) be verified based on the period of record used to develop operational forecasts or to train models to ensu re optimal results. This verification can be conducted easily using one or more of the online c limate analysis tools revi ewed in this report. In particular, the online plotting and analysis tools of the ESRL PSD and the Climate explorer of KNMI can be particularly useful in this rega rd since they contain several gridded climate datasets and indices. In additi on, both of these tools allow user-def ined time series to be used in analyses and, therefore, lend themselves conducive for produc ing results with modified periods of record for incorporation into operational forecast models for specified time series.

PAGE 197

197 References Asefa, T. and A. Adams. 2008. Investigation of intra-seasonal to multi-decadal variability in West-Central Florida rainfall. Report a nd Presentation. October 20, 2008. Gainesville, FL. Barnett, T.P. and R. Priesendorfer. 1987. Origin s and levels of monthly and seasonal forecast skill for United States surface air temperatures determined by canonical correlation analysis. Monthly Weathe r Review 115: 1825-1850. Enfield, D.B., Mestas-Nuez, A.B., and P.J. Tr imble. 2001. The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S. Geophysical Research Letters 28(10): 2077-2080. Gershunov, A. and T.P. Barnett. 1999. Interdeca dal modulation of ENSO teleconnections. Bulletin of the American Meteor ological Society 79(12): 2715-2725. Grantz, K., Rajagopalan, B., Clark, M., and E. Zagona. 2005. A technique for incorporating large-scale climate information in basin-sc ale ensemble streamflow forecasts. Water Resources Research 41, W10410, doi:10.1029/2004WR003467. Harshburger, B., Ye, H., and J. Dzialoski. 2002. Observational evidence of the influence of Pacific SSTs on winter precipita tion and spring stream discharge in Idaho. Journal of Hydrology 264: 157-169. Horel, J.D., and J.M. Wallace. 1981. Planetary-scal e atmospheric phenomena associated with the Southern Oscillation. Monthl y Weather Review 109(4): 813–829. Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woolen, J., Zhu, Y., Chellia h, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K.C., Ropelewski, C., Wang, J., Leet maa, A., Reynolds, R., Jenne, R., D. Joseph. 1996. The NCEP/NCAR 40-year reanalysis project. Bulletin of the American Meteorological Society 77(3): 437-471. Kelly, M.H. and J.A. Gore. 2008. Florida river flow patterns and the atlantic multidecadal oscillation. River Research Applications 24: 598-616. Leathers, D.J., Yarnal, B., and M.A. Palecki. 1991. The Pacific/North American pattern and United States climate. Journal of Climate 4(5): 517–528. Martinez, C.J., Baigorria, G.A., and J.W. Jone s. 2009. Use of climate i ndices to predict corn yields in southeast USA. Internati onal Journal of Climatology, In Press.

PAGE 198

198 McCabe, G.J., Palecki, M.A., and J.L. Betancou rt. 2004. Pacific and Atlantic Ocean influences on multidecadal drought frequency in the Un ited States. Proceedings of the Natural Academy of Sciences 101(12): 4136-4141. Obeysekera, J., Trimble, P., Neidrauer, C., Path ak, C., VanArman, J., Strowd, T., and C. Hall. 2007. Appendix 2-3: Consideration of long-term climatic variability in regional modeling for SFWMD planning and operations. 2007 S outh Florida Environmental Report. https://my.sfwmd.gov/pls/portal/docs/P AGE/PG_GRP_SFWMD_ SFER/PORTLET_PRE VREPORT/VOLUME1/appendices/v1_app_2-3.pdf Oplitz-Stapleton, S., Gangopadhyay, S., and B. Rajagopalan. 2008. Generating streamflow forecasts for the Yakima River Basin using large-scale climate pr edictors. Journal of Hydrology 341: 131-143. Ropelewski, C.F. and M.S. Halpert. 1986. No rth American precipitation and temperature patterns associated with the El Nio/Sout hern Oscillation (ENSO). Monthly Weather Review 114: 2352-2362. Schwing, F.B., Murphree, T., and P.M. Green. 2002. The northern oscillation index (NOI): A new climate index for the northeast Paci fic. Progress in Oceanography 53: 115-139. Smith, T.M. and R.W. Reynolds. 2004: Improve d Extended Reconstruction of SST (1854-1997). Journal of Climate 17: 2466-2477. Smith, T.M., Reynolds, R.W., Peterson, T.C., and J. Lawrimore. 2008. Improvements to NOAA's historical merged land-ocean surface temperature analysis (1880-2006). Journal of Climate 21: 2283-2296. Sveinsson, O.G.B., Lall, U., Gaudet, J., Kushnir, Y., Zebiak, S., and V. Fortin. 2008. Analysis of climate states and atmospheric circula tion patterns that in fluence Qubec spring streamflows. Journal of Hydrol ogic Engineering 13(6): 411-425. Tootle, G.A. and T.C. Piechota. 2006. Relations hips between Pacific a nd Atlantic sea surface temperatures and U.S. streamflow variab ility. Water Resources Research 42, W07411, doi: 10.1029/2005WR004184. Trenberth, K.E. and J.W. Hurrell. 1994. Decadal at mosphere-ocean variations in the Pacific. Climate Dynamics 9: 303-319. von Storch, H. and F.W. Zwiers. 1999. Statistical Analysis in Climate Research. Cambridge University Press.

PAGE 199

199 Wallace J.M. and D.S. Gutzler. 1981. Teleconnecti ons in the geopotential height field during the northern hemisphere winter. Mo nthly Weather Review 109: 784–812. Wolter, K. and M.S. Timlin, 1993. Monitoring EN SO in COADS with a seasonally adjusted principal component index. Proceedings of the 17th Climate Diagnostics Workshop, Norman, OK, NOAA/NMC/CAC, NSSL, Okla homa Clim. Survey, CIMMS and the School of Meteorology., Univ. of Oklahoma, 52-57.http://www.cdc.noaa.gov/people/klaus.wolter/MEI/WT1.pdf Wolter, K. and M. S. Timlin. 1998. Measuring th e strength of ENSO ev ents how does 1997/98 rank? Weather 53: 315-324. Yin, Z.Y. 1994. Moisture condition in the sout h-eastern USA and tele connection patterns. International Journal of Climatology 14(9): 947–967.