Irrigation Demand. Florida Statewide Agricultural. Appendix E Technical Information

Transcription

1 2017 Florida Statewide Agricultural Irrigation Demand Appendix E Technical Information THE BALMORAL GROUP 165 Lincoln Ave Winter Park, FL

2 Appendix E Technical Information List of Tables Table E 1. Statewide Metered Data Records Statistics by Crop...E 1 Table E 2. Metered Data Records Statistics by District...E 2 Table E 3. Statistical Output and Coefficients of Analytical Model...E 3 Table E 4. Projected Crop Prices and Costs...E 4 Table E 5. Differences in Rainfall by Crop...E 6 Table E 6. Difference in Calculations from Model Dataset and Entire 2015 Projection ILG...E 7 Table E 7. Florida Department of Citrus Projection Scenarios...E 9 Table E 8. Forecast Citrus Acreage: FSAID Comparison to DOC...E 10 Table E 9. Dry to Average Year Ratio...E 12 Table E 10. Median Metered Irrigation Intensity Reported by Crop, E 18 List of Figures Figure E 1. Application of Analytical Model...E 4 Figure E 2. Modeled and Measured 2015 MGD, Top 10 Permits by WMD E 7 Figure E 3. Modeled and Measured 2015 MGD Permit Comparison E 8 Figure E 4. Example of ALG Polygons Edited with DEP Statewide Land Use Layer...E 14 Figure E 5. Example of ALG Polygons Edited by Changes in Aerial Imagery...E 14 Figure E 6. Example of ALG Polygons Edited with USDA Cropland Data Layer...E 15 Figure E 7. Shift in Types of Irrigation Equipment...E 17 Figure E 8. FRIS Data of Irrigation Amount (MGD) per Irrigated Acre...E 17 Figure E 9. Methodology for Applying FRIS derived Efficiency Improvement Slopes...E 18

3 Table E 1. Statewide Metered Data Records Statistics by Crop Primary Crop All Years Count INYR Count INYR Count INYR Count INYR Count INYR Count INYR Count INYR Count INYR Citrus , , , , Field Crops , Fruit (Non citrus) , Greenhouse/Nursery , Hay Potatoes Sod Sugarcane Vegetables (Fresh Market) , , , , All Crops (n) 1, , , , , , E 1

4 Table E 2. Metered Data Records Statistics by District WMD Primary Crop Acres MGD INYR NWFWMD Field Crops 38, NWFWMD Greenhouse/Nursery 3, NWFWMD Hay 2, NWFWMD Potatoes NWFWMD Sod 3, NWFWMD Vegetables (Fresh Market) 2, SFWMD Citrus 427, SFWMD Field Crops 3, SFWMD Fruit (Non citrus) 1, SFWMD Greenhouse/Nursery 7, SFWMD Hay 93, SFWMD Potatoes 12, SFWMD Sod 34, SFWMD Sugarcane 139, SFWMD Vegetables (Fresh Market) 173, SJRWMD Citrus 25, SJRWMD Field Crops 9, SJRWMD Fruit (Non citrus) 2, SJRWMD Greenhouse/Nursery 12, SJRWMD Hay 26, SJRWMD Potatoes 30, SJRWMD Sod 15, SJRWMD Vegetables (Fresh Market) 13, SRWMD Field Crops 46, SRWMD Fruit (Non citrus) 1, SRWMD Greenhouse/Nursery SRWMD Hay 13, SRWMD Potatoes 1, SRWMD Vegetables (Fresh Market) 10, SWFWMD Citrus 459, SWFWMD Field Crops 9, SWFWMD Fruit (Non citrus) 26, SWFWMD Greenhouse/Nursery 6, SWFWMD Hay 8, SWFWMD Potatoes 4, SWFWMD Sod 18, SWFWMD Vegetables (Fresh Market) 117, Total 1,800,424 1, E 2

5 Table E 3. Statistical Output and Coefficients of Analytical Model Ordinary least squares regression Weighting variable = none Dep. var. = INYRAC Mean= , S.D.= Model size: Observations = 22120, Parameters = 22, Deg.Fr.= Residuals: Sum of squares= e+11, Std.Dev.= Fit: R squared= , Adjusted R squared = Model test: F[ 21, 22098] = , Prob value = Diagnostic: Log L = ***********, Restricted(b=0) Log L = LogAmemiyaPrCrt.= , Akaike Info. Crt.= Autocorrel: Durbin Watson Statistic = , Rho = Variable Coefficient Standard Error b/st.er. P[ Z >z] Mean of X Constant SFAC (2.26) 0.46 (4.91) SJRAC SRAC SWFAC (2.15) 0.46 (4.67) RFAC (0.56) 0.02 (23.62) 4, RFETAC , CITAC FIELDAC (1,299.40) (2.87) FVEGAC FRUAC 1, GRENAC 1, HAYAC (15,636.39) 1, (14.70) 0.00 POTAC SODAC SUGARAC 11, CPIVAC CNURAC DRIPAC GRAVAC IMPSAC MICROAC The econometric model is used to generate coefficients from the metered data records. The coefficients are used to estimate water use for the remaining Florida farms that do not have metered data. The following example is provided to demonstrate the process. Assume a Citrus farm in Manatee County in the Southwest Florida Water Management District, with irrigated area of 6.19 acres. The farm has Microspray irrigation, the recorded average annual rainfall was inches, and the recorded average annual ET was The estimated net revenue was $9,238. A summary of the GIS attribute values for the specified farm are provided in Figure E 1. Applying the coefficients from FSAID VI to this specific farm s GIS attributes produces an estimated inches per acre per year. E 3

6 Figure E 1. Application of Analytical Model GIS Attribute Value for Demonstration Farm Name Variable (all values are for demonstration farm) Value ID Field ID Mana_31120 SWFAC Indicator of 1 if field is in SWFWMD, times field acres RFAC Rainfall (in inches, average annual, per nearest rain station), times field acres 3, RFETAC RainET interaction term (rain times ET), times field acreage 190, CITAC Net Revenue in millions (Net Price $1, times field acres, divided by one million) 0.11 MICROAC Indicator of 1 if field irrigation type is Micro Spray, times field acres Calculation of Farm level Estimated Water use for GIS assignment Name Field Specific Value Coefficient Product CONSTANT NA SWFAC RFAC 3, , RFETAC 190, , CITAC MICROAC ACREIN (Total in Acre Inches per Year) GALMIL (Total in Millions of Gallons per Year) 0.06 Estimated Inches/Acre/Year (GALMIL/365/72 acres* ) Table E 4. Projected Crop Prices and Costs Primary Crop Unit Citrus Crop Price $3,223 $2,923 $3,574 $4,291 $5,021 $5,733 $6,445 Chem. Cost $1,301 $1,431 $1,741 $2,117 $2,379 $2,674 $3,004 Field Crops Crop Price $665 $754 $684 $720 $753 $790 $826 Chem. Cost $253 $278 $313 $352 $370 $388 $408 Vegetables (Fresh Market) Crop Price $6,964 $6,951 $8,681 $9,912 $11,047 $12,598 $14,071 Chem. Cost $1,159 $1,275 $1,433 $1,611 $1,693 $1,779 $1,870 Greenhouse/Nursery Crop Price $16,416 $15,240 $14,306 $14,055 $13,866 $13,715 $13,589 Chem. Cost $547 $595 $669 $752 $790 $830 $873 Hay Crop Price $267 $422 $368 $370 $384 $399 $414 Chem. Cost $152 $167 $187 $211 $221 $233 $244 Fruit (Non citrus) Crop Price $5,092 $7,731 $8,323 $9,049 $9,719 $10,631 $11,496 Chem. Cost $1,340 $1,474 $1,793 $2,181 $2,451 $2,754 $3,095 Potatoes Crop Price $4,020 $3,324 $3,770 $4,267 $4,762 $5,246 $5,731 Chem. Cost $424 $466 $524 $588 $618 $650 $683 Sod Crop Price $4,473 $6,931 $4,356 $5,096 $5,836 $6,575 $7,315 Chem. Cost $857 $932 $1,048 $1,177 $1,238 $1,301 $1,367 Sugarcane Crop Price $1,414 $1,313 $1,442 $1,608 $1,837 $2,007 $2,177 Chem. Cost $408 $443 $498 $560 $588 $618 $650 E 4

7 FSAID Methodology As required by Florida Statute, observations of irrigation water use form the basis for the FSAID estimates of spatially distributed statewide irrigation demand. Metered and reported water use data is collected each year from the water management districts. Historical metered data extends from 2007 to The data is inspected to remove estimated records and outliers. District datasets are then formatted into a single statewide dataset. New permits are identified and added to the GIS spatial dataset, using updated permit boundary shapefiles supplied by the districts. Modified CUP permit data is similarly reviewed and GIS values updated as appropriate. Parameter values needed for estimating water demand are assigned to the newly added irrigated fields, including rainfall (assigned by latitude/longitude using a 2 km gridded dataset), evapotranspiration (ET; also assigned by latitude/longitude on a 2km grid), crop, irrigation system, and irrigated acres. Where metered data within a permit reflects multiple fields with multiple crops, water use is apportioned using mean AFSIRS inches/acre by crop. Quality control processes are completed before exporting GIS data in tabular format for analytical modeling. The model specification is estimated using Ordinary Least Squares regression analysis to generate coefficients from the actual water use for each field level variable. Variables include rainfall (assigned by Latitude/Longitude on a 4km grid), ET (assigned by Latitude/Longitude on a 2km grid), net price (crop specific average farmgate prices less IFAS/UGA average chemical costs), irrigation system type (gravity/seepage, microspray/microjet, center pivot, traveling gun, drip, container/nursery and impact sprinkler) and location attributes including Water Management District dummies; all input variables are scaled by field specific irrigated acres. In the 2017 model, approximately 22,000 farm observations are included in the model used to generate coefficients (including all years). The R 2 or statistical fit of the model output to actual data is After the model is run on the metered input dataset, the output coefficients are applied to each irrigated field in the ILG (approximately 26,000). The current year irrigation water use estimate is based rainfall and ET from the year matching the most recent water use data (in this case, 2015). For projection years, the field level variables are updated to reflect changes in five year increments based on countylevel declines or expansions in agricultural acreage, and crop changes indicated by long run worldwide agricultural price forecasts. Average rain and evapotranspiration ( ) values are used for projections and assigned to each field. If a crop change is indicated in a given period, a new net price is assigned to match the updated crop type. If the crop type does not change, the net price is simply updated to the new period s value. To simulate future conditions, the coefficients are applied to the updated field level values and aggregated. Water use is then calculated for each irrigated field. If the water use for a given irrigated field exceeds the prior period s water use, a portion of the water is retained for assignment to newly added acreage in future periods. In counties with declining acreage, irrigated fields are randomly selected for deletion by period, removing the associated acreage and water use from the model. E 5

8 Sensitivity Analysis Metered Water Use Metered or reported water use data was obtained from each Water Management District. Data was cleaned using a number of checks prior to use in the final dataset. Tukey inter quartile range (IQR) analysis, measurement of the standard deviation from the mean, and tercile analysis were processes used to discover outliers. Tukey analysis uses 1.5 times the IQR from the first and third quartiles to determine the range. The standard deviation methodology uses two standard deviations from the mean to identify potential outliers. It was determined that the 25 th percentile for the lower bound and the 90 th percentile for the upper bound appropriate to identify outliers in the metered data based on irrigation in/yr. These ranges were applied to crop irrigation intensity measures of inches per acre for metered measurements of water use and acreage. These variables were chosen over total water use, since outliers for this variable would depend heavily on farm size. In addition, analysis was conducted by crop type to identify outliers in irrigation intensity among crops with widely varying water application rates. Rainfall Because rainfall replaces irrigation water, it is an important component of water demand. Rain gage data was used in the initial FSAID I estimates. However, in some districts, this resulted in very few location specific measures of precipitation. Starting with FSAID II, NOAA s Advanced Hydrologic Prediction Service (AHPS) data has been used instead. This set of data is provided in the form of daily point shapefiles for the entire continental USA and Puerto Rico. Within Florida, there are over 12,600 points spaced approximately 3.5 to 4 km apart (point spacing decreases from north to south). The available period of record is from 2005 to 2015, or eleven years. Analysis was performed to compare the estimated net irrigation demand using both actual 2015 rainfall data and an average of 11 years ( ). Table E 5 shows the differences in annual rainfall by crop for the two time periods of rainfall data; overall, results varied by less than 1%. Average rainfall was used for the projection period of 2020 to Table E 5. Differences in Rainfall by Crop 2015 MGD 2015 MGD Crop 2015 Rainfall Avg Rainfall ( ) Difference Citrus % Field Crops % Fruit (Non citrus) % Greenhouse/Nursery % Hay % Potatoes % Sod % Sugarcane* % Vegetables (Fresh Market) % Total 2, , % E 6

9 Evapotranspiration For purposes of assessing sensitivity of the ET variable inputs for the model output versus the larger dataset, the mean ET was calculated by crop, using the model dataset and also using the entire 2015 projection ILG dataset (which is calculated with an 11 year average of ET). Table E 6 shows the results. Differences will reflect the spatial variation in the model dataset as contrasted with the greater distribution in the larger ILG dataset; overall the difference is within 2.3%. Table E 6. Difference in Calculations from Model Dataset and Entire 2015 Projection ILG Statewide ET (IN/YR) ET (IN/YR) Difference Crop Metered Dataset 2015 Projection ILG Citrus % Field Crops % Fruit (Non citrus) % Greenhouse/Nursery % Hay % Potatoes % Sod % Sugarcane % Vegetables (Fresh Market) % Total % Model Performance The water use model has an R 2 is in the high 0.70 range, and when residuals are calculated at the scale of county or planning area, tend to wash out such that area totals generally replicate actual water use well. For purposes of evaluating the model for particular farms, Figure E 2 shows a comparison of the 10 permits with the largest 2015 reported/metered water use compared to the FSAID model estimates for those permits. Total differences (modeled reported) aggregated for all 10 permits within each District range from 1.8 MGD to 0.63 MGD. Figure E 2. Modeled and Measured 2015 MGD, Top 10 Permits by WMD E 7

10 A plot of modeled and metered/reported data by permit for 2015 are shown in Figure E 3 with a one to one line; this provides a graphical assessment of model performance as seen in the general clustering near the line. Figure E 3. Modeled and Measured 2015 MGD Permit Comparison E 8

11 Citrus Evaluation Citrus irrigation is the greatest variable in long term projections of Florida agricultural irrigation demand, both in acreage and in water demand. Annual average citrus prices are published and forecast by USDA, but only historical pricing is provided at the state level. While historically a higher share of Florida s citrus tends to be sold as processed than fresh, producers report that fresh sales are increasing in the last few years. Processed fruit prices have averaged about 66% of national prices for the past ten years. Using USDA forecasts, citrus acreage and water demand would increase significantly over the next 25 years. Given the issues relating to greening in Florida, and intense foreign competition, there is significant uncertainty about future citrus production in Florida. Production plummeted 24% in the past year, and while short term production was expected to decline before recovering in later years, the precipitous decline was not incorporated into Department of Citrus or USDA outlooks. In the prior year report, a long term decline in citrus acreage over 25 years was projected, with variation in between, but a net effect of 11% fewer acres by the end of the planning period. The projections took into account a confidence interval on price forecasts (further detail below), a linear trend on price forecast in the outer years, and an upper bound on overall production. Extensive discussion was held with academic experts in citrus production forecasting, Florida Department of Citrus Chief Economist Marisa Zansler, and producers, resulting in the following components of citrus evaluation: Department of Citrus forecasts production based on multiple scenarios that reflect potential changes in replanting rates (50%, 100%, 125% and 255% the latter considered the recovery rate ) and yield (184 boxes/acre and 200 boxes/acre); see Table E 7. The three higher replanting scenarios are reported as likely. The average of the three, and the 125% replanting scenario, were used along with both yield levels as comparison to the FSAID model s predictions of acreage. The lowest production is reflected in the fourth combination (125% replanting rate, 184 boxes; see Table E 8. The FSAID model predicts less acreage than DOC in each case. As a result, no upper bound was used on citrus acreage projections. Table E 7. Florida Department of Citrus Projection Scenarios Florida Department of Citrus (DOC) Projection Scenarios (bold considered likely scenarios) Declining Yields@ 100% plant rate Constant 50% plant rate Constant 100% plant rate Constant 125% plant rate Constant 255% plant rate Increasing 50% plant rate Increasing 100% plant rate Increasing 125% plant rate Increasing 255% plant rate E 9

12 Table E 8. Forecast Citrus Acreage: FSAID Comparison to DOC Scenario 4: 184 Boxes per Acre at 125% Planting Rate Citrus Acreage FSAID Predicted 577, , , , , ,284 DOC Implied 577, , , , , ,008 %CHG 0% 10% 13% 19% 25% 32% Difference 54,955 72, , , ,724 The long term accuracy of prior price estimates was evaluated. Previous estimates of citrus prices per box were researched; on average, the first three years of forecasts tended to underestimate prices by a few percentage points, while the later years tended to overestimate. In general, repeated forecasts reflected about a 13% overestimate for forecasting periods more than five years out from forecast date. A confidence interval of 13% was applied to the forecast prices beginning with The official forecast period for the Department of Citrus extends ten years and reflects price increases of just over 1% per year. In last year s report, numerous algorithms were tested to extend price forecasts beyond the official forecast period using a variety of trending algorithms. Linear and power trend forecasts tended to most closely track the official forecast trajectories. In the end, based largely on input from Dr. Zansler (who produces the Department of Citrus forecasts), a linear trend was used to extend the forecast to The same process was used this year. The current ILG reflects a reduction of 50,000 acres in Citrus for current conditions. Over the short run, the decline is likely to continue, but newer nursery stocks that exhibit more resilience to greening and other pests are beginning to be used in replantings such that within 5 7 years, acreage is expected to stabilize and increase. The ending conditions vary little 566,000 acres versus 555,000 acres in the prior forecast. However, because the beginning point is lower, the net decline appears smaller 4% versus 11% previously. The net change in water volume is an increase of 4%. E 10

13 FSAID Methodology for Dry Year Water Demand Estimation Florida statute requires agricultural demand estimates be developed for use in regional water supply planning to include average year and 1 in 10 year drought demands. It is not clearly defined if the 1 in 10 refers to field specific annual rainfall or to the simulated irrigation demand. In FSAID IV, the approach is to base the dry year estimates on modeled irrigation demand that occurs with a 1 in 10 frequency. This approach reflects the reality that the years of high irrigation demand might correspond to years with normal or above average rainfall because of the difference in timing of rainfall events and cropping seasons. The relevant water supply planning statute language is as follows: From Statue Regional water supply planning: A quantification of the water supply needs for all existing and future reasonable beneficial uses within the planning horizon. The level of certainty planning goal associated with identifying the water supply needs of existing and future reasonable beneficial uses must be based upon meeting those needs for a 1 in 10 year drought event. Agricultural demand projections used for determining the needs of agricultural self suppliers must be based upon the best available data. In determining the best available data for agricultural self supplied water needs, the district shall consider the data indicative of future water supply demands provided by the Department of Agriculture and Consumer Services pursuant to s and agricultural demand projection data and analysis submitted by a local government pursuant to the public workshop described in subsection (1), if the data and analysis support the local government s comprehensive plan. Any adjustment of or deviation from the data provided by the Department of Agriculture and Consumer Services must be fully described, and the original data must be presented along with the adjusted data From Statue Department of Agriculture and Consumer Services; agricultural water conservation and agricultural water supply planning: Estimates of historic and current water demands must take into account actual metered data as available. The 1 in 10 dry year demand is estimated in FSAID IV using the modeled irrigation demand that occurs with expected frequency of 1 in 10. This corresponds to the 90th percentile irrigation demand over the 17 years of record from The annual irrigation demand for each of the ILG polygons was simulated using AFSIRS, and the 17 years of annual results were used to calculate the 1 in 10 irrigation demand. A ratio of dry year (1 in 10 irrigation demand) to average year irrigation demand as simulated by AFSIRS was calculated for each polygon and was then summarized by crop group within each District. The FSAID model, based on the field level historical water use, is used to estimate the average year irrigation demand. Estimates of Dry Year Water Demand are calculated using a ratio of dry to average 1 in 10 irrigation needs. Based on a statewide batch run of AFSIRS, the 90 th percentile irrigation demand by crop was determined. The dry to average ratios used throughout the estimates are shown in Table E 9. E 11

14 Table E 9. Dry to Average Year Ratio Primary Crop NWFWMD SFWMD SJRWMD SRWMD SWFWMD Statewide NWFWMD Citrus Field Crops Fruit (Non citrus) Greenhouse/Nursery Hay Potatoes Sod Sugarcane Vegetables (Fresh Market) Statewide overall 1.34 E 12

15 ALG Statewide Revision Description and Methodology There were three major updates to the ALG for FSAID IV. The first process was to remove all of or portions of nonagricultural features to reflect recent land use changes. The second process was to add ALG features (agricultural areas) that were not previously in the FSAID ALG. The third process was to provide an updated, uniform crop type classification for the ALG. The Department of Revenue (DOR 2015) parcel data was used in order to remove non agricultural areas from the ALG. The Cropland Data Layer (CDL, from USDA) was also used to help identify non agricultural features. Figures E 4 through E 6 below illustrate some examples of ALG features removed as result of being identified as non agricultural based on DOR parcels or CDL. Substantial quality control and manual review was required in order to assess agricultural areas in parcels without an agricultural description. For example, vacant commercial, governmental, institutional, and other parcel land uses may have agricultural lands even though the parcel legal description is not agricultural. This means that the ALG updates were largely based on decision rules linked to parcel descriptions, but there are numerous instances in which a judgment based on aerial imagery was required. Adding agricultural areas (where the previous ALG version from FSAID III did not have agriculture features) was done based on the DOR 2015 parcel data, the newest ( ) Statewide Landuse Layer, and the SFWMD land use updates in UEC, LWC, UKB, LKB. There were very few ALG features added that were not already in the FSAID III ALG. The bulk of the spatial changes resulted from land use changes recognized using the DOR 2015 parcels. Classifying the crop types in the ALG was based on a hierarchy of spatial datasets. Attributes in the ALG describe the crop type and the source of the crop type classification. The following list identifies the spatial datasets (in order of decreasing ranking in terms of hierarchy) that was used to assign agricultural land use descriptions. This means that wherever multiple data sources overlap, the one closest to the top of the list was used. In some places, manual review was required to correctly classify crop type. FSAID ILG crop for currently irrigated areas in the ALG SFWMD land use layers in the following regions: UEC, LWC, UKB, LKB SJRWMD 2015 Ag Layer FLUCCS land use from Statewide Land Use Land Cover FDOR parcel data CDL 2015 Manual assignment using aerial imagery (NAIP 2015) or Division of Plant Industry (DPI) active citrus to identify abandoned citrus E 13

16 Figure E 4. Example of ALG Polygons Edited with DEP Statewide Land Use Layer ALG shape (not irrigated); corresponds to agricultural area from DEP statewide LU layer; 4/1/2014 aerial imagery shown Figure E 5. Example of ALG Polygons Edited by Changes in Aerial Imagery ALG shape (ALG_6585) removed (resulting from intersect with non ag parcels from DOR 2015); 2/4/2016 aerial imagery shown E 14

17 Figure E 6. Example of ALG Polygons Edited with USDA Cropland Data Layer Manatee County, ALG_6703: identified for ALG removal using CDL area classified as developed (gray color). E 15

18 Irrigation efficiency improvements The efficiency of an irrigation system represents the relationship between the amount of irrigation water retained in crop root zones and the total amount of water withdrawn from a source for irrigation. Improvements to irrigation efficiency are generally the result of the following changes: Management or scheduling improvements (possibly resulting from implementation of soil moisture sensors or weather stations or other monitoring equipment) Equipment change to irrigation equipment that reduces non beneficial water use Equipment maintenance and upgrades that improve uniformity and/or efficiency (resulting from nozzle replacement, adding pressure regulators, or other equipment improvements) Estimating future improvements in irrigation efficiency requires estimating the field level improvements in efficiency and estimating the number of fields or farms that are expected to achieve those efficiency improvements. The advantage of this approach is that it can be linked to on farm data where irrigation efficiency improvements have been made in the past. However, a challenge with this approach is estimating the numbers of farms that can be expected to realize similar efficiency improvements in the future. The approach used in FSAID relies on observed historical trends in irrigation efficiency in agriculture to predict future improvements. The Farm and Ranch Irrigation Surveys (FRIS) of the USDA were utilized as the data source for statewide irrigated area and total volume of water applied. The advantage of this approach is that it avoids the uncertainty of estimating at the field level exactly what type of management change would be made and how many farms or fields would be expected to make that change. It also recognizes the equipment and management changes made by producers which happen outside of any particular conservation initiative. The time series of the FRIS derived ratio of irrigation water applied to irrigated area was used to develop two exponential functions of efficiency improvements, where water per area is used as a proxy for efficiency. The function derived from the more recent FRIS data ( ) is assumed to be representative of efficiency improvements resulting mostly from management/scheduling changes. The exponential function derived from the entire time series ( ) is used to represent the combined impacts of irrigation system equipment changes and management improvements. Figure E 7 illustrates the shift in types of irrigation equipment. It can be seen that there is a sizeable shift from gravity systems to drip and micro spray systems. There appears to be relative stability in the equipment trends beginning with the 2003 FRIS. E 16

19 Figure E 7. Shift in Types of Irrigation Equipment Both FRIS derived efficiency improvement functions are used in the FSAID estimates (Figure E 8). The approach of using two functions allows for less efficiency improvements to be projected in the coming decades for newly irrigated fields or for existing irrigated fields that have drip or micro spray systems. It is assumed that newly irrigated fields or fields with drip or micro spray systems would achieve efficiency improvements largely resulting from management improvements. Figure E 8. FRIS Data of Irrigation Amount (MGD) Per Irrigated Acre E 17

20 Other fields (that are not new, or have not yet upgraded their equipment) could be expected to achieve efficiency improvements of a larger amount resulting from combinations of equipment and management improvements; this is represented by the exponential function built from the entire FRIS time series of water applied per irrigated area. A diagram of these decision rules is illustrated by Figure E 9. Figure E 9. Methodology for Applying FRIS Derived Efficiency Improvement Functions Limits on irrigation efficiency are defined using the metered/reported irrigation data. The median irrigation intensity (in/year) by crop type was used in order to not allow additional efficiency improvements at the field level if the irrigation intensity dropped below the median value for the crop type based on historical water use data (Table E 10). This means that for an individual citrus field, for example, if irrigation decreases below 9.6 in/yr as a result of the conservation trend, then no further irrigation conservation would be estimated for that field. Table E 10. Median Metered Irrigation Intensity Reported by Crop, Primary Crop INYR Citrus 9.6 Field Crops 7.1 Fruit (Non citrus) 23.7 Greenhouse/Nursery 21.8 Hay 5.9 Potatoes 11.9 Sod 6.7 Sugarcane 19.4 Vegetables (Fresh Market) 12.9 E 18

21 Soils represented in the FSAID model The FSAID model is developed using a large sample of metered/reported irrigation water use data. The historic water use data reflect the impacts of the physical (soils, weather, irrigation system, crop, etc.) and managerial (irrigation timing/scheduling, multi cropping) attributes that influence irrigation water use. The form of the FSAID model was chosen to allow the model to be fitted to water use data, per Florida Statute, and to allow for physical and economic variables to influence future water use. The variables included in the FSAID model include irrigation system type, location (latitude, longitude, and Water Management District), and crop type, among others. Early versions of the FSAID model included a variable representing soil type; however, the combination of soil, irrigation system, location, and crop resulted in a non functional result the model could not be fitted to the data due to collinearity among soils / crop / irrigation / location and water use data. In other words, the effect of soil types is captured in the model through the combination of crop / irrigation / location variables. Because soils are a static variable, and the intent is to identify variation among attributes that influence irrigation quantities, a static description of soils does not add critical information to the estimate of future irrigation demand. Soil is included explicitly in the decision making process for spatial distribution of future water demand soils data is used to identify fields that are likely candidates for conversion from unirrigated to irrigated land, and is used to identify likely cropping, based on prevalent combinations of crop/soil/irrigation equipment within a County or District. The application of the FSAID model is to estimate future water demands. Variation in FSAID model input variables is designed to reflect future conditions. E 19