Water Journal : Water Journal November 2015
WATER NOVEMBER 2015 36 Feature Article As water utilities and others consider supply constraints and planning horizons, understanding where water will be most needed is of utmost importance. Of particular importance is water demand for agricultural irrigation purposes, which consumed 10,730,000 ML in 2014 in Australia, according to the Australian Bureau of Statistics. While several models exist in Australia to predict water use for individual farms, predicting future agricultural water use in Australia at the national level is more challenging. Water availability is a sensitive issue across the globe and the United States (US) is no different. Recent work there has shown how large-scale models can be developed to predict future irrigation demands based on the combination of physical and economic factors that impact water use in agriculture. In Florida, despite dense urban development, 40 per cent of the total water use, or about 3,800,000 ML per year, results from agricultural irrigation (Marella, 2014). In recognition of this large consumptive footprint amid heightened water scarcity, The Balmoral Group recently developed a model for the entire state of Florida, showing the prediction potential for a large geographical context. Florida legislature recently set planning rules for agricultural water supply that require the Florida Department of Agriculture and Consumer Services (FDACS) to develop geographically speci c water use estimates that incorporate metered and other water use data into a 20-year horizon, across the entire state. In the past, Florida's ve water management districts (WMDs1) used their own individual models to estimate agricultural water use separately for each region. As a result, statewide estimation was fraught with inconsistency. For example, a crop in one district might be said to require 70 per cent more water than the same crop one kilometre away, but in another district due to differences between estimation approaches. To overcome this type of inconsistency, The Balmoral Group was tasked with estimating agricultural water use for all the farms across the state; this is the Florida Statewide Agricultural Irrigation Demand (FSAID) project. The Balmoral Group's model was completed as four related steps. First, a dataset of all agricultural lands was prepared in a geographic information system (GIS) and irrigated lands were identi ed. Water use estimates were developed using econometric techniques to link current and historical biophysical factors, irrigation water use data, and crop prices at the farm level. Revenue projections were then used to simulate future conditions. Finally, auto-regressive techniques were used to identify areas of growing and declining irrigated acreage, and generate projected irrigated area and water use. DEVELOPMENT OF IRRIGATED LANDS AND AGRICULTURAL LANDS GEODATABASES The bene t of a geodatabase with individual elds is that it can be utilised at any scale, facilitating re nements to re ect temporal changes and controlling for crop differences. Recognising this, GIS databases of all agricultural lands in Florida, as well as irrigated lands, were created for the FSAID project (The Balmoral Group, 2015). An Irrigated Lands Geodatabase (ILG) was populated for 2015 conditions, using data from a number of sources. Agricultural areas were identi ed as irrigated or non-irrigated through manual and automated processes. The GIS was used to visually review and compare aerial photography (NAIP2 and Google Earth), USDA's Cropland Data Layer (CDL)3, and permit data (reading permit documents to extract crop type, irrigation system and irrigated area), to correctly classify eld geometry, crop type and irrigation system. The resulting GIS mapped an estimate of 7,000 square kilometres of irrigated area in 2015, and about 36,000 square kilometres of total agricultural land as shown in Figure 1. IRRIGATION ESTIMATES Once the current irrigated and non-irrigated agricultural areas were established, the next step was to estimate current water use. About one-third of farms in Florida have metered water use, and this data is reported to the WMDs. Data was obtained for each District for three full years of irrigation, allowing for comparison of irrigation practices under a variety of climatic and market conditions. Demand for irrigation water was then modelled in an econometric regression linking total annual water use to biophysical factors, climate and market factors (see Table 1). For each farm, this included compiling data for each of the three years -- if metered water use data was available for 2007, 2010 and 2013. The model included variables representing the crop mix, eld size, irrigation equipment used, and weather conditions observed for each year having measured ESTIMATING AGRICULTURAL IRRIGATION DEMAND USING ECONOMICS AND ENGINEERING MODELS The Balmoral Group was recently tasked with estimating agricultural water use across every farm in the state of Florida in the US. The project, titled Florida Statewide Agricultural Irrigation Demand (FSAID), was completed in a four-step process outlined here by Valerie Seidel and Paul Yacobellis. 1 Northwest Florida (NWFWMD), St. John's River (SJRWMD), Suwannee River (SRWMD), Southwest Florida (SWFWMD) and South Florida (SFWMD). 2 United States Department of Agriculture National Agricultural Imagery Program. 3 The USDA's Crop Data Layer (CDL) is a gridded dataset (30-metre resolution) that classi es crop type based on satellite data and ground truth data, updated annually based on satellite data during the peak growing season (April to September).
Water Journal September 2015
Current Feb 2016