SPATIAL ESTIMATION OF REGIONAL CROP EVAPOTRANSPIRATION M.A. Hashmi, L.A.Garcia, D. G. Fontane ABSTRACT. Current methods of computing regional crop evapotranspiration (ET), the prime variable in estimating irrigation demand, largely ignore the spatial variability of ET parameters, thus introducing errors. In this research, a method was developed to estimate regional ET while considering the spatial variability of parameters. To consider this variability, spatial databases were developed for agricultural land-use, relevant climatic parameters, and topographic data using geographic information systems (GIS). Analytical GIS Junctions of map algebra and map overlay were used to calculate regional crop ETfor the Cache la Poudre Basin in Colorado. The Cache la Poudre Basin was selected as the study area due to the availability of ground truthed land-use survey data. This research uses land-use classification, climatic variables and elevation adjustments to determine the value of using a spatial/GIS approach instead of current nonspatial approaches. Keywords. Geographic information systems (GIS), Evapotranspiration, Water demand. Regional crop evapotranspiration. Agricultural land-use classification. Spatial analysis. A ccurate estimation of water demand in an irrigation network system is critical to a system's agricultural as well as nonagricultural objectives. A recent study, Mizyed (1990), has quantified the improvements in the performance of an irrigation system in response to improved estimation of water demand. For the Mahaweli System in Sri Lanka, water demand errors with a standard deviation of 40% caused a 12% increase in the mean annual energy shortages. Current nonspatial methods for calculating evapotranspiration (ET) could potentially produce water demand errors with a standard deviation in the same order of magnitude as the Mahaweli System (40%). Water demand is usually estimated using a simple water balance equation that determines changes in soil water by measuring water inputs and withdrawals. Effective rainfall, applied irrigation, and upward groundwater movement are the principal water inputs, whereas ET and drainage are the prime water withdrawal components. The climatic data used to measure ET are highly variable spatially. For instance, temperature and wind can vary greatly in distances of only a few kilometers. In addition to climatic data, land use can vary from field to field, thus affecting crop ET rates. Most of the current water demand models are non-spatial models. Nonspatial models are based on the concept that a measured value represents a homogeneous Article was submitted for publication in January 1994; reviewed and approved for publication by the Soil and Water Div. of AS AE in January 1995. The authors are Mansoor A Hashmi, Design Engineer, National Engineering Services Pakistan (Pvt.) Ltd., New Garden Town, Lahore, Pakistan; Luis A. Garcia, ASAE Member Engineer, Director, Integrated Decisions Support Group and Assistant Professor, Agricultural and Chemical Engineering Dept., and Darrell G. Fontane, Associate Professor, Civil Engineering, Colorado State University, Fort Collins. Corresponding author: Luis A. Garcia, Agrciultural and Chemical Engineering, Colorado State University, Fort Collins, CO 80523; telephone: 303-491-5144. area around the measurement point, which is generally not true in areas with large climatic or topographic variations. The approach of this research was to model the spatial variability of some of the parameters used to calculate regional ET and estimate the expected benefit of using a spatial approach. Digital elevation maps were entered directly into the system as spatial data. Point data, such as climate data were modeled spatially by using interpolation techniques. Agricultural land-use classification was done using remotely sensed data along with extensive ground truthing to produce a digital-spatial representation. This approach was facilitated by geographic information systems (GIS), a technology in the field of spatial data management and analysis. In this research, the prime functions of GIS have been to perform complex area overlays and modelling analysis, develop and prepare spatial input data, provide a conversion and standardization for digital land forms, and allow for post-processing of output in a graphical format. The objective of this article is to develop a spatial/GIS method for estimation of regional ET and to characterize factors that affect the accuracy thus achieved using the spatial/GIS approach. SPATIAL/GIS APPROACH FOR REGIONAL ET ESTIMATION Depending on the location, local ET has been shown to vary greatly over short distances. For example, Nixon et al. (1963) measured ET in a California coastal valley located 23 miles inland, and found that ET was 1.5 times greater than a value measured 10 miles inland from the Pacific ocean. This research shows that local calculations of ET can be expected to vary from location to location depending on topography and climate. In other words, ET can vary spatially. However, Trimmer (1980) concluded that within a 67 mile radius of a weather station located in a relatively flat area in the Nebraska high plains, the ET data were representative of crop water use. Therefore, regional ET values can be expected to vary, except when VoL.38(5):1345-1351 Transactions of the ASAE 1994 American Society of Agricultural Engineers 0001-2351 / 95 / 3805-1345 1345