Prediction accuracy for projectwide evapotranspiration using crop coefficients and reference evapotranspiration

The Imperial Irrigation District is a large irrigation project in the western United States having a unique hydrogeologic structure such that only small amounts of deep percolation leave the project directly as subsurface flows. This structure is conducive to relatively accurate application of a surface water balance to the district, enabling the determination of crop evapotranspiration (ETc) as a residual of inflows and outflows. The ability to calculate ETc from discharge measurements provides the opportunity to assess the accuracy and consistency of an independently applied crop coefficient-reference evapotranspiration (K-c ET0) procedure integrated over the project. The accuracy of the annual crop evapotranspiration via water balance estimates was +/-6% at the 95% confidence level. Calculations using K-c and ET0 were based on the FAO-56 dual crop coefficient approach and included separate calculation of evaporation from precipitation and irrigation events. Grass reference ET0 was computed using the CIMIS Penman equation and ETc was computed for over 30 crop types. On average, K-c-based ET computations exceeded ETc determined by water balance (referred to as ETc (WB)) by 8% on an annual basis over a 7 year period. The 8% overprediction was concluded to stein primarily from use of K-c that represents potential and ideal growing conditions, whereas crops in the study area were not always in full pristine condition due to various water and agronomic stresses. A 6% reduction to calculated K-c-based ET was applied to all crops, and a further 2% reduction was applied to lower value crops to bring the project-wide ET predicted by K-c-based ET into agreement with ETc WB. The standard error of estimate (SEE) for annual ETc for the entire project based on K-c following the reduction adjustment, was 3.4% of total annual ETc, which is considered to be quite good. The SEE for the average monthly ETc was 15% of average monthly ETc. A sensitivity analysis of the Computational procedure for K-c showed that relaxation from using the FAO-56 dual K-c method to the more simple mean (i.e.. single) K-c curve and relaxation of specificity of planting and harvest dates did not substantially increase the projectwide prediction error The use of the mean K-c curves, where effects of evaporation from wet soil are included as general averages, predicted 5% lower than the dual method for monthly estimates and 8% lower on an annual basis, so that no adjustment was required to match annual ET derived from water balance. About one half of the reduction in estimates when applying the single (or mean) K-c method rather than the dual K-c method was caused by the lack of accounting for evaporation from special irrigations during the off season (i.e., in between crops).