Stochastic soil moisture estimation and forecasting for irrigated fields

A methodology is developed for estimating and forecasting soil water depletion and crop evapotranspiration, with explicit consideration of modeling errors and stochastic inputs. The water balance of an irrigated field and a time series model for reference crop evapotranspiration are formulated in state‐space form, with soil moisture depletion and reference evapotranspiration as state variables. The Kalman filter is used to generate estimates and forecasts of the state variables, together with statistical information on their associated errors. Model calibration and validity tests are performed with two independent data sets from locations in Colorado. Each set includes several years of reference crop evapotranspiration data calculated from climatological observations, one season of soil moisture measurements, and concurrent irrigation applications. The estimates, forecasts, and error covariance information provided by the model can allow irrigation decisions to be made with explicit consideration of the inherent risks of crop damage or failure under limitations in water, energy, labor, and capital.