Using Forecasts of WRF Regional Model to Improve the Accuracy of Reference Evapotranspiration Estimation

An accurate estimation of reference evapotranspiration is crucial for optimal irrigation scheduling and management. Also, achieving accurate medium range forecasts of effective parameters in estimating reference evapotranspiration is a key component for dynamic irrigation scheduling. This study was aimed to investigate the effect of using Weather Research and Forecasting Model (WRF) regional forecasts to increase the accuracy of reference evapotranspiration estimation. Consequently, the precision and accuracy of the model, and the outcome of forecasts performance at 24, 48, 72, 96, and 120-hours were evaluated to estimate the reference evapotranspiration. For this purpose, the output of the model for four stations including Qazvin, Esmaeil-Abad, Karaj and Hashtgerd were extracted for a period of three months (May-July, 2018) with a 10-days average of the base period. The weather data of 2018 at these stations with the corresponding ones were compared afterwards. The results indicated that the 10-days reference evapotranspiration (average of all stations) in the study period, according to the base period were -20.9, -8.12 and 7.83 percent, respectively. These variations reflects the deviation of the reference evapotranspiration value in the study period in comparison with the base period, indicating the need for using medium-range forecasting in order to correct reference evapotranspiration estimates. The range of determination coefficient (R2) of model output was obtained to be between 0.813 and 0.921. Due to the statistics, the model output for all stations and the lead-time forecasting periods of 24, 48, 72, 96 and 120 hours can be evaluated with high accuracy and its application would enhance the accuracy of reference evapotranspiration estimates. According to the results, not only in terms of time coordination, but also in terms of quantity, there was a high similarity between the estimated values of evapotranspiration derived from its post-statistical output of the (WRF) with calculated values.