Using Panel Data Technique to Estimation Reference Evapotranspiration

Evapotranspiration is the most important part of the hydrological cycle, which plays a key role in water resource management, crop yield simulation, and irrigation scheduling. The purpose of this research was to estimate the reference evapotranspiration using ‘panel-data’ models. Panel-data multivariate analysis endows regression analysis with both spatial and temporal dimensions. This study was carried out using weather data of 9 synoptic stations of Khorasan Razavi during 1971-2000. Data were divided randomly into two sub-sets, 75% for model development and 25% for model evaluation. The panel-data models were developed using the monthly mean air temperature and monthly mean wind speed as inputs in order to estimate monthly reference evapotranspiration. The results indicated that the two-way fixed effects models were superior. The statistical index (RMSE = 9.85, MAE = 7.38 and R2 = 0.99) revealed the effectiveness of this model. In addition, these results were compared with the results of ordinary least squares regression and Hargreaves-Samani equation which showed the superiority of the panel-data models.