Abstract:Field evapotranspiration is an important parameter for crop water requirements estimation and farmland water management, which affects the distribution of soil water and heat parameters during the wintertime and potentially affects crop growth in next spring. So far, Penman-Monteith (PM) model, Priestley-Taylor (PT) model and simultaneous heat and water (SHAW) model are widely used for surface evapotranspiration estimation. However, only the SHAW model is used for the estimation of surface evapotranspiration in wintertime, and the applicability of PM and PT models in the winter still needs to be verified. The accuracy and applicability of the above three models were analyzed and evaluated for estimating surface evapotranspiration in wheat fields overwinter. Parameters in models were modified by considering the latent heat during water-ice phase change. Meteorological parameters and actual evapotranspiration during the winter of 2011—2012 and 2012—2013 were collected in an experimental field, Changping County, Beijing, China. The estimated values using the default or empirical parameters of the three models were compared with the actual measured values, showing that the accuracy of PT model for evapotranspiration was the highest (RMSE was 0.159mm for the PT model, 0.697mm for the PM model and 0.390mm for the SHAW model), but PM and PT models overestimated the evapotranspiration, which were attributed to the solid-liquid and gas-liquid phase changes of soil surface. In order to improve the accuracy of the estimation, a water stress coefficient in the PT and PM models was introduced to modify the three model parameters by using the data of the first year of winter. The results showed that the estimated accuracies of evapotranspiration were significantly improved by the modified PM model (RMSE was 0.159mm during 2011—2012) and the SHAW model (RMSE was 0.280mm during 2011—2012). The modified models were used to estimate the surface evapotranspiration in the winter of 2012—2013. The results showed that the accuracy of the three models remained high and acceptable (RMSE was 0.267mm for PT model, 0.252mm for PM model and 0.253mm for SHAW model). In comparison, the PT model had a simple form and required less parameter input. Therefore, the PT model can be preferentially chosen when estimating surface evapotranspiration during wintertime.
