Analyzing water table depth fluctuation response to evapotranspiration involving DRAINMOD model

Abstract

A proper prediction of the water table (WT) depth in hydrological models highly depends on the accuracy of evapotranspiration (ET) estimation. In this study, the response of WT fluctuations to changes in ET was investigated involving the DRAINMOD water management model. The WT depth data were measured during three canola growing seasons (2011-12, 2015-16 and 2016-17) in a paddy field with two subsurface drainage systems (drain depth of 65 cm with two spacing of 15 m and 30 m) in the north of Iran. For the growing seasons, ET was estimated through 17 equations including five temperature- based, four radiation- based, three pan evaporation- based, four combination- based and one mass transfer- based equations. The estimated ET values was then applied in the calibrated DRAINMOD model to simulate the corresponding WT depths in the drainage systems. The reliability of various ET estimation methods and the model predictions were evaluated using determination coefficient (R2), root mean square error (RMSE), error percentage (PE) and mean absolute deviation (MAD). Compared to the FAO- Penman-Monteith equation, the FAO-24 radiation (R2 = 0.90, RMSE = 0.31 mm d-1, MAD =0.71 mm d-1 and PE = - 0.95%), Blaney- Criddle (R2 = 0.93, RMSE = 0.38 mm d-1, MAD =0.76 mm d-1 and PE = - 6.24%), Irmak (R2 = 0.87, RMSE =0.87 mm d-1, MAD =0.41 mm d-1 and PE = - 11.09%) and FAO- 24 pan (R2 = 0.86, RMSE =0.45 mm d-1, MAD =0.80 mm d-1 and PE = - 16.18%) provided the best estimations of ET. However, the best prediction of the WT depths were obtained for the Rohwer and pan evaporation- based equations. These methods considerably improved the reliability of the DRAINMOD predictions in comparison with the Thornthwaite method which is a default method for ET calculation in the model. Based on the results, the Rohwer and FAO-24 pan equations are recommended as suitable methods for estimating ET in the DRAINMOD model for the study area.