Jalal Yarahmadi; Seyed Majid Mirlatifi; Ali Shamsoddini; Majid Delavar
Abstract
Evapotranspiration has a key role on spatial and temporal distribution of available water, as vital component of water balance. ET ground measurements at large scale has limitation, so, different methods have been developed to estimate actual ET based on remote sensing data. The purpose of this study ...
Read More
Evapotranspiration has a key role on spatial and temporal distribution of available water, as vital component of water balance. ET ground measurements at large scale has limitation, so, different methods have been developed to estimate actual ET based on remote sensing data. The purpose of this study was to evaluate accuracy of actual ET estimation for MOD16, MYD16 and SSEBOP global database models in monthly and seasonal time scales for different land use and wet, dry and normal climate conditions at Karkheh Dam Basin. First, SWAT model was calibrated and verified based on data of hydrometric stations included: river discharge, base flow and aquifer storage. After ensuring the accuracy of SWAT model performance in estimating water balance components at the studied basin, simulated actual evapotranspiration values were used to evaluate the temporal-spatial accuracy of actual evapotranspiration data of global database models. Results showed that all three models underestimate actual evapotranspiration values with a significant difference from the SWAT model results. The RMSE and MBE values varied from 15 to 21.74 and -15.93 to -8.19 mm on a monthly scale and from 40.17 to 59.32 and -47.74 to -19.36 mm on a seasonal scale, respectively. The concordance between the actual evapotranspiration results of the global database models and the simulated SWAT model values in the dry year is lower than wet and normal years. Although the results of the SSEBOP model had less error than SWAT model, the actual evapotranspiration time variations of the MOD16 and MYD16 models were more consistent with the time series data of the SWAT model. The results also showed that in the dominant agricultural basins, the SSEBOP model and in the forest and pasture basins, the MYD16 and MOD16 models have less error.
