Uncertainties in Estimation of Basin-Scale Actual Evapotranspiration Using SEBAL
2021
Owlia, Amir Hossein | Sima, Somayeh
Estimation of actual evapotranspiration (Eta) at the basin-scale using SEBAL, as one of the commonly used satellite-based models, are affected by uncertainties associated with the selection of the anchor pixels, satellite sensors, and the spatial extent of the input satellite images. This study was conducted to investigate the impacts of aforementioned uncertainties on the actual evapotranspiration estimates in Urmia Plain (in the northwest of Iran) using MODIS and Landsat8 satellite imaginaries and the PySEBAL and MPySEBAL (Modified version) models. Validation results using lysimetric data during 2010-2011, showed that MPySEBAL (with cold pixels on well-irrigated vegetation) has less RMSE up to 70% as compared to PySEBAL model. Moreover, in the heterogeneous areas such as Urmia Plain, MODIS data with less spatial resolution leads to a 33 percent overestimation of daily Eta compared to Landsat 8 results. Furthermore, introducing a satellite image at its original extent rather than cropping the study area will result an uncertainty in the daily Eta estimates up to 8%. Comparing the relative impacts of the three sources of uncertainties indicated that the selection of the anchor pixels based on the surface temperature and NDVI thresholds, the spatial resolution of the sensors, and the spatial extent of the input images introduce the largest uncertainties respectively. Findings of this study can be used to enhance the accuracy of satellite-based Eta models and estimation of the irrigation water consumption from filed to basin scales.
显示更多 [+] 显示较少 [-]