Estimation of daily evapotranspiration and irrigation water efficiency at a Landsat-like scale for an arid irrigation area using multi-source remote sensing data

Abstract

The estimation of land-surface evapotranspiration (ET) at high spatial and temporal resolutions is important for management and planning of agricultural water resources, but available remote sensing data generally have either high spatial resolution or high temporal resolution. To overcome this limitation, we evaluated the use of a data fusion scheme, Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM), to determine the surface parameters needed to estimate daily ET at a Landsat-like scale (100 m). In particular, we fused Moderate Resolution Imaging Spectroradiometer (MODIS) data with Landsat Enhanced Thematic Mapper Plus (ETM+) data in analysis of the Heihe River Basin (HRB), an arid region of Northwest China. The surface parameters were then used to drive the revised Surface Energy Balance System (SEBS) model to estimate daily ET at a spatial resolution of 100 m for this an arid irrigation area during the crop growth period (April to October) in 2012. The results showed that the daily ET estimates had a mean absolute percent error (MAPE) of 12% and a root mean square error (RMSE) of 0.81 mm/day relative to ground measurements from 18 eddy covariance (EC) sites in the study area. The validation results indicated good accuracy for land cover types of maize and vegetables, a slight overestimation for residential and wetland sites, and a slight underestimation for orchard site. Our comparison of the input parameter fusion approach (IPFA) and the ET fusion approach (ETFA) with field measurements indicated the IPFA was superior than the ETFA for land surfaces with high spatial heterogeneity. Furthermore, our high spatiotemporal ET estimates indicated that irrigation water efficiencies of the irrigation districts (mean: 70%) and villages (mean: 62%) had large spatial heterogeneity. These results point to the need for calculating ET at a high spatiotemporal resolution for monitoring and improving irrigation water efficiency at local scales. Our findings suggest that the proposed framework of estimating daily ET at a Landsat-like scale using multi-source data may also be applicable to other heterogeneous landscapes by providing a foundation for management of water resources at the basin or finer scales.