Evapotranspiration estimation for oasis transect in middle reach of Heihe river basin based on remote sensing

Abstract: The Heihe River Basin is the second largest inland river basin in the arid area of Northwestern China, and the oasis in its middle reach has 95% of the total arable land and uses 68% of the total water resources in whole basin. Because evapotranspiration (ET) from the oasis is one of the most important components in water consumption, it is very important to study the ET in various land uses for water resource management and allocation in whole basin. The objective of this study was to estimate ET values for different land uses at a spatial scale of 16 km × 20 km in the middle reach oasis of Heihe river basin using remote sensing method. The main challenge of this study was to find a method to extrapolate instant ET values, which were estimated from the remote images, to monthly or seasonal actual ET (AET) values. In this study, a representative transect was selected to analyze the spatial distribution and temporal change of ET for different land cover types in the middle reach of the Heihe river basin. Seven qualified Landsat TM images for the studied transect were obtained during June to August in 2011. The surface energy balance model of SEBAL-METRIC and meteorological data were used to calculate the transient ET and daily reference ET (ET0) when the Landsat passed the studied region. The ratio of instant ET/ET0 was extrapolated to daily value using a sinusoidal function, while the daily ratio of ET/ET0 was assumed to be constant for a short period of 5 to 24 days for estimating a cumulative ET for different land uses in the transect. The comparison of daily ET values obtained by an eddy covariance system and remote sensing method showed the SEBAL-METRIC model combined with Landsat TM image and ET0 can provide reliable estimation of ET for different land cover types with an average relative error of 10.2%. The ET values varied from different land cover types and different irrigation schedules, with the high ET values in water area, farmland and forest, in contrast to low ET values in grassland and desert. The average ET values for farmland and forest during the studied period were 340 mm and 328 mm, respectively. They were higher than that for grassland of 214 mm and desert of 97 mm. ET variation for farmland was closely related to crop growth, and the highest ET value appeared on June 27th when both maize and wheat needed a large amount of water. By contrast, the ET values for desert were greatly influenced by the occurrence of precipitation events, as the vegetation was sparse and the most important limitation factor for evapotranspiration in this area was the low soil water content. Besides, the ET values of forest did not change significantly (p>0.05) during this period, with approximately 4 mm/d in sunny days. This study showed that METRIC hold substantial promise to estimate ET values for different land cover types in the middle reach of the Heihe river basin throughout plant growing periods. Besides, the accuracy of ET estimation in the oasis of Heihe middle reach could be improved when the hot extreme pixels in arable land and desert were respectively chosen due to different soil types.