Comparison of ET from two remote sensing-based surface energy models in south central Nebraska

摘要

Quantification of crop water use through evapotranspiration (ET) is very important for water resources management and irrigation scheduling. Two energy balance approaches were applied to estimate ET in south central Nebraska: (1) Mapping EvapoTranspiration at high Resolution with Internal Calibration (METRIC), and (2) Simplified Surface Energy Balance (SSEB) model. While both use remotely sensed thermal data for estimating ET, the METRIC is much more complex than the SSEB and widely used in the USA. Both ET models use the hot and cold pixel approach for scaling ET between driest and wettest pixels within an agricultural setting. Four Landsat 5 and four Landsat 7 images acquired for the 2007 growing season over path 29/row 32 were used to produce instantaneous ET. Images were selected with the least cloud cover and distributed temporally to cover different crop growth stages. ET data measured with a Bowen ratio energy balance system was used to assess the estimated ET accuracy. Statistical comparison of estimated daily ET against observed data gave an r~2 of 0.73 and 0.66 for METRIC and SSEB, respectively. Both models produced the same root mean square error of 1.5 mm/d or 35% of the observed average daily ET. However, statistical comparison between METRIC and SSEB-based ET gave an r2 of 0.96, indicating that SSEB performance is comparable to METRIC. Considering the minimal amount of ancillary data required for implementation and excellent performance in predicting daily ET, the SSEB approach is a promising tool for mapping ET.