干旱区生态环境知识资源中心

Most of the existing satellite-based evapotranspiration algorithms have been developed using fine-resolution Landsat TM and ASTER data. However, these algorithms are often applied to coarse-resolution MODIS data. In this paper, we investigate the feasibility of this approach using data from ASTER overpasses of the Las Cruces region in New Mexico, which consisted of irrigated and bare land surface conditions in a desert climate, and a satellite-based evapotranspiration algorithm which has been developed for the Las Cruces region using fine-resolution ASTER data. The evaluation technique consists of comparing evapotranspiration calculated in two ways: first, calculate at the ASTER resolution and average to the MODIS resolution; and second, calculate directly at the MODIS resolution by aggregating the ASTER data to the MODIS resolution. Results show that applying the satellite-based algorithms, which are developed at ASTER resolution, to MODIS resolution leads to ET estimates that (1) preserve the overall spatial pattern (spatial correlation in excess of 0.90), (2) increase the spatial standard deviation and maximum value, and (3) have modest conditional bias: underestimate low ET rates (<1 mm/day) and overestimate high ET rates; the overestimation is within 20%. The results emphasize the need for exploring alternatives for estimation of ET from MODIS. (C) 2009 Elsevier B.V. All rights reserved.