Comparison of Some Split-window Algorithms to Estimate Land Surface Temperature from AVHRR Data in Southeastern Tehran,

Abstract

Land surface temperature (LST) is a significant parameter for many applications. Many studies have proposed<br />various algorithms, such as the split-window method, for retrieving surface temperatures from two spectrally<br />adjacent thermal infrared bands of satellite data. Each algorithm is developed for a limited study area and<br />application. In this paper, as part of developing an optimal split-window method in the southeast of Tehran province,<br />Iran, four commonly applied algorithms to retrieve the LST from AVHRR were compared. This study was carried<br />out in a wheat farm site located in the Pakdasht Agricultural Region. Measurements of LST over the farm were made<br />with a manual infrared radiometer at the time of NOAA overpass for 18 days of May to June 2004. These days were<br />cloud free over the study area. A total of 18 NOAA images were acquired for the days that LST measurements were<br />made. The temperatures derived by the different split-window algorithms were compared to ground truth<br />measurements. The performance of the split window algorithms was checked with three statistical indices: root mean<br />square error (RMSE), mean bias error (MBE) and coefficient of determination (R2). The results showed that the<br />Ulivieri split-window algorithm produced the lowest value of RMSE and MBE (2.71 and 0.26 K, respectively) and<br />its highest value of R2 (0.92) gave more accurate results than the other algorithms.