J. Sci. Agri. Technol. (2022) Vol. 3 (1): 1-7
DOI: https://doi.org/10.14456/jsat.2022.1
Abstract
Land Surface Temperature (LST) has long been monitored and studied; however, the most reliable method of estimating the LST has yet to be examined regarding the mixed land-use types over a small city. This research explores an optimum method for Land Surface Temperature (LST) estimation in a city using the data from LANDSAT-8. Four favored LST retrieval approaches, the Radiative Transfer Equation-based method (RTE), the Improved Mono-Window method (IMW), the Generalized Single-Channel method (GSC), and the Split-Window algorithm (SW), were used to estimate the LST over Buriram Town Municipality, Thailand. The calculated LST from these four methods was compared with ground-based temperature data of 100 measured sites over the study area on the same date and time as the employed Landsat-8 data. The lowest Normalized Root Means Square Error (NRMSE) was considered to identify the optimum method of the LST estimation. The SW algorithm provides the lowest NRMSE value (0.114), followed by the RTE (0.171), the IMW algorithm (0.181), and the GSC (0.219). As a result, the SW algorithm is the optimum method in LST estimation for Buriram Town Municipality. The SW algorithm mainly eliminates atmospheric effects based on differential absorption in two thermal bands, which have shown the smallest error in the retrieval of LST. The explored optimal method will benefit GIS specialists working for Buriram local government to conduct the best practice to monitor the LST over the city. The other local governments could consider the SW algorithm to monitor the LST over their small cities with similar contents.
Keywords: split-window, Landsat-8, land surface temperature, thermal bands, Buriram
Received: October 13, 2021. Accepted: February 14, 2022.
