Retrieval of Land Surface Temperature from Landsat 8 OLI and TIRS: A Comparative Analysis Between Radiative Transfer Equation-Based Method and Split-Window Algorithm
Sk Ajim Ali,Farhana Parvin,Ateeque Ahmad
DOI: https://doi.org/10.1007/s41976-022-00079-0
2022-12-01
Remote Sensing in Earth Systems Sciences
Abstract:The system of observation and capturing the earth resource features have been improving with the scientific revolution and technological development in remote sensing techniques. In comparison with the previous Landsat series, Landsat 8 OLI and TIRS (Operational Land Imager and Thermal Infrared Sensor) is the latest applications of thermal infrared sensor for the Landsat project offers two adjacent thermal bands that has a great advantage for retrieving land surface temperature. In this study, an effort was made to compare two different approaches of land surface temperature retrieval method from TIRS data including the radiative transfer equation (RTE) and the split-window algorithm (SWA). The objective of this study was to estimate land surface temperature from TIRS data of Landsat 8 using different techniques and compare with actual ground temperature for pre-monsoon, monsoon, and post-monsoon season to determine accurate technique and thermal band. In this regard, twelve ground stations such as New Delhi, Noida, Ghaziabad, Bulandshahr, Gurugram, Faridabad, Muradnagar, Safdarjung airport, Indira Gandhi international airport, Rajiv Chowk, Dadri, and Kirti Nagar were marked on Landsat 8 product with Path 146 and Row 40. Based on analysis, the result shows that the radiative transfer equation (RTE) using band 10 has highest accuracy with the lowest root mean square error (1.0334 ℃, 1.5189 ℃, and 1.4197 ℃, respectively for pre-monsoon, monsoon, and post-monsoon), while RTE using band 11 and split-window algorithm (SWA) using band 10 and 11 has lower accuracy with higher root mean square error (> 2.0 ℃ in all cases). Thus, it is recommended that for those methods LST retrieval using single band, band 10 using RTE has higher accuracy than band 11 and split-window algorithm.