Land surface temperature retrieval for climate analysis and association with climate data

The aim of this study is to demonstrate the relationship between the long years’ monthly average (LYMA) land surface temperature (LST) and the LYMA air temperature (T_a), the total precipitation (P_t), and the relative humidity (RH). Data from 27 meteorological stations in the Eastern Thrace region and corresponding thermal infrared images from Landsat-5 (TM) and Landsat-7 (ETM+) were used in this study. Simple regression models were developed for each meteorological station to predict the LYMA T_a, P_t and RH based on the LST values. The resulting LST-based prediction models were judged based on the correlation coefficient (r) and root mean square (RMSE). The average correlation and RMSE for the LST-based T_a were r = 0.959 and RMSE = 1.771 °C. The average correlation and RMSE for the LST-based Pt were r =-0.863 and RMSE = 10.098 mm. The average correlation and RMSE for the LST-based RH were r =-0.932 and RMSE = 1.875%. The results indicate that LST can be a good estimator for LYMA T_a, P_t and RH, and LYMA T_a is positively, LYMA Pt and LYMA RH are negatively correlated with LYMA LST.