In this research, the LST-Ta relationship was tested and evaluated to estimate Ta for 15 years (2004-2018) for summer and winter months through various predictive models (linear, quadratic, cubic and exponential) in Peshawar basin. Peshawar Basin in the foothills of Himalayas (northwest Pakistan) with semi-arid to arid type of climate having very limited and uneven distribution of weathers stations. Air temperature is usually measured at ground-based meteorological stations covering close vicinity of the measurement site, hence exhibits limited applicability for vast areas with heterogeneous surfaces. Remotely sensed TIR (thermal infra-red) imagery are top choice for geoscientists to estimate air temperature data at high spatio-temporal sclaes. However, such approaches need to be thoroughly evaluated globally. Results from the current study revealed that linear and quadratic models performed well to estimate air temperature during winter and summer with more plausible results. The Pearson correlation coefficient (r) with values 0.78 & 0.62 (summer), and 0.59 & 0.78 (winter) were observed at two weather stations. This study will provide technical details for developing remotely sensed techniques for estimating spatio-temporal air temperature data using same or other sensors. The spatio-temporal outcome could be easily incorporated in various models in the domain of climate research, global warming and agriculture. The model adopted in this study should be further investigated across different ecosystems having different meteorological and climatic conditions. (c) 2022 COSPAR. Published by Elsevier B.V. All rights reserved.