Estimation of Spatio-Temporal Near Surface Air Temperature from Land Surface Temperature (LST) Generated by Landsat 8 Satellite Data

Abstract Predicting air temperature with high accuracy is useful for many agricultural, environmental, hydrological, and ecological applications, especially for global climate change studies. Air temperature is traditionally obtained from measurements at mobile or stationary meteorological stations and their distribution is determined by applying various interpolation methods. However, using of these data is limited and insufficient in regions such as Türkiye where the spatial distribution of stations is limited, containing many micro-climatic regions, and where the air temperature is very important, especially for agricultural applications. In this respect, being able to predict the near-surface air temperature with high accuracy with satellite-based observations constitutes the motivation of this study. Many studies tried to estimate air temperature using surface temperature data. However, the obtained accuracies were limited as medium or low-resolution satellite data were used and varied according to the regions. The aim of this study is to predict the land surface temperature and air temperature with high accuracy to be produced from high spatial resolution Landsat 8 images. For this purpose, two data sets were used, namely the LST data obtained from Landsat 8 and the air temperature data measured with the meteorological station. Linear regression models were created by comparing these two data sets and Pearson correlation was applied to determine the correlation between them. The results showed that there was a significant relationship between the produced LST and the predicted air temperature for all stations, this relationship was estimated in the 95% standard deviation range, and the difference between the two values was mostly below 1°C. This result showed that the applied method is suitable for estimating the air temperature near the surface. It has been observed that this method can be used with high accuracy, especially in regions with insufficient number of meteorological stations, variable air temperatures in short distances, and different altitudes.