Development and Validation of a Daily Maximum Temperature Estimation Algorithm Using Landsat-8

Daily Maximum Temperature (DMT) is important variable for a wide range of field such as agriculture, climate, and disaster management. There is Automated Weather Station (AWS) operated by Korea Meteorological Administration (KMA). AWS gathers the data such as temperature and rainfall. However, AWS is not installed with consideration of land cover. Therefore, there is a lack of temperature information for areas where AWS is not installed. In this study, we made an algorithm to generate daily maximum temperature image using stepwise regression to fill the lack. We set the AWS DMT as dependent variable, and Landsat-8 Land Surface Temperature (LST), Digital Elevation Model (DEM), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), and Normalized Difference Built-in Index (NDBI) as independent variables. We used selected variables, which are LST, NDWI, and NDBI, and made an algorithm. Produced images were compared and validated with ground observation data. Correlation between AWS and Landsat-8 DMT is 0.76, and averaged RMSE is about 2.36eC). It is considered that the accuracy of our algorithm is verified because correlation and RMSE are within the error range of the previous studies.