Spatial Downscaling of Land Surface Temperature (LST) of MODIS in Irrigated Areas by TOTRAM and OPTRAM Soil Moisture Estimation Models

The limitations of satellite sensors make it impossible to access thermal bands with high spatial and temporal resolution simultaneously. Therefore, downscaling methods are essential because they provide simultaneous access to thermal data with high spatio-temporal resolution. LST parameter is critical in agriculture, it is one of the most important in estimating the amount of evapotranspiration and significantly impacts crop growth. LST images of the MODIS with spatial resolution of 1000 meters are available daily. Still, the low spatial resolution in these images is a limitation that makes it impossible to use for agricultural management. On the other hand, high humidity changes in irrigated regions cause errors in LST downscaling process. this research was carried out with the aim of downscaling the LST of the MODIS sensor from 1000 meters to OLI resolution from Landsat 8 satellite (30 meters) in irrigated regions. In the first stage, the DisTRAD model was implemented for the downscaling of the LST of MODIS in Amirkabir and Mirzakoochak Khan Farms. The results show the poor performance of the DisTRAD model in the downscaling of LST from 1000 meters to 30 meters. Next, to check the results of LST downscaling of the MODIS by TOTRAM and OPTRAM soil moisture estimation models, the root mean square error (RMSE) statistic was used. The results indicate that the average value of RMSE in the downscaled images by the OPTRAM-TOTRAM model shows a decrease of about 2.53°C compared to the DisTRAD model. Also, the average value of RMSE in the first six months of the year when irrigation has been done, compared to the DisTRAD model, shows a decrease of about 4.11 °C. As a result, the use of the OPTRAM-TOTRAM model offers much better performance than the DisTRAD model in downscaling the LST of the MODIS in irrigated regions.