Land surface temperature retrieval and oversaturation correction using Landsat 8, MODIS Terra, and Sentinel 3

The importance of land surface temperature (LST) for local to global scales ecosystem poses as a challenge for developing nations like Indonesia, whereas the insufficient distribution of ground observation to measure LST. Another issue of LST retrieval is the oversaturated pixel of the optical sensor which later causes the abnormal value of NDVI for the LST retrieval. Therefore, the purpose of this study is to retrieve land surface temperature which later will be validated using available weather stations and to correct the oversaturated pixel during the retrieval process. The method used to perform oversaturated pixel of Landsat 8 product is done by using the abnormal NDVI value (below -1 and above 1) as the mask, which later used to remove the oversaturated digital number (DN) value (>65535). Moreover, the method used to retrieve the LST is the conversion of oversaturation corrected DN to radiance combined with surface emissivity for Landsat 8, and LST end-product (level 2) of MODIS Terra and Sentinel 3. The result of the oversaturation correction using the abnormal NDVI value was able to remove the oversaturated DN, thus produced NDVI with quite an acceptable range value which later counts as the crucial part to determine the surface emissivity for LST retrieval. In addition to LST retrieval, LST from Landsat 8 (both in 30m and 1km resampled spatial resolution) has the highest agreement with the ground measurement using air temperature data from the four available weather stations. In conclusion, the oversaturation correction of the Landsat 8 product is crucial for calculating surface emissivity which later used for more reliable LST retrieval. The occurrence of the oversaturated pixel in the optical sensor still presents even the product is passed the Landsat Quality Assessment Tools (cloud-free, no radiometric saturation). Moreover, the available weather station used to measure the air temperature cannot fully validate the LST retrieval from satellite observation, which points out the importance of ground-based LST measurement especially in developing nations for both economic and ecological management.