Pixelizing and Validating Land Surface Temperature from Band 10 of Landsat 8

Abstract

The availability of long-term satellite imageries of Landsat since the early 1970s has contributed to a dimensional perspective between the in-situ observatory and ex-situ remote sensors regarding assessing the differences in Land Surface Temperature (LST). This study, therefore, has conducted digital image processing based on Algorithm for Automated Mapping in retrieving the LST and comparing it with the Air Temperature from the weather observatory in Savar, Dhaka, on three distinct occasions. The LST has been obtained from Landsat 8 thermal properties (assigned as band 10 TIRS) and the weather observatory data from Jahangirnagar University campus. The study reveals that there is negligible variation in the timeframe and that it is relatively almost the same fluctuations in the air temperature from the weather observatory. In relation to the Normalized Difference Vegetation Index (NDVI) and Emissivity values, the variability of retrieved LSTs has been evaluated. By processing the time series data, the margin difference evaluated from pixel-based value deviation and point data from the exact location of the weather observatory that was used to determine the relevance of obtaining Land Surface Temperature which can be changed by taking into account the atmospheric influence and weather conditions of seasonal fluctuations. The goal of the current study has been derived from the Raster Calculation method to retrieve data to validate with air temperature. The method is effective by providing a difference of ±0.5˚C on the study, according to the difference between the LST and air temperature, which indicates a prominent phase of retrieving temperature data for the vast accumulation of climate-based research.