Geostatistical Modeling of Air Temperature Using Thermal Remote Sensing

摘要

Geographic Information Systems and spatial interpolation are the most often used geographic sciences for spatial analysis and visualization of temperature to use in hydrological studies. According to dependency of nature of thermal bands data to temperature, using thermal remote sensing images as auxiliary data can be useful in air temperature spatial interpolation. In light of these considerations, we used Landsat thermal bands together with Kriging and Co-kriging geostatistical methods for four seasons to interpolate mean temperature in Northeast of Iran as a region with low density of gauge distribution. Using Landsat (instead of for instance MODIS) is firstly to provide requirement of mentioned science. Secondly, help to provide deeper understand in case of “climatic neighborhood” concept. To assess the efficiency of the method cross validation indicators were used. Thermal images used in this study increase the accuracy for the winter and autumn in comparison to unused outputs. The provided results for spring and summer were good too. Also, the spatial impacts of thermal images on the results of autumn and spring are significant. This research indicated that using thermal images as auxiliary data have potential to improve spatial prediction accuracy and quality. At the end, we know that number of our observation stations are too low and considering the Kriging requirements like normal distribution and stationarity is toilsome but we should consider that this problem exist in the regions with low density of gauges and should find a way to enhance the air temperature interpolation in these cases.