Evaluation of the effect of spatial and temporal resolutions for digital change detection: case of forest fire

One of the most important subjects in remote sensing is digital change detection which identifies the differences between the before and after an event in spatial context. Even though lots of new satellite images have been launched to use with improvement of their resolutions, there needs to utilize at least two satellite images one of which must be acquired before the event happened. It is rather difficult to always find out proper spatial resolutions for change detection. The aim of this study is to evaluate and investigate the effect of spatial and temporal resolution on change detection. In this respect, the forest fire case was chosen as an event with five different satellite images (i.e. IKONOS, WorldView-2, ASTER, Landsat7 and Landsat8) having different acquisition time which is almost 10-year range. In this study, different spatial resolutions vs temporal resolutions were examined on different change detection algorithms which are image differencing, image rationing, NDVI differences, principle component analyses and minimum noise fractions (MNF). These data sets were tested on Adrasan area (Antalya, Turkiye) for change detection of forest fire. The obtained results has been shown that change detection using proposed MNF method for Landsat data sets have the highest accuracy with the value of 95.65%. Then the other high accuracy was obtained as 92.89% in MNF method for ASTER data sets. In addition, the other method (i.e. image ratio) is another high accuracy as 92.82% obtained for IKONOS and WorldViev-2 images. Finally, the relation between temporal resolution and spatial resolution has been generated as a graphical representation with spatial kernel size/filtering.