Deforestation detection using multitemporal satellite images

Indonesia a megadiverse country which has the largest area of forest with the richest biodiversity in the world. Unfortunately, the forest-loss in this country is quite big, especially in Kalimantan and Sumatera. Therefore, a robust method for deforestation method has to be developed to monitor deforestation to minimize the forest-loss in this country. In this paper, we proposed a method for deforestation detection called Multitemporal Deforestation Detection (MDD). The basic idea of this method is to utilize the difference of reflectance values on the target image and original image. Band selection was used to select bands in developing the algorithm. To improve the accuracy of the results, NDVI and dNBR were combined to the algorithm. As a result, the MDD can detect deforestation from forest to heterogeneous land cover such as open land, land clearing for plantation, urban, road and small road, and burnt area accurately. The advantage of the MDD is that it does not interfere cloud and cloud shadow on the original image. As it is an automatic algorithm, it can support in detecting deforestation for a big dataset.