Comparison of Three Classification Methods of Remote Sensing Techniques for the Purpose of Categorizing Habitat Types of Khar Yamaat Nature Reserve

Land cover change is one of the major problems facing the world, and as human needs increase, so does the inappropriate use of land (Geist et al., 2008), and climate change, precipitation variability, temperature rise, and the frequency of extreme natural disasters are the primary factors influencing land cover change (Bordi et al., 2004; Nandintsetseg & Shinoda, 2013). Therefore, we will investigate how the land cover in the Khar Yamaat Nature Reserve changes between 2014 and 2019. Using Landsat satellite images from September 2014 and September 2019, the land cover change was classified in three ways using the pixel-based supervised classification method: Maximum likelihood classification, K-Nearest neighbor, and Minimum distance classifier. Based on common plant species and environmental factors, land cover categories were divided into nine categories: mountain steppe, dry steppe, shrubland, degraded area, agriculture, sand, lake, river, and meadow. To detect land cover changes, the "Maximum likelihood classification" that accurately depicts the spatial distribution of each land cover class and has a high statistical significance was chosen. During the six-year period from 2014 to 2019, the area of dry steppes decreased by -42,232 hectares or 42.8%, mountain steppes decreased by -3,797 hectares (3.84%), rivers decreased by -1,000 hectares (1.1%), meadows decreased by -1,581 hectares (1.6%), and agriculture decreased by -723 hectares (0.7%), while degraded area increased by 39480. For 2014 and 2019, the overall accuracy was 95.96 (k=0.94) and 93.39 (k=0.93).