Identification of Land Criticality with the Application of Deep Learning in West Lahat District Using Sentinel-2A Imagery

Land is an important factor in human life. In addition of land use that continue to increase every year. Land use is an element of meeting needs. This situation often makes the condition of the land around it questionable the content and level of land productivity. Land whose productivity level is lost can cause critical land to occur. Coupled with the occurrence of uncontrolled development, land productivity has decreased. By using the application of remote sensing, it is able to monitor land conditions, one of which is by using Sentinel-2A data. Sentinel-2A image data was selected to identify the condition or distribution of critical land and critical land parameters that has the most influence on criticality level of the land with Sentinel-2A imagery with a spatial resolution of 10 meters for Red, Green, Blue, and Near-Infrared canals to perform NDVI classification processing. closely related to vegetation. Based on the Regulation of the Director General of Watershed Management and Social Forestry Number: P.4/V-SET/2013 concerning the Technical Guidelines for the Preparation of Spatial Data for Critical Lands, there are 5 parameters for determining the criticality of the processed land as indicators, including the level of erosion distribution, productivity land, land management, slope, and vegetation density. Based on the results of the study, the researchers found that the distribution of critical land in Lahat Regency was 19 hectares or 0.56%, the critical class was 36,090 hectares or 10.1%, the critical potential class was 142,140 hectares or 42.1%, the class which was slightly critical is 156,860 hectares or 46.5%, and non critical class is 3 hectares or 0.074%. for very critical class. These results can be seen with the parameter that most affects the criticality of the land is vegetation density.