Mitigation & Identification for Local Aridity, Based of Vegetation Indices Combined with Spatial Statistics & Clustering K Means

Abstract

This research aims to develop new methods of mapping aridity risk zones and their potential impacts on land fires using a combination of indices to identify land fires such as the CSI and NBR and the indices for NDVI and SAVI. The research location is in Gunung Merapi National Park (TNGM) and Gunung Merbabu National Park (TNGMb), in Central Java Province and Yogyakarta. The data used in this research is Landsat 8 OLI image year 2010-2018, DEM data from ASTER image in TNGM and TNGMb area using landsat 8 OLI image specification. The research was conducted in 3 stages such as pre-processing, image data extraction and post-processing. Global and Local Moran's Analysis on NDVI, SAVI, CSI and NBR vegetation indices data can be used as an indicator of aridity and potential land fires. The experiments show that the average is in class 4 including the moderate greenish classification. Moderate greenish is interpreted that the study area is overgrown with meadows, shrubs, barren, sandy, rocky areas and a low population of vegetation canopy that shows that the area is on the surrounding mountain peaks. The results of the analysis shows Positive Spatial Autocorrelation, the phenomenon of aridity has spatial connectivity between observed regions. Analysis of K Means on the high vegetation density conditions shows that the weight of the distance between the vegetation data to the centroid is shorter, therefore the data is concentrated on a region. In low vegetation density conditions, the weight of the distance between the data to the centroid is increasingly wide, therefore the data looks more widely distributed.