Identification and classification of cloud computing-based vegetation index values on several lands used in Bogor Regency, Indonesia

Abstract

Understanding the threshold value classification from various vegetation types may help distinguish spectral reflectance differences in detailed land use studies. However, conducting all of the processes requires relatively large resources regarding manual computation, which could be surpassed by cloud computing. Unfortunately, in Bogor Regency, there is still a lack of research that studies the threshold value of various vegetation types related to forestry and plantation sectors. Land use categories were classified, and threshold values were determined, especially for selected vegetation types including teak, oil palm, rubber, pine, bamboo, and tea based on several vegetation indices in Bogor Regency using the Cloud-Computing platform. The data source was retrieved from 10-meters Sentinel-2 Satellite median imagery of January 2019 - June 2021. Land use maps were generated using Random Forest Algorithm from composite images. Meanwhile, the threshold value of each vegetation type was calculated from the average and standard deviation of NDVI, SAVI, EVI, ARVI, SLAVI, and GNDVI index. The result of the study showed forest and plantation area covers about 158,168.13 ha or 48.92 % of the study area. NDVI was found suitable to identify teak, SLAVI for rubber and pine, EVI for bamboo and tea, and GNDVI for oil palm vegetation.