Land cover identification using Pleiades satellite imagery by comparison of NDVI and BI method in Jatinangor, West Java

Abstract

Land cover monitoring is important to maintaining human needs and ecological function balance that will be able to observe, analyse, and present land use form. Some methods that can be implemented to this case are Normalized Difference Vegetation Index (NDVI) and Bare soil Index (BI). Land cover identification is using Pleiades images in 2018 for Jatinangor area, with land cover classes were built in five class, divided by producer's and user's accuracy and validated with field survey results. Accuracy of NDVI method is 69.56% with kappa 0.57 also producer's accuracy is 73.94% water bodies, 84.88% built land, 67.37% open field, 98.79% crops field, and 98.79% vegetation. User's accuracy in a row for each class are 26.86% 82.78%, 20.01%, 76.15%, and 75.37%. The Accuracy of BI method is 69.17% with kappa 0.52. Producer's accuracy for each class is 73.28% water bodies, 70.71% land build, 52.83% open filed, 63,04% crop field, and vegetation with 81.29%. User's accuracy in a row for each class is 17.05%, 81.29%, 21.44%, 65.13%, and 88.54%. Validation of NDVI more fit in the field than BI that can be concluded NDVI has a good result than BI.