Generating Land Use/Land Cover Change Index for Satelite Images Data Verification (Case Study: West Java Province)

The majority of land use/landcover (LULC) data were derive from satellite image data of varying resolution. These images were interpreted to classify the LULC in a particular area at a certain period. However, the process of visual interpretation might produce a pattern of land class change that does not match actual condition because it always has the potential to create different classification outcomes if compared using image data of higher resolution. Thus, it becomes highly important to verify satellite image data of the study area by generating an index of LULC change. In this research, proportional stratified random sampling method was used. In this method, each sample of recorded LULC change was verified using a high-resolution image. The verification results were then tested using one-tailed t-test with 90% confidence level. The product was an index of LULC Change where 1 (the number one) signified that change did/might happen and 0 (the number zero) signified that change did/might not happen. From this research, LULC Change index with a resolution of 30 meters for the area of West Java Province was obtained. Cellular Autamata-Markov Chain model was used to test the application of this index to predict land cover in 2015. The most accurate result (accurate value = 40.3797%) was obtained from using Von Neuman nearest neighbors method with 7x7 size. This value was 0.9509 higher than the value resulted from LULC prediction without using the index.