MODELING FOREST ENCROACHMENT AND LAND COVER CHANGE IN MOUNT MASIGIT KAREUMBI, WEST JAVA

• Main Author: SUSILOWARDANI (NIM : 11514032), DWI
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/26798
• Institution: Institut Teknologi Bandung

<p align="justify"> Encroachment is one of the driving factors of deforestation in Indonesia. Forest encroachment commonly occurs in conservation areas in Indonesia, including in Mount Masigit Kareumbi (MMK). Formulating strategies in dealing with encroachment inside the area requires basic information about the distribution of encroached areas and variables influencing the encroachment. The study aims to identify encroached areas inside MMK during 2001-2015, to model risk of forest encroachment, and to predict land cover changes in the period of 2015 - 2029. Study area covers the whole of MMK with an area of 12420.7 hectares. Encroached area was identified using on screen digitizing technique using Landsat 7 ETM + 2001 and Landsat 8 OLI 2015 images. Seven predictor variables were used for forest encroachment modelling, i.e. distance from cultivation land (X1), elevation (X2), distance from the road network (X3), population density (X4), slope (X5), distance from settlement (X6), and distance from the river (X7). We applied a binary logistic regression analysis for this modelling using IDRISI Selva with stepswise backward elimination method. Land cover change was modelled using an integration of Markov Chain and Cellular Automata (CA) analyzes. This study identifies the distribution and the total encroached area in MMK, i.e. about 428 ha or 3.56% of the area. The encroached areas cover 18 villages: Sindanggalih, Mekarbakti, Cilembu, Cimarias, Pasanggrahan, Sukajaya, Cipameungpeuk, Cikondang, Tanjunghurip, Bangbayang, Kaduwulung, Sundamekar, Cimarga, Cipeuteuy, Buana Mekar, Pangeureunan, Tanjungwangi, and Sindulang. The best binary logistic regression equation is logit p = 0.8156 - 0.003713 * X2 + 0.000862 * X4 + 0.004310 * X5 - 0.000665 * X6. This model performs well with the value of Receiver Operating Characteristics (ROC) of 87%, and Area Under ROC Curve (AUC) of 88%. CA Markov Chain analysis for land cover prediction in 2029 shows that area of forests, shrubs, and agricultural land in MMK are 9443 hectares, 2017 hectares, and 953 hectares respectively. <br /> <p align="justify">