<![CDATA[Gunung Halimun Salak National Park (GHSNP) is one of the most biodiversity-rich protected areas in Java, yet it remains highly vulnerable to deforestation and forest degradation. This study examines forest cover dynamics from 1994 to 2024 and projects village-level encroachment risk for 2034. Landsat 5 TM, Landsat 7 ETM+, and Landsat 8–9 OLI imagery were processed in Google Earth Engine to generate the Normalized Difference Fraction Index (NDFI) using spectral mixture analysis of GV, NPV, soil, and shade fractions. Changes in NDFI (ΔNDFI) were used to classify degradation, deforestation, regrowth, and intact forest. Encroachment risk mapping was modeled using a 3 × 3 kernel neighborhood with two analytical approaches: the sum of risk weight and the majority of risks around. Forest cover declined by 19,424 ha between 1994 and 2004, largely driven by illegal encroachment linked to governance uncertainty in 2003. An increase of 6,678 ha during 2004–2014 reflects the impact of restoration initiatives and strengthened area protection, although a subsequent decline of 1,992 ha occurred between 2014 and 2024 due to renewed encroachment. Model evaluation indicates low predictive performance for both kernel methods (Precision 4%). Despite this limitation, areas of elevated risk consistently appeared along forest edges near settlements and footpath access routes. Citorek Kidul was identified as the village most susceptible to encroachment in 2034. Improving the accuracy of encroachment prediction will require the integration of socio-economic drivers and advanced machine-learning approaches capable of capturing the complex and non-linear patterns of forest encroachment. ]]>
Copyrights © 2025
