Kajian Dampak Deforestasi terhadap Tingkat Kerawanan Longsor Berbasis Sistem Informasi Geografis di Wilayah Pulau Sumatra.

Abstract

Deforestasi yang terjadi di Pulau Sumatra berpotensi meningkatkan kerawanan longsor melalui perubahan tutupan lahan yang memengaruhi stabilitas lereng. Penelitian ini bertujuan menganalisis perubahan tutupan lahan selama periode 2015–2025, mengkaji pengaruh deforestasi dan variasi curah hujan terhadap kerawanan longsor, serta mengevaluasi kesesuaian hasil pemodelan dengan kejadian longsor historis. Analisis perubahan tutupan lahan dilakukan di Google Earth Engine, sedangkan pemodelan kerawanan longsor dilakukan dengan metode weighted overlay. Empat skenario disusun melalui kombinasi bobot tutupan lahan normal dan modifikasi serta curah hujan tahunan dan ekstrem. Hasil penelitian menunjukkan bahwa terjadi kehilangan tutupan hutan terbesar sebesar 393.303,85 ha. Pemodelan kerawanan longsor menunjukkan bahwa peningkatan bobot tutupan lahan dan penggunaan curah hujan ekstrem cenderung meningkatkan luas area kerawanan tinggi. Skenario 1 menghasilkan akurasi tertinggi sebesar 78% yang menunjukkan model dengan bobot tutupan lahan modifikasi dan curah hujan ekstrem memiliki kemampuan terbaik dalam merepresentasikan distribusi kejadian longsor historis.

Deforestation on Sumatra Island has the potential to increase landslide vulnerability through changes in land cover that affect slope stability. This study aims to analyze land cover changes during the 2015–2025 period, examine the influence of deforestation and rainfall variations on landslide vulnerability, and evaluate the suitability of the modeling results with historical landslide events. Land cover change analysis was conducted in Google Earth Engine, while landslide vulnerability modeling was conducted using the weighted overlay method. Four skenarios were developed through a combination of normal and modified land cover weights and annual and extreme rainfall. The results showed that the largest forest cover loss occurred at 393,303.85 ha. Landslide vulnerability modeling showed that increasing land cover weights and the use of extreme rainfall tended to increase the area of high vulnerability. Skenario 1 produced the highest accuracy of 78%, indicating that the model with modification land cover weights and extreme rainfall had the best ability to represent the distribution of historical landslide events.