Hubungan Fragmentasi Lanskap dan Kelembapan Permukaan dengan Suhu Permukaan di Cagar Biosfer Giam Siak Kecil-Bukit Batu

Abstract

Cagar Biosfer Giam Siak Kecil-Bukit Batu merupakan lanskap gambut yang mengalami tekanan penggunaan lahan dan perubahan biofisik permukaan. Penelitian ini bertujuan menganalisis tutupan lahan, kelembapan permukaan, suhu permukaan, fragmentasi lanskap, serta hubungan antarvariabel. Analisis menggunakan citra Landsat 9 tahun 2025, klasifikasi random forest, normalized difference moisture index (NDMI), land surface temperature (LST), metrik patch density (PD), edge density (ED), dan aggregation index (AI) pada grid 500 m × 500 m, serta Korelasi Spearman untuk menguji hubungan antarvariabel. Hasil klasifikasi mencapai overall accuracy 97,29% dan Kappa 0,967. Tutupan lahan didominasi kelapa sawit, hutan, dan lahan terbuka. Zona inti memiliki NDMI tertinggi dan LST terendah, sedangkan area buffer 5 km berpola sebaliknya. NDMI berhubungan negatif sedang dengan LST (? = -0,678), sedangkan PD, ED, dan AI masing-masing berhubungan lemah dengan LST (? = +0,285. +0,296. -0,293). NDMI menjadi hubungan bivariat paling kuat dengan LST, sementara metrik fragmentasi lanskap menjadi konteks spasial pemantauan tekanan termal kawasan.

Giam Siak Kecil-Bukit Batu Biosphere Reserve is a peatland landscape facing land-use pressure and surface biophysical changes. This study aimed to analyze land cover, surface moisture, land surface temperature, landscape fragmentation, and their relationships. The analysis used 2025 Landsat 9 imagery, random forest classification, normalized difference moisture index (NDMI), land surface temperature (LST), landscape metrics of patch density (PD), edge density (ED), and aggregation index (AI) within 500 m × 500 m grids, and Spearman Correlation to examine the relationships among variables. The classification achieved 97.29% overall accuracy and a Kappa coefficient of 0.967. Land cover was dominated by oil palm, forest, and open land. The core zone had the highest NDMI and lowest LST, whereas the 5 km buffer area showed the opposite pattern. NDMI had a moderate negative relationship with LST (? = -0.678), while PD, ED, and AI each had a weak relationship with LST (? = +0,285. +0,296. -0,293). NDMI had the strongest bivariate relationship with LST, while landscape-fragmentation metrics provided spatial context for monitoring thermal pressure in the reserve.