Visualisasi Perubahan Tutupan Lahan Akibat Bencana Lumpur Lapindo Pada Wilayah Kerja Pusat Pengendalian Lumpur Sidoarjo Menggunakan Metode Object-Based Image Analysis (OBIA) Periode 2006-2025
Nabila Aryuni Nuroktavia, Ir.Ruli Andaru, S.., M.Eng.,Ph.D., IPU
2026 | Skripsi | TEKNIK GEODESI
The Lapindo mudflow disaster, which has been ongoing in Sidoarjo Regency since 2006, continues to cause land cover change in the surrounding area to this day. However, no visualization of land cover change covering the entire disaster period has been made available on a continuous basis, so that the pattern and rate of mudflow inundation change since the early eruption phase have not been comprehensively observed. This research aims to visualize land cover change caused by the Lapindo mudflow disaster within the 2023 working area boundary of the Sidoarjo Mudflow Mitigation Agency (Pusat Pengendalian Lumpur Sidoarjo/PPLS), covering the study period of 2006–2025, through the application of the Object-Based Image Analysis (OBIA) method on the Google Earth Engine (GEE) platform.
This research employs an integration of Landsat 7, Landsat 8, and Sentinel-2 imagery to map land cover within the PPLS working area for the 2006–2025 period. Object segmentation was performed using the Simple Non-Iterative Clustering (SNIC) algorithm, while classification was carried out using the Random Forest algorithm with 250 decision trees. Object features were extracted using spectral indices selected based on the characteristics of each sensor, comprising a total of six indices, namely NDVI, MNDWI, BSI, TCW, NDBI, and UI. Landsat imagery was classified into two classes (Mudflow and Non-Mudflow), while Sentinel-2 imagery was classified into five classes (Mudflow, Vegetation, Built-up, Water Body, and Bare Soil). Classification results were evaluated using Overall Accuracy, Kappa Coefficient, and mean Intersection over Union (mIoU).
The results reveal three phases of land cover dynamics within the PPLS working area: an acute expansion phase (2006–2010) with a 153.44 percent increase in mudflow area, a slow expansion phase (2010–2015) that reached the highest cumulative extent with a 3.60 percent increase, and a stabilization phase (2015–2025) with a 2.83 percent decrease in extent. During the acute expansion phase, the largest conversion occurred from the Non-Mudflow class to the Mudflow class, indicating the direct impact of the mudflow eruption on the surrounding land cover. During the stabilization phase, the Sentinel-2 classification results indicate an ongoing ecological recovery process, marked by a 25.36 percent increase in vegetation extent and a decrease in bare soil extent over the 2020–2025 period. Based on accuracy testing using the Confusion Matrix, Overall Accuracy values ranged between 85–95 percent with a Kappa Coefficient of 0.700–0.900, meeting the minimum land cover mapping accuracy standard of 85 percent set by the United States Geological Survey (USGS). The spatial validity of the mapping results was further supported by mIoU values that consistently increased from 0.7513 for Landsat 7 in 2006 to 0.9260 for Landsat 8 in 2025. The final research output, comprising land cover change caused by the Lapindo mudflow within the PPLS working area, is presented as an interactive dashboard through the GEE App platform, which is publicly accessible.
Kata Kunci : GEE, Lumpur Lapindo, OBIA, Random Forest