Electronic Theses and Dissertation

Ekosistem mangrove di wilayah pesisir Kota Banda Aceh menghadapi tekanan besar akibat urbanisasi yang terus berkembang. Penelitian ini bertujuan memetakan distribusi spasial ekosistem mangrove menggunakan kombinasi tujuh indeks vegetasi NDVI, MSR-RE, DVI, IRECI, SR-RE, ClRE, dan RVI yang diintegrasikan dengan algoritma machine learning RandomForest berbasis platform Google Earth Engine (GEE). Citra satelit Sentinel-2A digunakan sebagai sumber data utama, didukung 50 titik validasi mangrove untuk pelatihan dan validasi model. Hasil klasifikasi menunjukkan total luas tutupan mangrove mencapai 257 Ha atau sekitar 8% dari total luas penelitian (3.201 Ha), dengan sebaran tertinggi di Kecamatan Syiah Kuala (97 Ha), diikuti Kuta Raja (57 Ha), Kuta Alam (52 Ha), dan Meuraxa (51 Ha). Distribusi mangrove terkonsentrasi di sepanjang garis pantai dan muara sungai, namun terjadi fragmentasi signifikan yang mengindikasikan tekanan antropogenik. Evaluasi akurasi menghasilkan overall accuracy 0,88 dan Kappa Coefficient 0,77 (substantial agreement), membuktikan bahwa integrasi multi-indeks vegetasi dengan RandomForest merupakan pendekatan yang efektif, akurat, dan efisien untuk pemetaan serta pemantauan ekosistem mangrove secara berkelanjutan.

Mangrove ecosystems in the coastal areas of Banda Aceh are facing significant pressure due to ongoing urbanization. This study aims to map the spatial distribution of mangrove ecosystems using a combination of seven vegetation indices NDVI, MSR-RE, DVI, IRECI, SR-RE, ClRE, and RVI integrated with the RandomForest machine learning algorithm on the Google Earth Engine (GEE) platform. Sentinel-2A satellite imagery was used as the primary data source, supported by 50 ground truth points for mangrove areas to train and validate the model. The classification results show that the total area of mangrove cover reaches 257 ha, or approximately 8% of the total study area (3,201 ha), with the highest concentration in Syiah Kuala District (97 ha), followed by Kuta Raja (57 ha), Kuta Alam (52 ha), and Meuraxa (51 ha). Mangrove distribution is concentrated along the coastline and river estuaries, but significant fragmentation indicates anthropogenic pressure. Accuracy evaluation yielded an overall accuracy of 0.88 and a Kappa Coefficient of 0.77 (substantial agreement), proving that the integration of multi-vegetation indices with RandomForest is an effective, accurate, and efficient approach for the mapping and sustainable monitoring of mangrove ecosystems.