Mapping mangrove forest distribution on Banten, Jakarta, and West Java Ecotone Zone from Sentinel-2-derived indices using cloud computing based Random Forest

Ekosistem mangrove merupakan kawasan yang sangat potensial, umumnya berada di kawasan ekoton (kombinasi kawasan intertidal dan supratidal), dimana terdapat interaksi antara perairan (laut, air payau, dan sungai) dengan kawasan daratan. Indonesia khususnya wilayah Banten dan Jawa Barat memiliki kawasan mangrove yang sangat luas dan saat ini terancam alih fungsi lahan. Apalagi pemetaan sebaran hutan bakau menggunakan platform Google Earth Engine berbasis Cloud Computing kurang dipublikasikan. Oleh karena itu, penelitian ini dilakukan dengan memperkenalkan sebaran hutan mangrove yang melibatkan metode algoritma klasifikasi Random Forest (RF), dan mencari modifikasi indeks yang terbaik. Uji kombinasi dilakukan dengan melibatkan indeks NDVI, EVI, ARVI, SLAVI, IRECI, RVI, DVI, SAVI, IBI, GNDVI, NDWI, MNDWI, dan LSWI. Sebaran mangrove terdapat di tiga provinsi (Jawa Barat, Banten, dan DKI Jakarta) yaitu seluas 933,54 ha (8,372%), 1.537,89 ha (18,231%), dan 8.184,82 ha (73,397%). Dari 70 pengujian kombinasi, indeks LSWI (K13, Type-A) merupakan kombinasi dengan tingkat akurasi terendah sebesar 58,45% (Overal Accuracy) dan 39,59 (Kappa statistik), dan kombinasi K23 (SAVI-MNDWI-IBI) merupakan kombinasi yang terbaik yaitu 96,48% dan 92,79. Hasil dan rekomendasi dalam penelitian ini diharapkan dapat digunakan sebagai acuan dalam menentukan kebijakan perlindungan kawasan mangrove dan referensi untuk penelitian selanjutnya.Mangrove ecosystem is a very potential area, generally located in ecoton areas (a combination of intertidal and supratidal areas), where there is interaction between waters (sea, brackish water, and rivers) with land areas. Indonesia, especially the Banten and West Java regions, have vast mangrove areas and are currently under threat of land conversion. Moreover, mapping the distribution of mangrove forests using the Google Earth Engine platform based on Cloud Computing is less published. Therefore, this research was conducted by introducing the distribution of mangrove forests which involved the Random Forest (RF) classification algorithm method, and looking for the best modification of the index. The combination test was carried out by involving the NDVI, EVI, ARVI, SLAVI, IRECI, RVI, DVI, SAVI, IBI, GNDVI, NDWI, MNDWI, and LSWI indexes. There is a distribution of mangroves in three provinces (West Java, Banten, and Jakarta) which are 933.54 ha (8.372%), 1,537.89 ha (18.231%), and 8,184.82 ha (73.397%). Of the 70 combination tests, the LSWI index (K13, Type-A) is the combination with the lowest accuracy rate of 58.45% (Overal Accuracy) and 39.59 (Kappa statistic), and the combination of K23 (SAVI-MNDWI-IBI) is a combination the best are 96.48% and 92.79. The results and recommendations in this study are expected to be used as a reference in determining policies for the protection of mangrove areas and a reference for further researc