Pemetaan Sebaran dan Kerapatan Mangrove di Teluk Pangpang, Banyuwangi Menggunakan Maximum Likelihood dan Artificial Neural Network

Abstract

Mangrove menjadi ekosistem pesisir terbesar kedua di Indonesia serta memiliki peranan penting dalam siklus hidup biota laut sebagai nursery ground, spawning ground dan feeding ground. Penelitian ini bertujuan untuk memetakan sebaran dan kerapatan mangrove di Teluk Pangpang menggunakan citra Sentinel 2B. Metode Hemispherical Photography digunakan untuk memperoleh data tutupan kanopi mangrove serta analisis Normalized Difference Vegetation Index digunakan untuk memetakan sebaran kerapatan mangrove. Hasil uji akurasi 2 klasifikasi menunjukkan bahwa metode artificial neural network menghasilkan nilai Overall accuracy 92,73% (kappa 0,84) dan lebih baik daripada metode maximum likelihood dengan nilai Overall accuracy 89,09% (kappa 0,76). Analisis Normalized Difference Vegetation Index (NDVI) menghasilkan 3 kelas kerapatan mangrove. Luas yang dihasilkan adalah 925,8 ha dengan mangrove rapat 672,99 ha, mangrove sedang 61,83 ha, dan mangrove jarang 178,75 ha. Pengujian korelasi data survei tutupan kanopi dengan nilai NDVI didapatkan nilai korelasi (r) sebesar 0,893287 dan nilai determinasi (R2) sebesar 0,797961.

The mangrove ecosystem is the second largest coastal ecosystem in Indonesia, playing a crucial role in the life cycle of marine biota. It serves as a nursery ground, spawning ground, and feeding ground, making it an essential component of the marine ecology. This study aimed to map the distribution and density of the mangroves in Pangpang Bay, utilizing Sentinel 2B imagery. To gather data on mangrove canopy cover, the Hemispherical Photography method was employed, while the Normalized Difference Vegetation Index analysis was utilized to map the mangrove density distribution. The results of the classification accuracy tests revealed that the artificial neural network method produced an overall accuracy value of 92.73% (kappa 0.84), surpassing the maximum likelihood method, which scored an overall accuracy value of 89.09% (kappa 0.76). The Normalized Difference Vegetation Index (NDVI) analysis yielded 3 classes of mangrove densities, resulting in a total area of 925.8 ha, with 672.99 ha of dense mangroves, 61.83 ha of medium mangroves, and 178.75 ha of sparse mangroves. A correlation analysis of the canopy cover survey data with the NDVI value produced a correlation value (r) of 0.893287 and a determination value (R2) of 0.797961.