| dc.description.abstract | Over the past two decades, mangrove forests have undergone substantial deforestation.
North Sumatra Province, as one of Indonesia's largest repository of mangrove forests,
covering approximately 57,490 hectares, primarily situated along the East Coast of
Sumatra. Notably, Pulau Sembilan Village, nestled within Pangkalan Susu District,
Langkat Regency, North Sumatra, emerges as a notable hub of mangrove growth,
featuring a thickness range spanning from 100m to 1700m. Despite its considerable
expanse, this ecosystem faces imminent threats posed by human-induced activities such
as illicit logging, waste accumulation, and the overarching impacts of global climate
change. Consequently, this study employs the Gaussian Process Regression for
Classification (GPRC) methodology to delineate mangrove planting zones within Pulau
Sembilan Village, Langkat Regency, North Sumatra. Leveraging water content data
alongside diverse environmental features, the GPRC model undergoes training
employing three distinct kernels: Radial Basis Function (RBF), Matern, and Rational
Quadratic. The training regimen entails 25 iterations per kernel, with performance
assessments conducted via accuracy, F1-Score, precision, and recall metrics.
Outcomes indicate the Rational Quadratic kernel yielding the highest accuracy at
98.29%, followed by the Matern kernel at 98%, and the RBF kernel at 96.58%.
Subsequent testing underscores the superiority of the Rational Quadratic kernel,
attaining an accuracy rate of 98.15%. These findings underscore the efficacy of GPRC
in classifying mangrove planting zones, with the Rational Quadratic kernel emerging
as particularly adept. Ultimately, this research contributes valuable insights towards
mangrove planting zone management and offers a robust foundation for future
investigations into GPRC-based classification methodologies. | en_US |
