Mapping bamboo species in Agusan del Norte, Caraga region using remote sensing and GIS techniques

Abstract

Mapping naturally occurring bamboo is challenging due to the mingling with other land uses. This study used high-resolution Planet Scope satellite imagery with a 3-meter spatial resolution to map bamboo distribution using two classifiers: Support Vector Machine and Random Forest. The comparative analysis showed that SVM outperformed Random Forest, achieving higher accuracy rates in bamboo classification. Specifically SVM achieved producer's and user's accuracy rates of92.86% and 83.87%, respectively, demonstrating its precision in identifying bamboo species. The study estimated 1,840 hectares of bamboo and 4.44 million bamboo culms in the province, with 39.79% located in the Municipality of Buenavista. A bamboo database was created within GIS software linked to Microsoft SQL Server Management using field survey data and the classified map. Mapping the bamboo species in the province is crucial due to the ecological and economic significance of bamboo. Accurate mapping helps in understanding the distribution of bamboo species, identifying areas for conservation and sustainable harvesting. This research highlights the effectiveness of SVM in classifying bamboo species from satellite imagery and the potential of remote sensing and GIS technologies for efficient natural resource management in tropical regions like Agusan del Norte.