Determining the Optimal Location of Endangered Species Habitats Using Remote Sensing and Species Distribution Models to Protect Biodiversity in Indonesia

 

Granted with the world's third-largest area of tropical rainforest, Indonesia is called a mega-biodiversity country with the second-highest level of biodiversity in the world. The diversity of flora and fauna in Indonesia is classified as rare and endangered due to forest fires, climate change, and anthropogenic factors. Strategies to protect biodiversity are required to address this situation. However, studies on conservation status are still lacking and limited to certain species because they have unique characteristics, so they do not always respond well to proposed strategies. Spatial modeling of potentially suitable habitats is essential in effective biodiversity conservation management. Using machine learning algorithms, more than 500 species points occurrence and ten environmental predictors consist of weather and climate aspects, topography, vegetation cover, air pollution, and fire prediction points from future climate model data to predict potential habitats for suitable species. From remote sensing data also analyzes the predictor variables that influence it. This study is resulting in predictions of flora and fauna habitat based on the Random Forest algorithm with suitable and unsuitable values. The novelty in the results of this study provides spatial modeling of the habitats of rare and endangered species so that policymakers can immediately take practical conservation actions to protect species from the threat of extinction.

Keywords: biodiversity, machine learning, remote sensing, and species distribution model.