Deforestation Mapping using Random Forest on The Google Earth Engine in the Forest Management Unit of the Larona and Ajatappareng Regions, South Sulawesi. Nuranisa Harfiana1,2, Syamsu Rijal1, and Munajat Nursaputra1*
1 Laboratory of Forestry Planning and Information System, Faculty of Forestry, Universitas Hasanuddin
2 Law Enforcement of the Environment and Forestry Region of Sulawesi
*Corresponding author: munajatnursaputra[at]unhas.ac.id
Abstract
The calculation of deforestation in Indonesia based on the land cover data series of the Ministry of Environment and Forestry is very dynamic. In the 2016-2017 period was 480 thousand Ha- the 2017-2018 period was 439.4 thousand Ha. However, it increased again in 2018-2019 to 462.5 thousand Ha and decreased again to 115.5 thousand Ha in 2019-2022. Information regarding deforestation rates at the site level, such as in forest management units spread across the Sulawesi region, is also very much needed to determine the condition of the forests in that area. Deforestation events can be measured with a mapping system, but it takes a long time if done with conventional image interpretation. The presence of a geospatial-based cloud computing platform such as the Google Earth Engine (GEE) provides a new option for researchers and policymakers who are interested in efficiently analyzing remote sensing data. GEE provides a classifier based on machine learning that used for multi-temporal land use mapping as a basis for mapping deforestation. By using the Random Forest method in GEE, it can be seen on the study area, namely the Larona and Ajatappareng forest management units in South Sulawesi, the deforestation rates were 3,964.5 Ha and 345.89 Ha on 2010-2021 period. The land cover classification accuracy test results in GEE also obtained an accuracy value with a kappa value of 0.96.
Keywords: Deforestation, GEE, Random Forest
Topic: Topik D: Climate change impacts for biodiversity
