Journal
GEOGRAPHIA TECHNICA
Volume 17, Issue 1, Pages 46-57Publisher
GEOGRAPHIA TECHNICA ASSOC - GT Assoc
DOI: 10.21163/GT_2022.171.04
Keywords
Peat thickness; Carbon stock; Machine learning; Remote Sensing; Tropical peatland
Categories
Ask authors/readers for more resources
This study aims to map the spatial distribution of carbon stock based on peat thickness modeling using machine learning algorithms. The results showed that the Random Forest model performed the best among other models in estimating peat thickness.
Indonesia is one of the countries with the largest area of tropical peatlands in the world. These wide peatlands have a vital role in the carbon cycle and carbon storage in huge quantities, thus strict conservation in the area is necessary. One effort to carry out conservation is to understand the spatial distribution of carbon stock in peatlands. This study aims to map the spatial distribution of carbon stock based on peat thickness modeling using machine learning algorithms, i.e., Random Forest (RF), Quantile Regression Forests (QRF), and Cubist. A case study was conducted in a part of Lake Sentarum National Park, Indonesia where human interference is still limited. Digital elevation model (DEM) and synthetic aperture radar (SAR) data were included as the input variables. The results showed that RF performed the best among other models to estimate peat thickness (mean = 1.877 m) with an RMSE (root mean square error) of 0.483 m and an R-2 of 0.786. DEMs are the most important parameters in our analysis compared to SAR data. Based on the best model, we estimated the total volume of 5,112,687 m(3) in the study area, produced at 12.5 m resolution, which was then converted to a total of carbon stock at 0.337 +/- 0.106 Mt carbon.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available