Work

Estimation of Stand Volume Using Empirical Modeling Based on Multi-sensor Imagery with Random Forest Regression

Year: 2026
Type: article
Abstract: Stand volume estimation is the process of measuring the volume of tree stands, which serves as an indicator of forest ecosystem structure and biomass. This assessment supports various applications, including the implementation of reduced impact logging, a sustainable forest management strategy that employs selective harvesting based on tree maturity to preserve ecological integrity. However, measuring resources in large, complex forests with limited access is challenging for detailed resource measurement. This study was conducted in Central Kalimantan Province, as one of the production forest land uses with a naturally growing silviculture system. Forest canopy density (FCD) is employed to determine the classification of field samples. Furthermore, multi-sensor imagery was employed to derive input variables for stand volume estimation. Among the 19 predictors, the most influential were the FCD transformation, GEMI, and SAVI indices from Sentinel-2, along with bands 12 (SWIR) and 5 (red edge). Results indicated that variables derived from passive sensors contributed most significantly to the accuracy of the volume estimation model. The optimal stand volume estimation model generated from multi-sensor images had an R² value of 0.824 and RMSE of 9.7 at m-try 4 and n-tree 100. Estimation of stand volume using random forest obtained results in each sample plot size of 30 m² plots ranging from 0.05 m³ to 55.4 m³.
Language: English
Subfield: Forestry
Open Access status: gold
Estimated APC paid: $65
OpenAlex ID: W7167922166
FWCI:
Cites: 44
Cited by:
Related to: 10