Regression Models for Estimating Aboveground Biomass and Stand Volume Using Landsat-Based Indices in Post-Mining Area

Abstract

This paper describes the use of remotely sensed data to measure vegetation variables such as basal area, biomass and stand volume. The objective of this research was developed regression models to estimate basal area (BA), aboveground biomass (AGB), and stand volume (SV) using Landsat-based vegetation indices. The examined vegetation indices were SAVI, MSAVI, EVI, NBR, NBR2 and NDMI. Regression models were developed based on least-squared method using several forms of equation, i.e., linear, exponential, power, logarithm and polynomial. Among those models, it was recognized that the best fit of model was obtained from the exponential model, log (y) = ax + b for estimating BA, AGB & SV. The MSAVI had been identified as the most accurate independent variable to estimates basal area with R² of 0.70 and average verification values of 16.39% (4%32.66%); while the EVI become the best independent variable for estimating aboveground biomass (AGB) with R2 of 0.72 and average of verification values of 18,10% (9%-28.01%); and the NDMI was recognized to be the best independent variable to estimate stand volume with R2 of 0.69 and average of verification values of 24.37% (-15%-38.11%).