Forest above-ground woody biomass estimation using multi-temporal space-borne LiDAR data in a managed forest at Haldwani, India

Abstract

Measurement of forest aboveground biomass is critical to account for carbon budgeting, carbon flux monitoring, biodiversity health monitoring, and climate change studies. Therefore, there is a crucial requirement to develop methods for improvement in forest biomass estimation. This study uses recently launched Global Ecosystem Dynamic Investigations (GEDI) mission LiDAR data combined with field-measured biomass and geospatial analysis to estimate forest aboveground woody biomass (AGB) in the managed forest. From a total of 300 predictor variables obtained from GEDI height metrics (100) and derived products (200), an exhaustive search is performed using regsubsets function in the leaps package, which uses Adjusted correlation coefficient (Adj.R2) statistical measure to select suitable predictor variables for AGB estimation. We constrained the search algorithm with a maximum number of variables to be equal to or lower than four to avoid over-fitting. Therefore we selected the best set of variables with the least root mean squared error (RMSE) and high Adj.R2, and developed an AGB model using regression analysis. The developed model predicted AGB with R2=0.75, RMSE = 32.06 Mg/ha, and relative RMSE = 27.26%. The results obtained in this study illustrate the potential of GEDI LiDAR data to model forest AGB with improved accuracy.