Biomass model development for carbon stock estimation in the tropical forest of Eastern India: an allometric approach

Abstract

Allometric regression models are one of the common methods of carbon stock estimation based on growing stock data conversion to estimates of above ground biomass (AGB). Therefore, allometric model selection is important functional aspect that has considerable influence on accuracy of biomass estimation. As destructive sampling is restricted in our study area, the site specific biomass model is developed for the first time based upon the forest inventory data that includes measurements of diameter at breast height (DBH) and tree height (H). To minimize the error in AGB estimation, intensive sampling was done where 78,201 individual tree were enumerated (6034 quadrats laid over 1207 plots). 20 locally abundant tree species were assessed. Tree volume and biomass were calculated and examined for best fit allometric model for the area. Species-specific models were established which best fits with the DBH as predictor variable. For multi-species models, inclusion of wood density (WD) enhanced the model fitness with increased adjusted R2 by 99.9%. Significant variations in predicted and observed values were noticed while considering the regional and pan-tropical models (model prediction error − 614.364 to 288.304%). Therefore, development of local models would provide more accurate AGB estimates. Best fit multi-species allometric model in our study is represented by ln (AGB) = a + b ln (DBH) + c ln (H) + d ln(WD). The equation developed for tropical forest of Eastern India applicable for Sal zone of Bihar is ln(AGB) = − 0.886 + 2ln(DBH) + ln(H) + ln(WD).