| Abstract: | Tree biomass plays an important role in sustainable management and in estimating forest carbon stocks. The objective of this study was to select the best model for measuring stem biomass of Acacia auriculiformis in the study area. Data from five hillocks and 120 individual trees from each hillock were used in this study. Twelve different forms of linear, power and exponential equations were compared in this study to select the best model. Two models (VI and XI) were selected based on R 2, adjusted R 2, the Akaike information criterion, F-statistics and the five assumptions of linear regression. Model VI was discarded based on the Durbin-Watson value of autocorrelation of the residuals, then the ARIMA (2, 0, 1) model was used to remove the autocorrelation from the model and the final bias-corrected model XI was derived. The model was validated with a test data set having the same range of DBH and stem height of the training data set on the basis of linear regression, Morisita's similarity index, and t-test for mean difference between predicted and expected biomass. A comparison between the best logarithmic and non-linear allometric model shows that the non-linear model produces systematic biases and overestimates stem biomass for larger trees. The overall results showed that the bias-corrected logarithmic model XI can be used efficiently for estimating stem biomass of A. auriculiformis in the northeastern region of Bangladesh. |
