Ridge Regression Model for the Estimation of Total Carbon Sequestered by Forest Species

Abstract

In this chapter, the study of total carbon sequestration has been taken for Acacia catechu, because of its highest relative dominance, relative density, relative frequency and importance value index in forest land-use system. The ridge regression methods have been used to estimate the total carbon sequestration (dependent variable) of this species in Kandi belt of Jammu region of Jammu and Kashmir state located in the foothill zone of Jammu Shivaliks. The explanatory variables diameter at breast height (DBH), height, stem biomass (SB), branch biomass, leaf biomass, below-ground biomass and total above-ground biomass (TB) were chosen for the study. It has been observed that the explanatory variables were highly correlated and indicates the presence of multicollinearity and hence estimates based on the classical ordinary least square method are not precise. Ridge regression method was attempted to deal with the problem of multicollinearity. The results show that the optimum value of ridge constant is 0.02 and the variables DBH, SB and TB are significant variables to increase the total carbon content sequestered by the species Acacia catechu. The estimates of parameters through ridge regression technique were more stable and more reliable than ordinary least square on the basis of size, sign and significance of the regression parameters.