Prediction models for a young Kawayan tinik (Bambusa spinosa Roxb.) plantation in Nueva Ecija, Philippines

Rosalie  Mendoza; Ramon Razal; Canesio Predo; Willie Abasolo; Roberto Visco

Rosalie C. Mendoza Department of Forest Products and Paper Science, College of Forestry and Natural Resources, University of the Philippines Los Baños
Ramon A. Razal Department of Forest Products and Paper Science, College of Forestry and Natural Resources, University of the Philippines Los Baños
Canesio D. Predo Institute of Renewable Natural Resources, College of Forestry and Natural Resources, University of the Philippines Los Baños
Willie P. Abasolo Department of Forest Products and Paper Science, College of Forestry and Natural Resources, University of the Philippines Los Baños
Roberto G. Visco Institute of Renewable Natural Resources, College of Forestry and Natural Resources, University of the Philippines Los Baños

Keywords: bioenergy, linear regression analysis, non-linear regression analysis

Abstract

The objective of this study is to generate prediction models for different characteristics (clump and culm variables, aboveground dry biomass, chemical properties, and gross calorific values) of Kawayan tinik (Bambusa spinosa Roxb.) as a function of clump age. Biomass yield models were developed over a three-year-period that can serve as basis for making decisions in the planning and design of bamboo energy plantations. Linear and non-linear models best fit the data. Models considered were linear regression (double log model) and non-linear regression specifically exponential, logarithmic, and Gompertz. Prediction models with high coefficient of determination (R2 ) values, adjusted R2 , low root mean square error (RMSE), and significant coefficients were generated for all tested dependent variables except for culm dry biomass. Of all the models used, the exponential model fitted best with the data, specifically the exp2, followed by exp2a, and then the linear regression model. In terms of dry biomass, multiple linear regression model in double natural logarithmic form showed that culm number per clump and culm diameter are important variables in estimating culm dry biomass. Branches+leaves dry biomass can be predicted by culm number per clump, culm diameter, and clump age. These prediction models can be used to estimate the bamboo characteristics included in the study given the same clump age and site conditions.