A new model for cork weight estimation in Northern Portugal with methodology for construction of confidence intervals

Cork, a unique biological material, is a highly valued non-timber forest product. Portugal is the leading producer of cork with 52% of the world production. Tree cork weight models have been developed for Southern Portugal, but there are no representative published models for Northern Portugal. Because cork trees may have a different form between Northern and Southern Portugal, equations developed with southern data do not always predict well in the North. An analysis of eight tree variables revealed the interaction between tree circumference (outside bark at 1.3 m) (CSC) and debarking height (H-DEB) to be the best predictor variable. A nonlinear exponential function was chosen over a simple linear function due to a slight curvature exhibited in the plot of weight over CSC x H-DEB.. The paper concludes with a section on construction and use of confidence intervals about the mean and individual predictions from the nonlinear regression, plus a method to place bounds on an aggregate estimate for a stand. (C) 2001 Elsevier Science B.V. All rights reserved.