Diameter Distribution P. occidentalis_P11

Slide 1:

Pinus occidentalis is a major economic tree species in the Dominican Republic, contributing close to 97% of the total annual volume harvested within the country. To model growth of P. occidentalis forest stands in three different ecological zones within La Sierra region in the Dominican Republic, a diameter-distribution model was employed. Characterization of these stands will allow us to predict the number of trees by diameter class at future points in time, allowing forest managers to define the optimum economic rotation for particular sites in the region, detect the excess or lack of specific wood products, and make multiple and valuable decisions regarding the management of P. occidentalis in the region. We chose to model the diameter-distribution using the three-parameter Weibull function based on its flexibility and creditability of its diameter predictions. Weibull parameters were estimated for each of 25 stands from three ecological zones. Estimated Weibull parameters were then directly predicted using linear mixed regression functions with age, site quality and density as predictor variables. The prediction equations were employed to predict Weibull distributions and derive yield per hectare estimates for each stand, which were compared to observed yields. Developing a Diameter-Distribution Based, Whole Stand Growth Prediction System for Pinus occidentalis Stands in La Sierra, Dominican Republic. S.W. Bueno and E. Bevilacqua, Department of Forest and Natural Resources Management. SUNY_ESF. Syracuse, NY 13210 The objectives for this research were to: 1) determine if the Weibull probability density function (pdf) is flexible enough to describe dbh distributions of P. occidentalis stands found in the D.R.; 2) determine which, if any, stand level characteristics are suitable predictors of Weibull pdf parameters; and 3) explore the precision and accuracy of stand yield estimates derived from predicted diameter distributions. Area of Study The study area is a region of approximately 2,000 km 2 in the north central portion of Cordillera Central, Dominican Republic. The data available for model development were from twenty-five P. occidentalis natural stands established on three different life zones in La Sierra region, Dominican Republic, nine stands in the humid zone, six stands in the intermediate zone, and ten in the dry region. The dry life zone corresponds to the formal denomination Subtropical Dry Forest; the humid life zone is denominated formally as Subtropical Very Humid Forest; and the intermediate zone, located between the two previously mentioned, has been denominated as Subtropical Humid Forest (Holdridge, 1987). Introduction Method Objective Results Conclusion Sensitive Analysis Best model and covariance structure for the 3-parameter Weibull function in “La Sierra”, D.R. Parameter Covariance Model Param . Estim . Std. Error Prob. Shape Unstruct . 6 Age 0.209 0.066 0.002 Scale Unstruct . 2 Intercept 30.859 3.486 <0.0001 TPH -0.0234 0.006 0.0004 Location Unstruct . 8 Intercept 10.792 2.218 <0.0001 Age - 0.115 0.055 0.0450 TPH - 0.011 0.001 <0.0001 BAPH 0.203 0.042 <0.0001 Fit Statistics for the Nonlinear Mixed Procedure in Height-DBH Relationships Weibull Parameter Estimates STATISTIC ZONE HUMID INTERMEDIATE DRY MSE 8.081 9.645 1.481 R 2 0.81 0.66 0.96 Bias 2.239 2.591 1.294 RMSE 2.843 3.106 1.217 MAD 2.239 2.591 1.294 Comparison of Actual and Estimated Stand Yields The diameter distribution model will allow to predict the number of trees by diameter class of P. occidentalis at future points in time based on current stand attributes. The model will allow forest managers to define the optimum economic rotation for particular sites in the region, detect the excess or lack of specific wood products and take multiple and valuable decisions regarding the management of P. occidentalis in the region. Weibull Parameter Estimation Obtaining the Best Height-DBH Model Residual vs. Predicted Values: Height-DBH Models Residual vs Predicted Values, H-DBH Exponential Function Modified Logistic Function Best Height-DBH Models Found Volume by DBH Class Average Yield (m 3 per hectare) Zone Observed Predicted Humid 274 198 Intermediate 169 156 Dry 214 159 Overall 219 171