Integrating Micro-Scale Timbering Events and Decisionmaking into Landscape Models Using Logistic and Multilevel Regression

Abstract

The objectives of this study were to identify potential drivers of micro-scale commercial timbering events in West Virginia, USA, and to develop regression-based models of these events for use in integrated landscape models. Logistic and multilevel regression techniques are used to model timbering events at the forest stand and tree scale. At the stand scale, the models indicated that forest stand value and ownership were key drivers of stand selection decisions, whereas measures of infrastructure, population growth, and distance to mills explained some variance. At the tree selection scale, the value of the tree, ownership regime, and stand value were key drivers of tree removal decisions, whereas measures of human development explained some variance. Overall, private lands were more than twice as likely to be selected for timbering events as public lands, other factors being equal, and the opposite effect was seen for tree removal intensity rates. The output from such models can be used to simulate the evolution of timber harvests at a fine scale, providing a critical input to integrated socioeconomic and biophysical landscape models. This micro-scale modeling approach has potential for broader application in other locations because the key independent variables are readily derived from public databases.