WWETAC Projects

Project Title: Evaluation of models used to predict postfire tree mortality

Principal Investigators: David Shaw, Stephen Fitzgerald, and Travis Woolley, Department of Forest Engineering, Resources and Management; Lisa Ganio, Department of Forest Ecosystems and Society, Oregon State University, Corvallis, Oregon

dave.shaw[at]oregonstate.edu

Collaborator: Terry Shaw, emeritus USDA Forest Service, Pacific Northwest Research Station, WWETAC, Prineville, OR

Key Issues/Problems Addressed:

Tools that generate a better understanding of post-fire tree mortality are increasingly important to forest management and planning due to a growing number of large fires occurring throughout the western United States. Multiple models that predict post-fire tree mortality have been developed, yet a comprehensive analysis and synthesis of these available methods is lacking.  Logistic regression models have been used as the basis of simple field tools and have contributed to larger fire effects models.

Setting and Approach:

This study reviewed literature on post-fire mortality prediction methods that use logistic regression models (based on empirical data) for coniferous tree species in the western United States.  Various sources were reviewed, including peer-reviewed journal articles (1986-2006), general technical reports (1980-2007), and 29 journal articles and technical reports [prescribed fire (18), wildfire (10), prescribed fire and wildfire (1), validation of previous models (2)]. The selected material was analyzed to identify and evaluate methods used to develop, evaluate, and interpret logistic regression models; explanatory variables used in logistic regression models; factors influencing scope of inference and model limitations; model validation, and management applications.

Key Findings:

  • More than 100 logistic regression models in 33 published papers have been developed to predict post-fire tree mortality for 19 coniferous species following both wild and prescribed fires.
  • The most widely used explanatory variables in post-fire tree mortality logistic regression models were the measurements of injury to the crown (e.g., crown scorch) and stem (e.g., bole char). Prediction of post-fire tree mortality was improved when crown and stem variables are used together.
  • Post-fire tree mortality prediction will benefit from more consistent definition of terms used to define model variables, model validation, and direct incorporation of physiological responses that link to process modeling efforts.

Impacts/Applications:

A synthesis and comparison of existing post-fire tree mortality prediction models (and related citations) was completed.

Publication:
Woolley, T., L.M. Ganio, D.C. Shaw, and S. Fitzgerald. 2009. A framework to evaluate post-fire tree mortality logistic models. Tall Timbers Research Miscellaneous Publication No. 16

WWETAC Project ID:  FY07TS29

treefire