A synthesis of literature on evaluation of models for policy applications, with implications for forest carbon accounting

Forest modeling has moved beyond the realm of scientific discovery into the policy arena. The example that motivates this review is the application of models for forest carbon accounting. As negotiations determine the terms under which forest carbon will be accounted, reported, and potentially traded, guidelines and standards are being developed to ensure consistency, accuracy, transparency and verifiability. To date, these guidelines have focused on definitions, data, and reporting, not models. The goal of this paper is to synthesize literature that may inform the development of guidelines for the application of models in areas with policy implications, such as forest carbon accounting. We discuss validation, verification, and evaluation as applied to modeling, and review common components of model evaluation. Peer review, quantitative analysis of model results, and sensitivity analysis are the most widely used approaches to model evaluation. US judicial and legislative perspectives on criteria for model acceptability are summarized.