Back to the future: assessing accuracy and sensitivity of a forest growth model

Abstract

The Forest Vegetation Simulator (FVS) is a widely used computer model that projects forest growth and predicts the effects of disturbances such as fire, insects, harvests, or disease. Land managers often use these projections to decide among silvicultural options and estimate the potential effects of these options on forest conditions. Despite FVS's popularity, the long-term accuracy of the model's projections has rarely been tested, nor has the sensitivity of model output to input values been rigorously assessed.

Susan Hummel, a research forester with the Pacific Northwest Research Station, revisited a site in southern Washington that had burned after she first sampled it. By using her original data as input values, she activated the Fire and Fuels Extension of FVS to simulate forest growth and assessed how accurately the model could predict present-day, postfire conditions.

Hummel and her collaborators also tested the sensitivity of model projections when using actual weather data as inputs compared to the model defaults. The model best captured total surface fuels that remained after the wildfire when actual wind speed and fuel moisture values from the fire were used. Projections of dead trees and amounts of surface fuels were strongly influenced by the source of fire weather data, whereas the projections were not strongly influenced by assumptions about forest health and disease.

Citation

Hummel, Susan; Meznarich, Paul. 2014. Back to the future: assessing accuracy and sensitivity of a forest growth model. Science Findings 165. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 6 p.