Key points are not available for this paper at this time.
Abstract Although forest managers have used the National Fire-Danger Rating System (NFDRS) since 1972, neither that system nor the 1978 revision to that system were validated in the Northeastern United States. We compiled a data base consisting of 2,558 wildfires from seven Northeastern locations. Using weighted binomial regression, we examined the capability of the NFDRS to predict four measures of wildfire occurrence: probability of a fire day (i.e., a 24-hour day with at least one reported fire), number of fires per fire day, number of fires per day, and probability of a large-fire day. As a control, we also tested the predictive capability of two meteorological elements and five indices from three other fire-danger rating systems. The ignition component of the NFDRS and the Fosberg Fire Weather Index (FFWI) were the best predictors of the probability of a fire day. The NFDRS spread component, the initial spread index from the Canadian Fire Weather Index, and the FFWI were the best predictors of the number of fires per day. Significantly poorer results were obtained for the number of fires per fire day and the probability of a large-fire day. Forest Sci. 29:679-696.
Haines et al. (1983) studied this question.