Fire Size

The following predictors were found to have significant effects (P-value <<0.05) on the multinomial probabilities of a fire reaching one of three size classes: spatial location; size of beetle infestations in previous 3 years; size of spruce budworm infestations 4 to 6 year ago; average monthly temperature; and average monthly PDSI.

Fig. 1 - Probability of fire

The estimated spatial pattern of fire sizes (Figure 1) seems to indicate northern Washington and eastern Oregon as some of the regions with the highest probabilities of a fire getting large after controlling for all other predictors in the model. Again, these results appear to be consistent with observed changes over the last 60–100 years in fire frequency and size in interior forests of the Pacific Northwest (Hessburg and Agee 2003).

Fig. 2 - Odds of a fire

According to the present data, beetle infestations that are 3 to 6 years old do not seem to be significantly associated with fire size; however, the odds of a fire getting large seem to increase significantly when the total size of beetle infestation in the last 3 years is greater than 5000 trees killed per km² (Figure 2, top panels). Canopies with residual dead needles and limbs in recently killed trees may contribute to crown conditions that significantly influence fire behavior. Following a simulated outbreak, Ager and others (in press) found that beetle-caused tree mortality increased surface fuel loadings and the potential for stands to sustain crown fire in the simulation. A different association was seen between fire size and size of spruce budworm infestation. No significant association was apparent between recent defoliations (area defoliated in previous 3 years) and sizes of fires nearby. However, a significant decrease was seen in the odds of a fire getting large as the size of older defoliations (total area defoliated 3 to 6 years ago) increased from a total of 100 ha to 300 ha (Figure 2, bottom panels). Lack of foliage following defoliation 4 to 6 years prior may actually reduce the risk of crown fire. Although the decrease in odds appears to continue as the size of defoliated area increases past 300 ha, the standard error for this range of defoliation is too large to warrant a precise conclusion.