The Influence of Mountain Pine Beetle Damage on the 1988 Yellowstone Fires Heather J. Lynch Paul R. Moorcroft
Remote sensing of fire & insect damage mountain pine beetle damage
covariates regression coefficients to be estimated Spatial Logistic Regression Model: What is the relationship between previous insect damage and the probability of burning in 1988? burn/no burn weather geographic factors insect damage latitude longitude
mean 95% confidence interval Palmer Drought Severity Index A more negative PDSI is associated with more severe drought. Therefore, as we might expect, more severe drought is associated with a higher probability of burning. ˆ (estimated model coefficients) Results after model selection, 3 variables remained (=autocorr)
mean 95% confidence interval Results (cont.) ˆ (estimated model coefficients) northeast- facing aspect northeast facing slopes are more likely to burn than southwest facing slopes.
mean 95% confidence interval Results (cont.) ˆ (estimated model coefficients) repeated mountain pine beetle damage Repeated mountain pine beetle damage of any intensity during is associated with an increased risk of burning in the 1988 fires.
mean 95% confidence interval Results (cont.) ˆ (estimated model coefficients) autocorrelation coefficient The coefficient associated with the strength of auto-correlation is most dominant.
non-spatial model Left: burn prob. (red = burned)Right: mis-classified pixels (pink)
auto-correlation only Left: burn prob. (red = burned)Right: mis-classified pixels (pink)
best-fit model with a random border: 61.0% of all pixels correctly predicted Left: burn prob. (red = burned)Right: mis-classified pixels (pink)
best-fit model with a known border: 87.3% of all pixels correctly predicted Left: burn prob. (red = burned)Right: mis-classified pixels (pink)
mountain pine beetle activity significantly affects the spatial patterning of forest fires, but only after a time lag of ~15 years mountain pine beetle activity impacts fire risk primarily through a change in stand structure and not as a direct result of increased fuel loading Conclusions Publications: Lynch, H.J. RA Renkin, R.L. Crabtree & P.R. Moorcroft (2006). The Influence of Previous Mountain Pine Beetle Activity on the 1988 Yellowstone Fires. Ecosystems 9:
HyMap (5.7 m resolution, 126 wavelength bands, airplane mounted) Satellite-derived estimates of forest insect damage non-photosynthetic vegetation in red