Citation

Dugan AJ, Baker WL. Int. J. Wildland Fire 2014; 23(1): 58-68.

Copyright

(Copyright © 2014, International Association of Wildland Fire, Fire Research Institute, Publisher CSIRO Publishing)

DOI

10.1071/WF12162

PMID

unavailable

Abstract

Accuracy of small-area, fire-interval estimation methods has been inadequately assessed, thus we conducted modern calibration and historical testing of the traditional composite-fire-interval and a newer all-tree-fire-interval method for estimating population mean fire intervals. We tested in eight areas, at four scales, using 30 small plots across ponderosa pine forests on the South Rim of Grand Canyon National Park. In modern calibration, individual-plot all-tree-fire-intervals were equal to population mean fire intervals in all plots. Across the eight areas, a mean-plot version of the all-tree-fire-interval method never failed, whereas mean-plot versions of composite-fire-intervals failed in 37.5-100% of areas. Pooled composite-fire-intervals, the traditional method, failed in all subareas. In historical testing, pooled and mean-plot all-tree-fire-interval methods and two variations of a mean-plot composite-fire-interval method had the lowest mean relative errors. Again, pooled composite-fire-intervals performed poorly across the eight areas. Overall, in modern and historical tests, the mean-plot all-tree-fire-interval method outperformed all others, but highly filtered mean-plot composite-fire-intervals were fairly accurate in historical tests. Both could be reliable methods, if replicated in small plots averaged over 600-1000-ha landscapes, but for small areas, the all-tree-fire-interval method outperformed others. However, for general use, there may be more value in spatially explicit, landscape-scale methods, rather than any small-area method.