Risk Assessment: A Forest Fire Example

David R. Brillinger; Haiganoush K. Preisler; John W. Benoit

doi:10.2307/4356185

Treesearch

Risk Assessment: A Forest Fire Example

Formally Refereed

Authors:

David R. Brillinger, Haiganoush K. Preisler, John W. Benoit

Year:

2003

Type:

Scientific Journal

Station:

Pacific Southwest Research Station

DOI:

https://doi.org/10.2307/4356185

Source:

Statistics and Science: A Festschrift for Terry Speed. 40: 177-196.

Download (PDF 1.73 MB)

Download (PDF 1.73 MB): https://research.fs.usda.gov/download/treesearch/69349.pdf

Abstract

The concern of this paper is obtaining baseline values for the number of forest fires as a function of time and location and other explanatories. A model is developed and applied to a large data set from Federal lands in the state of Oregon. To proceed the data are grouped into small spatial-temporal cells (voxels). Fires are rare so there are many of these voxels with no fires. In fact there are so many such cells that in the analyses presented a sample is taken to make the work manageable. The paper sets down a likelihood for the sampled data and fits a generalized additive model involving location, elevation and day of the year as explanatories.

Keywords

Forest fires, generalized additive model, Oregon, risk analysis, sampled data, wildfires

Citation

Brillinger, David R.; Preisler, Haiganoush K.; Benoit, John W. 2003. Risk Assessment: A Forest Fire Example. Statistics and Science: A Festschrift for Terry Speed. 40: 177-196.