|
Modeling the spatial patterns of wildfires ignitions in Southern Californian Mediterranean ecosystems.
Nicolas
R.
Faivre, UC Irvine, Dept. of Earth System Science, nfaivre@uci.edu
(Presenter)
Yufang
Jin, University of California, yufang@uci.edu
Michael
L.
Goulden, UC Irvine, mgoulden@uci.edu
James
T.
Randerson, UC Irvine, Dept. of Earth System Science, jranders@uci.edu
Fire occurs as a function of suitable conditions involving an ignition agent and adequate weather and fuel conditions. Modeling attempts of fire occurrence and burned area have provided a good understanding of the physical and climatic factors that constrain or promote fire spread and recurrence. However, information on how humans influence ignitions and burned area patterns is still lacking at a scale compatible with integrated fire management. This research provides new insights into the key physical, climatic, and human drivers associated with fire ignition probability, their interaction and relative importance across southern California. A 30-year exploratory analysis of one-way relationships identified that major determinants of ignitions frequency patterns included the distance to roads and housing, and the slope percentage. We then combined all significant covariates to model ignitions occurrence and frequency within National Park forests using Logistic and Poisson regression approaches. We successfully explained approximately 50% of the spatial variability of ignitions at a 3km resolution using a zero-inflated negative binomial regression. Our results further established that landscape-level fire activity was mostly determined by human factors while biophysical factors had a subsidiary influence. Using the model developed initially for National Forests, we extended the predictions outside the study domain where ignitions are not reported. This study offers spatially explicit information of fire ignitions risk in southern California, which is essential to identify values at risk and implement adequate prevention and suppression fire management strategies.
Presentation:
2013_Poster_Faivre_33_69.pdf (11997k)
Presentation Type: Poster
Session: Poster Session 1-B
(Tue 4:30 PM)
Associated Project(s):
Poster Location ID: 33
|