Seventh Symposium on Fire and Forest Meteorology

1.1

Climate and predictability of Alaskan wildfires

Peter Bieniek, Univ. of Alaska, Fairbanks, AK; and M. Shulski and G. Wendler

Wildfires burn an average of more than 4,000km2 each year in Alaska, but vary widely from year to year and can be several times higher. These fires, started by human and natural causes, can endanger life and property when they approach populated areas; this is why fire management is critical. An important component for the effectiveness of fire management is the ability to predict the start and spread of fires in advance. The purpose of this study is to examine area burned on a seasonal scale in Alaska, and its relationship with atmospheric factors on a composite monthly and seasonal scale to develop a climatology of conditions that control the area burned during the fire season. This may lead to an improved seasonal forecast of the strength of fire seasons one or two months in advance. Mean sea level pressure, surface air temperature, total column precipitable water, 500hPa and 700hPa geopotential height, 700hPa specific humidity and 1000-500hPa layer thickness were obtained from the NCEP/NCAR Reanalysis. Seasonal and monthly composite anomalies for the various atmospheric variables from the reanalysis were determined for seasons with the top and bottom 10% of area burned. These composites show significant differences in placement and sign of anomalies throughout the seasons and months. Years with high area burned show statistically significant positive anomalies in all variables with maxima centered on or near Alaska during summer months. Years with little area burned show a prevalence of negative anomalies compared to the high seasons with a notable anomaly visible in moisture fields over northern Africa. Correlations with the seasonal area burned data set from the Alaska Fire Service and the corresponding monthly averaged atmospheric variables from March through August from the reanalysis reveal significant correlations for multiple variables in different months. Highly significant parameters were then selected as potential predictors for consideration in a multiple regression fitting. Results show a correlation coefficient of 0.78 when applying the March-April time frame for forecasting the area burned the following summer. The correlation coefficient increased to 0.91 when meteorological parameters of the total season (spring and summer) were included. We can conclude from these results that not only is there a relationship between monthly averaged atmospheric variables throughout the spring and summer and area burned in Alaska, but it is also possible to predict the area burned in advance considering only the preseason conditions by using a multiple regression fitting.wrf recording  Recorded presentation

Session 1, Impacts of Weather and Climate on Wildfire
Tuesday, 23 October 2007, 1:30 PM-3:00 PM, The Turrets

Next paper

Browse or search entire meeting

AMS Home Page