Weather and Soil Moisture Impacts on Oklahoma Wildfires from 2000 to 2012

Oklahoma wildfire data from 2000 to 2012 were obtained from the Office of the State Fire Marshal and analyzed with respect to weather and soil moisture data. At the time of this abstract, 1743 wildfires had been analyzed; by the time of the October conference, a total of 39,370 fires will have been included in the analyses. Fires were grouped according to growing (May through October) or dormant (November through April) season as well as six fire size classes (ranging from <=1 acre to >= 10,000 acres burned); in addition, for one of the analyses, fires were also grouped by month. Weather and soil moisture data from the Oklahoma Mesonet, the state's automated weather station network of 120 sites, were used in the studies as well as some drought indices such as the Palmer Drought Severity and Z indices. Two major analyses were conducted, the first relating the size class of individual fires with weather and soil moisture data from the nearest Mesonet site locations on the start day of each fire, and the second looking at the impact of statewide weather and soil moisture conditions on numbers of fires and acres burned over monthly and seasonal periods.

The first analysis looked at individual wildfires and the weather and soil moisture data from the start day of each fire (using the nearest Mesonet station to the fire location). Variables included were minimum relative humidity (RH), average wind speed, plant available water (40-cm soil column), and KBDI. Wildfires were grouped into six size classes as well as into dormant or growing season. For each weather/soil variable, a regression for each of the two seasons was performed between the average value of the variable for a given fire size class and the fire size class (1 through 6). For both dormant and growing season fires, the best correlations with fire size occurred with minimum RH. Wind speed was strongly correlated with fire size as well. Plant available water (PAW) and KBDI were better correlated with fire size during the growing season, which makes physical sense since less plant available water translates into lower live fuel moisture and live fuel loads, which lead to an intensification of fire behavior during the growing season.

The second analysis considered all wildfires >= 100 acres and looked at the influence of statewide weather and soil moisture conditions upon numbers of fires and fire size (acres burned). As far as time periods studied, in contrast to the first analysis which focused on conditions during the start day of each fire, fires were grouped by month and by season (dormant or growing). Average statewide weather and soil moisture conditions during the period in question (monthly or seasonal) were used in the correlation. In addition, statewide time-lag weather and soil moisture conditions during each of the preceding five months were included in the analysis (in the case of monthly data) and during each of the five preceding six-month periods (in the case of seasonal data). Weather and soil-related data included maximum temperature, dew point, minimum relative humidity, wind speed, precipitation amount, Palmer Drought Severity Index, Z Index, KBDI, and plant available water.

While the results from the second analysis are too many to report, some of the more interesting findings can be mentioned (in this abstract only the monthly time analysis will be covered). During the growing season, average minimum RH during the month of the fires had the greatest correlations with both numbers of fires and acres burned (r = -0.77 and -0.67, respectively). Closely following were two soil water variables, PAW and KBDI, with r values of -0.62 and 0.62, respectively, for numbers of fires, and -0.55 and 0.52, respectively, for acres burned. This shows the importance of soil moisture conditions during the growing season, as noted earlier. Interestingly, precipitation amount had much lower correlations (-0.38 and -0.33, respectively), indicating that soil moisture is more important than precipitation amount. Average maximum temperature was fairly significant, with correlations of 0.47 and 0.37, respectively. Wind speed was only a minor factor during the growing season (r = 0.03 and 0.08, respectively). During the dormant season, average wind speed during the month of the fires was the top variable, with correlations with numbers of fires and acres burned of 0.46 and 0.38, respectively. Closely following was average minimum RH with r values of -0.44 and -0.37, respectively. Average maximum temperature had corresponding r values of 0.30 and 0.22. In contrast to the growing season, soil moisture variables were less important, with PAW having r values of -0.26 and -0.19 and KBDI, values of 0.29 and 0.19. Despite being lower, these correlations were still higher than precipitation amount, which had r values of -0.06 and -0.09, showing that even during the dormant season, soil moisture is more predictive of wildfire behavior than precipitation. Correlations with data from the monthly time lags for either growing or dormant season fires will not be discussed here.

This presentation will focus on these two analyses, which by October will have the benefit of thousands more fires having been included in the database as well as dead fuel moisture (1-, 10-, 100-, and 1000-hr values from the fire danger model in OK-FIRE, Oklahoma's operational weather-based wildland fire management system). Dead fuel moisture should prove very significant in these studies. We will also discuss some interesting results from the time-lag correlations. One interesting point which can, however, be made at this time is that soil moisture is a much better predictor of fire behavior (numbers of fires and acres burned, in this case) than precipitation amount, both during the dormant and growing seasons, when it is especially important. With soil moisture networks increasing in spatial density across the country, soil moisture data should be considered for inclusion in fire research studies as well as in operational fire danger systems locally (such as OK-FIRE in Oklahoma) and nationally.