| This study examined how fire risk, a combination of fuels conditions and fire probabilities, varied across a large portion of Missouri, Illinois, and Indiana. Case studies were conducted to evaluate the fuel loading variability in Missouri Ozark forests, determine the temporal variability in fuel accumulation rates, and quantify the role of topographic roughness in fire regimes. Using knowledge gained from these case studies two regional scale studies were conducted describing 1) variability in fuel loading and hazard and, 2) fire probabilities. For the fuel hazard study a stepwise multiple linear regression model (r² = 0.36) predicted litter depth from five parameters: residence time (number of months since leaf fall (November)), topographic roughness index, elevation, precipitation, and slope exposure. Model estimates of litter depth were weighted with litter moisture content to produce maps of litter hazard index. Maps displayed spatial changes in litter hazard for different drought conditions. Overall, litter hazard appeared to be relatively homogeneous throughout the study area with greatest levels attained in southeastern Missouri. Month of year and drought condition are likely the most important parameters concerning fuel hazard. For the fire probability study a large set of fire occurrence records (> 12,000) for the period 1986 to 2008 were used to develop a predictive model of fire probability. CART and logistic regression analysis were used to identify variables related to ignition frequency and to model the spatial variability in fire occurrence probabilities. Eight model parameters were used in a predictive model that showed fire probabilities to be substantially greater in the southern Ozark Highlands compared to the northern Ozarks and most of Illinois and Indiana. Finally, fire probabilities were combined with fuel hazard indices to generate fire risk indices. Fire risk pertained to the risk of forests burning and indices were designed so that increased fire frequency (i.e., fire probabilities) and severities (i.e., fuel hazard) represented increased fire risk. Fire risk indices closely resembled fire probabilities and gave increased weights to both exposed slopes and increased drought conditions. Areas of highest fire risk were identified as being primarily located on Mark Twain National Forest lands. The model appears to have captured much of the variability observed in the modern fire locations, however likely did not characterize the variability associated with known cultural patterns related to fires. |
