Assessment of Pre- and Post- Katrina Fuel Conditions as a Component of Fire Potential Modeling for Southern Mississippi

concentrated on creating a linear additive GIS model designed to determine fire potential in southeastern United States. Although numerous fire potential models have been developed (Andrews and Queen 2001, Bonazountas et al. 2005, Hernandez-Leal et al. 2006), our model is specifically designed to address distinctive conditions of southern Mississippi. Most of the existing literature on fire modeling in the United States has been oriented toward the western United States. (Pew and Larsen 2001, Whitlock et al. 2003). Although fire potential generally is regarded as lower in the eastern United States, Mississippi has on average 3,760 wildfires each year that require personnel and resources to extinguish. The average number of wildfires was calculated based on 14 years (1991–2004) of historic fire data acquired from the MFC and in the analyzed period the minimum number of fires was 1,847, and the maximum number of fires reached 6,616/year. Review of the literature on fire modeling indicates that fire potential is related to four factors: climate, topography, anthropogenic influences, and vegetation (Burgan et al. 1998, Mistry and Berardi 2005). Climate often is considered the most important factor in fire models. It is also the most dynamic fire influence, affecting fire potential through precipitation, evaporation, wind, and lightning. Topography is an important fire variable, especially in the western United States; however, Zhai et al. (2003) showed that topography was an insignificant fire variable for Mississippi. Anthropogenic factors also play an important role in fire incidence. Humans affect wildfire ignition by altering the vegetative fuel load characteristics and by providing an ignition source (Pye et al. 2003). Altered fuel characteristics such as forest harvests, construction, management practices, and nature of the urban/ wildland interface all affect fire behavior and probability of occurrence (Zhai et al. 2003). Vegetation is a major component in fuel estimations, because fires tend to be more prevalent in some vegetation types than others. In Mississippi, fires occur more often in needle-leaf conifers, predominantly pine (Pinus sp.) and mixed coniferous and broadleaf deciduous stands than in broadleaf deciduous stands alone (Zhai et al. 2003). In addition, sudden changes in environmental conditions and substantial vegetation damage can contribute to rapid changes in fuel loads and increase in fire potential. The term “fire potential” refers to the final model and for this study is defined as the likelihood or probability that a given part of the landscape is susceptible to fire should an ignition source be available. The final fire potential model will include variables describing climate, ignition, and fuels and will incorporate Light Detection and Ranging, Moderate Resolution Spectroradiometer, and other remotely sensed data as modeling components. However, the focus of this study is to evaluate the fuel component of the fire potential model by assessing changes in forest fuel conditions from preand post-Katrina aerial imagery. In fire terminology such assessment refers to a fire hazard and indicates the state of the fuel, independent of weather or the environs in which the fuel is found (Hardy 2005). Development of GIS layers that enable rapid characterization of changes in forest fuel conditions is important for determining how fire hazard can change due to hurricane impacts. These changes may be especially important in light of the low rainfall amounts before and after Katrina. The drastic and rapid change in vegetation conditions after Hurricane Katrina constituted an immediate need to estimate the amount of timber damage as well as to assess implications of damage in determining fire potential. Preand post-Katrina fire hazard was compared for six counties in southern Mississippi using information on forest age classes, forest type, and damage categories. Our study incorporates a rapid-response sampling design with moderate resolution aerial imagery to estimate accurately the extent of timber damage. Characterization of the spatial distribution of timber damage enabled assessment of fuel conditions important for activities aimed at preventing additional damage to the state’s forest resources because of the increased likelihood of wild-