A Resilience-Based Approach to Human−Flood Interaction Modeling for Decision Support in Biloxi, Mississippi, USA

The coastal city of Biloxi, Mississippi, has been impacted by and recovered from flooding due to hurricanes and tropical storms throughout its history. It experienced near-total destruction during Hurricane Katrina in 2005 and about 25% percent of damaged areas never recovered, leaving vacant lots scattered throughout the city. Stakeholders from the city government are weighing options for the redevelopment of these areas and are considering offering tax incentives for building homes and businesses or converting the areas to green space to improve the flood resilience of the city. The success of either of these strategies will depend on the frequency and severity of storms and where residents choose to rebuild following the storms. The level of damage from flooding is influenced by development, leading to a two-way interaction between humans and their physical environment. We adapted a minimalistic sociohydrologic model of human-flood interactions to Biloxi with the goal of providing decision support to our stakeholders. In previous work, we conducted a quantitative resilience analysis for the Mississippi Coast following Katrina, that covered social, economic, institutional, infrastructural, community-based, and environmental dimensions of resilience. This analysis was retrospective, and found significant correlation between resilience, defined as the ability of areas to avoid losses and recover, and a number of quantitative variables. Our current work transforms this retrospective analysis to consider the dynamic evolution of resilience through time under different decision-making strategies and trajectories of external shocks (floods). Biloxi does not have physical flood protection structures, so we represented protection infrastructure as the stormwater drainage density, accounting for both built drainage systems and infiltration in green spaces. We analyzed GIS-based information on the city’s stormwater system and land cover to determine how drainage has changed through time focusing on the repair of the stormwater system after Katrina. We analyzed census data with additional information from other sources to increase temporal and spatial resolution to determine how the size and spatial distribution of Biloxi’s population has changed through time and evolved with the city’s flood history. We modeled the population size and the distance of its centroid from the coast, using the population data to determine model parameters. To develop and parameterize the socioeconomic aspects of the human-flood interaction model, we drew on our quantitative resilience analysis conducted for the area at the subcounty scale in the 13 years after Katrina. We face validated these model components with stakeholders from the region. Analysis of the post-Katrina drainage infrastructure recovery found that of 195 pipes that were damaged due to storm inundation and repaired, only 6 were upgraded. The upgrades occurred because the existing pipes were constructed with material that was no longer in use. This analysis, along with stakeholder feedback, indicated that the rebuilding of flood protection infrastructure depended more on federal reimbursement policies and availability of materials than on lessons learned from the flood event. However, the city did alter building requirements after Katrina to require an additional foot of freeboard above base flood elevation on new construction, an action that was selected primarily to increase Biloxi’s Community Rating System score. Revisions to the model will consider building elevation as an additional protection structure. The population data showed short-term sensitivity to floods, but the growth of Biloxi was more directly related to external socioeconomic drivers through most of the historical period, except for recovery from Katrina which resulted in a shift of population away from the coast. The response to this high-severity storm may indicate that there is a threshold response of population growth and distribution to flooding. However, stakeholder input indicated that changes to National Flood Insurance Program policy may have also played a role in individual-level decision-making about redevelopment. While the model is still being refined, initial results indicate that full, sustained redevelopment of Biloxi and complete loss of coastal development are both possible regardless of the redevelopment strategy used due to the highly stochastic nature of large hurricanes. However, the conversion of vacant lots to green space increased the likelihood that development in Biloxi could be sustained. Our results indicate that sociohydrologic models need to consider the effect of incentive structures, such as the National Flood Insurance Program and Community Rating System, when characterizing the response of population size and distribution to flooding. Future work will extend the modeling analysis to other Mississippi Gulf Coast communities to evaluate how different post-Katrina development strategies and Community Rating System scores impact the dynamic resilience of coastal communities.