Coastal communities are commonly attributed with lower geographic elevations and are often associated with higher population densities than that of inland communities [1
]. For example, in 2010, over 123 million people, or 39 percent of the American population, lived in coastal counties representing less than 10 percent of the United States (U.S.) land area (excluding Alaska). From 1970 to 2010, an average 39% population increase in coastal counties has been recorded. The average US coastal counties population density of 446 persons per square mile (in 2010) is expected to increase by 37 persons per square mile by 2020, whereas the entire US population density will only increase by 11 persons per square mile [2
]. Unfortunately, most coastal counties are vulnerable to natural disasters, particularly from intense hurricanes. Over the past decade, the U.S. Gulf Coast has been severely affected by multiple hurricanes, including Ivan, Katrina, Rita, Ike and many others. Recently, Superstorm Sandy devastated the New York/New Jersey region. North and South Carolina faced significant economic loss, environmental degradation and public life disruption from hurricane Joaquin. These coastal storms, combined with rising populations, often challenge disaster managers and city planners in protecting public property and human life [2
Public health, both physical and mental, is expected to be impacted not only during an active disaster but also after the disaster [3
]. For example, hurricane Katrina and the Deepwater Horizon oil spill in the Gulf Coast of America have shown chronic mental health effects. However, the extent of the effect was related to the extent of distribution to where people live and work as well as their family structure and social engagement [4
]. In the case of the oil spill disaster, mental health problems persisted even one year after the spill. Anxiety and depression were clinically significant, particularly for people who continued to sustain spill-related income loss [6
]. The literature demonstrated that elderly, female, African Americans, less educated, and non-home owners (or the poor) are more vulnerable and are disproportionately affected by Katrina-induced health problems years after the storm [3
]. It is undoubtedly true that there are a number of contributing factors that are likely to impact coastal communities. For example, in 2005, hurricane Katrina destroyed wetlands, leading to elimination of crucial buffer zones, leaving once protected infrastructure and neighborhoods exposed to more intense future storms. Among these populations, many residents are elderly, a group that is particularly vulnerable to coastal storms and flooding [8
]. For example, nearly 85% of people killed during, and in the immediate aftermath of, Hurricane Katrina were aged 51 and older, and almost half were older than 75 years of age. In addition, climate change is expected to accelerate flood risks in the coming decades as the sea levels rise due to global warming, which will further intensify storm surges [1
]. The extent of the impact from climate change may also depend on local conditions as well as the severity of climate change projections. For example, existing coastal neighborhoods may encounter increased and frequent localized flooding; drainage systems are likely to be overloaded more frequently and severely (as the dated drainage systems are not build for projected climate change), causing backups and street flooding; and people’s mobility may be hindered due to inundation of low-lying feeder roads in coastal areas [10
Proper coastal management to protect populations and community resources from potential flood damage requires a systematic assessment of vulnerability. Measuring the vulnerability of an area or a targeted population has been studied for decades to help community planning and emergency management. Vulnerability, often expressed using a vulnerability index, has been conceptualized in many different ways. When resources become scarce, such as during massive disasters, vulnerability measures are very important for effective allocation of resources to mitigate hazards (e.g., land use practices and building construction practices), improve emergency preparedness practices (e.g., detection and warning systems), help emergency response (e.g., food and medical assistance) and for recovery preparedness practices (e.g., diversified investments and hazard insurance). A number of previous studies have demonstrated various approaches for coastal vulnerability assessment (e.g., [9
]). These assessments are done at multiple spatial (global to local) and temporal (short term to long term) scales. Typically, a vulnerability matrix is developed with available data suitable to a specific problem [9
]. Unfortunately, no single universal metric or measurement tool can be developed or applied to fit all criteria. Developing any such tool is challenging due to ever-present definitional ambiguity, along with the dynamic nature of the temporal and spatial scales of analyses [18
Socioeconomic indicators in a community are most commonly used in measuring social vulnerability. While it is important to characterize population by broad categories of dominant socioeconomic indicators, it is also important to understand how each indicator combines with others to generate interactive vulnerabilities [19
]. This is noted by Cutter, et al
. (2009) [19
] as, “selecting a single variable (e.g., race, gender, or poverty) does not adequately capture communities described as African American female-headed households below the poverty level, because not all African Americans are in poverty; not all female-headed households are African American; and not all people in poverty are females or female-headed households
”. Previous research has shown that some of the commonly used socioeconomic indicators are strongly correlated. Thus, it is important to use either a composite measure of social vulnerability or a subset of these indicators to measure the vulnerability of a community [19
The vulnerability of a community to a flood hazard is commonly measured using socioeconomic indicators or calculating physical flood extents, however, their combined impact is often ignored. Geographical Information System (GIS) based approaches have been used to understand the coastal flood vulnerability by overlaying the hydrodynamic models predicted flood areas over land surface elevations. However, this approach does not incorporate socioeconomic vulnerability [21
]. There are few studies that consider combined socioeconomic and physical vulnerability [25
]. The vulnerability assessment is often complex, requiring significant amounts of data, such as surface elevation surveys and development of detail hydrodynamic models, which are expensive.
In this paper, we proposed a simple approach that combines socioeconomic indicators as well as physical flood extents in measuring the combined vulnerability of an area. Our approach of vulnerability measurement is demonstrated for coastal counties of Mississippi as a case study. To date, no comprehensive studies have been carried out to explore the impact of flooding on infrastructure and public health for these three coastal counties in Mississippi where thousands of vulnerable populations are exposed to periodic and intense storm flooding. Preparing for the broad range of anticipated effects of coastal storms and floods may help reduce the public health burden. Thus, the focus of this paper is to identify critical populations vulnerable to disasters in order to help planners and communities better understand the baseline health status of neighborhoods. While the vulnerability of Mississippi’s coastal community to current flooding conditions are presented in this paper, future research needs to explore the vulnerability of public health from projected climate change (e.g., sea level rise) and anticipated future flood scenarios.
As described in the methodology, an aggregate value of vulnerability for each census block has been calculated as an average of standardized indicators values normalized between zero and one. A final vulnerability score for each tract was calculated as the average of the two vulnerability values (socioeconomic and climatological). A thematic vulnerability map illustrating vulnerability of census tracts is shown in Figure 3
. The interpretation of the thematic map is that, during a flood event, red or dark brown-colored census tracts are at higher risk than the tracts with light colors. In this particular case, there are seven census tracts found to be very highly vulnerable (Z-score ≥ 1.5), whereas 20 tracts are found to be highly vulnerable (0.5 ≤ Z-score < 1.5). The majority of tracts (32) are in the intermediate vulnerability group (−0.5 ≤ Z-score > 0.5), 16 tracts are in low vulnerability group (–1.5 ≤ Z-score < –0.5) and 16 tracts are in very low vulnerability group (Z-score < −1.5).
Inland flooding caused by Pearl River, which drains to Gulf of Mexico, combined with higher population apparently cause the higher vulnerability of tracts in Jackson County relative to the other tracts along the Mississippi Coast. Although the most coastal tracts in Harrison County have considerable flooding, the overall combined vulnerability of these tracts is primarily low as the people in those tracts are mostly white, in the age group of 18 to 65, live in permanent structures and are educated. The majority of tracts in Hancock County are also highly vulnerable not only because of their lower elevation but also the population in the county is primarily under-educated and are low income people, who cannot afford to live in permanent housing or own a vehicle.
The methodology presented in this paper for identifying vulnerable groups is very flexible. For example, the census tracts can be divided into four levels of vulnerability instead of the five presented in Figure 3
. The vulnerability intervals can also be determined as equal quadruples of Z-scores or by other user specified criteria, which might produce varying vulnerability levels. For example, for the vulnerability assessment of the three coastal communities presented in this paper, if the vulnerability groupings were reassigned with Z-score ≥ 1.0 as very high vulnerability, 0.25 ≤ Z-score < 1.0 as high vulnerability, −0.25 ≤ Z-score < 0.25 as medium vulnerability, −1.0 ≤ Z-score > −0.25 as low vulnerability, and Z-score < –1.0 as very low vulnerability, the resulting tracks in each group would be 13, 20, 16, 19 and 13, respectively.
The vulnerability of a community exposed to a flood hazard determined based on a single indicator might be different than the actual vulnerability that is influenced by other factors. For example, the extent of disaster exposure may not be the only determining factor to assess the vulnerability of people in a community. Interaction of socioeconomic factors and the exposure to flood hazard are expected to impact the overall vulnerability. For example, vulnerability of census tracts taking only one indicator at time is illustrated in Figure 4
. As a note, vulnerability of census tracts indicated in Figure 4
change from dark brown shaded census tracts being the highest vulnerability to pale yellow shaded tracts with lowest vulnerability. The 100-year flood exposure as fraction of census tracts flooded (flood vulnerability) is presented in Figure 4
a. Based on flood exposure as the only vulnerability indicator, most of the census tracts in Harrison County are relatively highly vulnerable, which is different than the combined vulnerability of socioeconomic and flood vulnerability shown in Figure 3
. Similarly, normalized non-white population distribution among census tracts is shown in Figure 4
b, indicating the census tracts in Harrison County are found to have medium to high level of vulnerability. However, the combined vulnerability presented in Figure 3
indicated primarily medium to low vulnerability in Harrison County. Similarly, vulnerabilities of other socioeconomic indicators are presented in Figure 4
c–l. How individual variables indicate vulnerability of a community is discussed Section 2.2.1
The information presented in Figure 4
is not only useful for insight into the drivers of overall vulnerability of an area but is also useful to planners, emergency responders and others involved throughout a disaster cycle to target their actions if a specific indicator is of their interest. For example, if there is an epidemiological outbreak in which older (senior citizen) people are deemed to be impacted, then the medical emergency team can use the vulnerability map of indicator 65 years and above (Figure 4
f) to prioritize their emergency actions. As another example, during a high wind event, the cluster of mobile home units (Figure 4
h) will be of interest for decision makers and responders.
Post disaster, especially post Katrina, it was found that not all people are affected to same extent during and after a hazard [20
]. For example, even a decade later, debate about the varied effects of Katrina and the failure of planning and response in certain demographical communities of the disaster area is still ongoing. The debate is on a claim that the communities that are primarily populated with minorities and low-income people were more heavily impacted from Katrina and thus these areas should have received priority in disaster response and mitigation efforts. This past experience shows that several socioeconomic factors, such as race and ethnicity, income, living conditions and education level, are some of the important indicators of vulnerability of people to a disaster. In addition, pre-existing health conditions are also expected to determine the extent of health related impacts from a hazard. Communities with pre-existing a health history will be more vulnerable than other communities. In that sense, the combined vulnerability as a result of interaction between socioeconomic, health and flood extent are expected to be different than the vulnerability arising due to any individual indicator.
Historically, studies have demonstrated various approaches for coastal vulnerability assessment. These studies have often determined the vulnerability of a community based on a single socioeconomic indicator or in some cases a combination of socioeconomic indicators. For flood disasters, agencies most commonly assign the vulnerability of communities simply based on projected flooding of communities, ignoring socioeconomic conditions of the communities. There is limited information on community vulnerability assessment that combines socioeconomic and flood vulnerabilities. In this paper, we presented an assessment framework that combines socioeconomic vulnerability with flood vulnerability in determining the overall vulnerability of a community from flood disaster. This framework can be potentially incorporated into the development of more advanced decision support techniques. As a case study, the vulnerability assessment framework presented in the paper is demonstrated through calculation of vulnerability of census tracts in three Mississippi coastal counties and the results are presented as a thematic vulnerability map. Such thematic maps are useful for decision makers as well for the first responders who often make rapid decisions on needed action with limited access to data sources.
The approach used in this paper is based on a set of socioeconomic indicators that were used in contemporary research focusing on vulnerability from coastal floods. Often the socioeconomic indicators are expected to be interdependent with variation in one indicator’s value affecting the other indicator in the group. However, the framework presented in this paper is expected to minimize the bias that might arise from correlated indicators, as the framework includes calculating composite scores of socioeconomic indicators before combining with flood vulnerability score. In addition, the actual selection of indicators to apply for a given vulnerability assessment is expected to depend on many factors, most notably the purpose and scale of the vulnerability assessment and data availability. The socioeconomic indicators considered in this paper for assessing vulnerability for three coastal counties are primarily based on the fact that they are the most commonly used indicators in the literature for socioeconomic vulnerability assessment as well the data for these indicators are readily available from ACS. Users may select indicators that are relevant to their community. Each community has certain characteristics that make some indicators more suitable than others. For example, if a community has few or no mobile homes, then the people living in mobile homes may not be a required indicator in the community vulnerability assessment. Similarly, if the interest of assessment is to understand chemical or toxins exposure vulnerability during a flood disaster, then one need to consider presence of toxic material storage facilities, solid waste storage units, wastewater treatment units and hazardous chemical processing industries and their material storage locations in assessment. Some other socioeconomic and health indicators of interest may include accessibility of roads, availability of healthcare facilities, number and location of assisted living facilities, college student dormitories and workers dormitories, emergency shelters, etc.
The socioeconomic vulnerability calculation approach presented in this paper assumes all indicators are weighted equally in the vulnerability assessment. However, if the user has specific information that confirms that, for their community, some indicators have higher vulnerability than other, then the user can assign appropriate multiplication factors to adjust for higher vulnerability of some indicators over others.
As the framework presented in this paper is based on indicators, the accuracy of combined vulnerability depends on the accuracy of the indicators’ data. It is also important that potential discrepancies in the results be noted. However, as the approach allows for relative vulnerability comparison among the study units, uncertainties are assumed to be equally weighed. In this way, uncertainty is not removed, but is integrated into the assessment. Overall, the successful assessment of vulnerability is predicated on the generation of a comprehensive set of vulnerability metrics that fully and accurately describe the exposure [41
As a continuation of the current research presented in this paper, we plan to explore quantification of vulnerability in public health sector from projected climate change scenarios. Climate change is an area of intense research and has gradually evolved into a multidisciplinary field. In the public health sector, impact from climate change can be influenced by many factors, including the severity as well as other characteristics of storms, such as the exact timing and location of impact and the unique geographic and topographic characteristics of the affected area. The next phase of our study is to augment the systematic approach that we developed here with future climate change scenarios of coastal communities. The focus of our planned research will be to identify public health indicators (such as existing diseases, age, etc.) and health infrastructure in coastal counties to assess the public health vulnerability from projected climate change scenarios.