An Assessment of Social and Physical Vulnerability to Hydroclimate Extremes in Appalachia


 Appalachia is a cultural region in the southern and central Appalachian Mountains that lags behind the nation in several social vulnerability indicators. Climate projections over this region indicate that precipitation variability will increase in both severity and frequency in future decades, suggesting that the occurrence of natural hazards related to hydroclimate extremes will also increase. The objective of this study was to investigate the spatiotemporal patterns of drought and precipitation and determine how trends overlap with vulnerable communities across Appalachia. The study utilized trend analysis through Mann-Kendall calculations and a Social Vulnerability Index, resulting in a bivariate map that displays areas most susceptible to adverse effects from hydroclimate extremes. Results show the southwestern portion of the region as most vulnerable to increased precipitation, and the central-southeast most vulnerable to an increase in drought-precipitation variability. This study is among the first to utilize the boundaries defined by the Appalachian Regional Commission from a climatological perspective, allowing findings to reach audiences outside the scientific community and bring more effective mitigation strategies that span from the local to federal levels.


Introduction
Appalachia is a geographic region in the Appalachian Mountains with distinct cultural and economic characteristics. Appalachia receives ample precipitation throughout the year, with some areas receiving enough to be considered a temperate rainforest (Reinhardt and Smith 2007). Dozens of major rivers originate in the forests of Appalachia, making the region an essential water source for millions east of the Mississippi River. Although the region is considered to be 'water-rich', climate projections suggest that precipitation and drought variability is likely to increase in both severity and frequency in future decades (Carter et  The purpose of this study was to identify vulnerable populations that are most likely to see increases in drought and extreme precipitation. The study utilized the longest available record (1895-2016) of the Palmer Drought Severity Index (PDSI), a commonly used index that compiles precipitation, surface air temperature, soil moisture, and evapotranspiration rates to measure drought severity. We employed Mann-Kendall calculations to examine trends of PDSI values and a Social Vulnerability Index to identify clusters of at-risk communities. A bivariate mapping technique combined social and physical vulnerabilities to display areas that are most susceptible to the adverse effects from hydroclimate extremes. The ndings from this study contribute to the literature on drought in Appalachia and highlight the need for tailored programs that better address the needs of isolated and underrepresented communities. By utilizing the boundaries of the Appalachian Regional Commission (ARC), the authors anticipate that this study will attract the attention of government o cials who work closely with ARC and impact legislation that spans from the local to federal levels.

Background
Vulnerability to natural hazards is inherently geographical, where the degree of loss varies over different social groups, time and space (Cutter, Boruff, and Shirley 2003;Fuchs et al. 2018). There is substantial literature that shows the risk of impact by a natural disaster disproportionately affects disadvantaged populations (e.g., Cutter et al. 2003;Fuchs et al. 2018;Rufat et al. 2015). Indicators that most strongly correlate with heightened vulnerability generally relate to a person's ability to have access to resources and information (SAMHSA 2017). Appalachia has a history of marginalization and extreme poverty as a result of geographical isolation (Yarnell 1998). Although conditions have improved in recent years, the region's struggled past is still evident: Appalachia exceeds the national average in poverty, unemployment, elderly populations and disability rates, and lags behind in educational attainment and having access to internet and phone services (Pollard and Jacobsen 2020a). Rural Appalachia is also more disadvantaged when compared to other parts of rural America (Pollard and Jacobsen 2020b), indicating that these areas may be more susceptible to impacts from natural hazards than other parts of the country.
The most common natural hazards in Appalachia are typically related to precipitation, or lack thereof.
Heavy precipitation may lead to the destruction of crops, ooding, landslides, debris ows, and contaminated drinking water (CCES; NSSL; Trenberth 2008; Wooten et al. 2016). The Appalachian landscape harbors particular features that exacerbate these risks, such as: 1) frequent rainfall, 2) steep mountainous slopes that produce rapid runoff, 3) close proximity to rivers and low-lying areas, and 4) densely populated urban areas that experience increased runoff along impervious surfaces (NSSL; Wooten et al. 2016 The authors selected the Appalachian Regional Commission's boundaries as the study area to ll a literature gap that exists among examining climate impacts across the region. ARC is a federal partnership that was founded in the 1960s, at a time when poverty in Appalachia was the highest in the nation, in an effort to spur economic growth (ARC). Though there have been a number of studies that utilize ARC boundaries, most of these are grounded in the political/social sciences and only examine socioeconomic demographics (e.g., Bradshaw 1992; Isserman and Rephann 1995; Ulack and Raitz 1981).
To our knowledge, there are no studies that combine environmental and socioeconomic variables for the entire region. Given that adverse outcomes from extreme weather are highly correlated with economic status, utilizing ARC boundaries in this study directly aligns with ARC's missions, and will encourage legislative responses that better serve underrepresented communities.

Study Area
The Appalachian Regional Commission de nes Appalachia as spanning across 13 states and 420 counties from northern Mississippi to southern New York. There are ve subregions that will be used throughout this paper: Southern, South-Central, Central, North-Central, and Northern ( Fig. 1).

Data
The Palmer Drought Severity Index (PDSI) was used to measure physical vulnerability to hydroclimate extremes. PDSI compiles precipitation, surface air temperature, soil moisture, and evapotranspiration rates, and is most commonly used to assess long-term drought (Alley 1984 Component Analysis (PCA) was applied, resulting in a total of ve components that met Kaiser's criterion (Kaiser 1960). The results from this process were totaled and mapped across the region using the standard deviation classi cation method to represent overall social vulnerability (Fig. 4). The ve components were also examined individually to explore unique themes of each component (Fig. 5) Hydroclimate trends and social vulnerability maps were combined using a bivariate mapping technique.
This method used quantile classi cations to reveal high, moderate, and low rankings of vulnerability ( Table 5). The map displays the counties that are most susceptible to adverse effects of hydroclimate extremes (Fig. 6). The de-identi ed secondary data used for this analysis was publicly available and therefore, not reviewed by the Institutional Review Board (IRB) for human subject compliance.

Physical Vulnerability
Long-term trends (

Social Vulnerability
Our principal component analysis resulted in ve components that explained a total of 72.5% of variance across social indicators. These components were totaled and mapped in ArcGIS Pro to visualize overall social vulnerability across the region (Fig. 4). This map did not reveal many obvious clusters of vulnerability, though Mississippi had the highest proportion of vulnerable counties than any other state (70.9%). The southern, south-central, and central subregions exceeded 30% when the three highest vulnerability classi cation were compiled (Extreme-High, High, Moderate-High) ( Table 4). The northern and north-central subregions were deemed least vulnerable, as well as counties that surrounded metropolitan areas.

Bivariate Mapping
The bivariate map (Fig. 6) displayed counties that are most vulnerable to both hydroclimate extremes and socioeconomic factors. The map identi ed the central-southwestern portion of the study area as most vulnerable to increased precipitation, especially in Mississippi, Alabama, Tennessee, and western Kentucky. A handful of counties in northern Pennsylvania and New York also showed high increases in precipitation, but generally low social vulnerability. The central-eastern portion of the region (i.e., Ohio, West Virginia, Virginia, and North Carolina) had high socioeconomic vulnerability and low PDSI values, indicating that this area is most susceptible to increases in drought or drought-precipitation variability.

Discussion
Rural communities across Appalachia are increasingly vulnerable to changes in extreme weather. The aim of our study was to identify areas with both high social vulnerability and high physical vulnerability to increases in drought frequency or severe precipitation. To our knowledge, this is among the rst studies to characterize both social and physical vulnerabilities to hydroclimate extremes across the Appalachian region. Results revealed that the southwestern portion of the study area (i.e., Mississippi, Alabama, Tennessee, and western Kentucky) was most vulnerable to increased precipitation, while the centraleastern portion (i.e., Ohio, West Virginia, Virginia, and North Carolina) was most susceptible to increased drought-precipitation variability. Both areas exhibited high socioeconomic vulnerability which will likely exacerbate the impacts from natural hazards related to both hydroclimate extremes. Our social vulnerability index revealed that the southern, south-central, and central subregions contained the highest rates of vulnerability, with Components 1 (rural indicators), 4 (elder population), and 5 (access to phone services and vehicles) playing a strong role in highlighting the vulnerabilities of rural populations. Counties with the highest vulnerability classi cations were consistently identi ed as rural based on the Urban In uence Codes (ERS 2013). This pattern corresponds with past literature that has shown rural areas are typically more vulnerable than their urban counterparts (e.g., Cutter, Boruff, and Shirley 2003).
Findings highlight the need to improve outreach efforts directed to rural communities. Developing education programs that emphasize preparation and mitigation designed to reach even the most isolated populations would greatly increase resiliency in these areas.
Notably, positive and negative trends for the entire 121-year period were rather weak, where tau values on either spectrum did not exceed 0.2. This indicates that trends in either direction are slow-moving, and that a fair amount of climate variability still persists in these areas. In general, we observed a lack of signi cant trends in neutral or negative directions, which may indicate increased variability of both hydroclimate extremes are expected to continue.
Calculating trends in 30-year intervals revealed an interesting pattern of alternating wet and dry periods. However, the strength of these trends were still rather weak, with a maximum tau value 0.303 in Period A. This suggests that Appalachia still exhibits high hydroclimate variability, even at shorter temporal scales.
Drought trends in all four periods also reported much lower rates of statistical signi cance when compared with wetter precipitation trends. Counties showing drying trends contained less than 10% of statistical signi cance in each segment, while the rates of signi cant precipitation trends were consistently much higher, with a minimum rate of 22.5% in Period B and a maximum of 85.2% in Period C. This may be an indication of how drought functions across the region: drought events are inconsistent, whereas precipitation still occurs regularly despite the presence of drought. Wet periods may also be ± increasing in severity, where trends of increased precipitation jump from 70.5% in Period A to 85.2% in Period C.
El Niño patterns may also be responsible for these wet/dry cycles. El Niño oscillations dictate precipitation patterns in the U.S. and are largely responsible for extreme weather events (Meehl et al. 2007). For instance, the 1982-1983 El Niño was the strongest of the century, and produced extremely heavy rain and ooding across the eastern U.S. (Quiroz 1983;Williams 2015). It is likely that this played a key role in explaining why we saw such strong positive trends during Period C.
Results from this study support the ndings from climate projections across the region. In the southeast, precipitation is projected to increase, as well as the frequency of extreme events receiving at least 3 inches (

Limitations & Implications
There were some limitations with this study. Precipitation data from weather stations was omitted due to its complexity and inconsistent records across the region. Future studies could create a clearer climatological picture by incorporating these records across the region. Additionally, examining social vulnerability at the county level over-simpli es the complex distribution of vulnerable communities. For instance, a county that may have overall low vulnerability may have vulnerable populations that are clustered within speci c census-tracts. A more effective study should examine these facets at a ner scale (Simpson and Human 2008). Moreover, drought is a di cult phenomenon to adequately quantify using a single drought index such as PDSI. Future studies should consider using composite indicators of multiple drought indices to more fully evaluate dryness across the hydrological cycle. Nevertheless, a major takeaway for this study is that, while Appalachia's weather still exhibits high variability, precipitation occurs at a more consistent level than drought. Current trends may point to a future of increased precipitation and extreme events in the southwest portion of the study area, while areas with weaker trends may indicate higher drought-precipitation variability.
Despite these limitations, this study also provides several valuable contributions. The study adds to the literature focusing on drought variability in precipitation-abundant ecosystems like Appalachia. Drought is highly complex and di cult to assess, but it is crucial to understand how it functions in such climates. Acknowledging how drought uniquely impacts Appalachia will help water managers better prepare for future challenges surrounding water quality and distribution. Moreover, this study is among the rst to examine the intersection of physical and social vulnerabilities to hydroclimate extremes across the Appalachian region. By visualizing vulnerability across ARC boundaries, our study has the potential to attract the attention from a wide range of audiences to identify and target areas that may be most severely impacted by extreme weather events.

Conclusion
Our study offers a unique perspective examining the intersection of physical and social vulnerabilities to hydroclimate extremes across the Appalachian region. This study is among the rst to consider Appalachia as de ned by the ARC from a climatological perspective. As a federal economic partnership, the Commission's mission is to tailor policies that address Applachian-speci c issues and work towards building more resilient communities. By utilizing these boundaries, our study provides a comprehensive examination on how climate hazards affect Appalachian populations, which is directly relevant to ARC's missions. We anticipate our ndings will attract the attention of government o cials who work closely with ARC and impact legislation that spans from the local to federal levels. The results from this study highlight the need to address rural needs by developing stronger outreach and education programs that will assist residents in isolated areas to prepare for the effects of hydroclimate extremes.  Figure 1 Study area as de ned by the Appalachian Regional Commission with de ned subregions. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Mann-Kendall's tau values mapped using Palmer Drought Severity Index (PDSI) from 1895 to 2016. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.