How Well Are Cities In Georgia Equipped To Prepare For And Withstand Hurricanes?
Ryan Travis Nation, Sujin Lee, and Katherine Losada
2024-12-02
Introduction
Hurricanes are among the most devastating natural disasters, posing significant threats to communities through their physical, social, and economic impacts. As climate change intensifies, the frequency and intensity of hurricanes are projected to rise, placing greater strain on urban and coastal infrastructures. Understanding an area’s vulnerability to hurricanes is critical for improving preparedness, minimizing damage, and enhancing community resilience. This research seeks to explore the vulnerability of Atlanta and Savannah, Georgia–two cities with distinct geographic and social contexts, representing an inland and coastal city respectively. Specifically, we aim to assess whether these cities are equipped to withstand the increasing threat of hurricanes.
To investigate this issue, we focus on three hurricanes that have impacted Georgia: Hurricane Matthew (Category 5, 2016), Hurricane Irma (Category 5, 2017), and Hurricane Helene (Category 4, 2024). These storms provide a lens through which we can analyze past responses and preparedness in both inland (Atlanta) and coastal (Savannah) areas. We examine physical impacts such as peak flood levels, social impacts including changes in the social vulnerability index (SVI), and community preparedness based on public perceptions drawn from social media data. By synthesizing past data and incorporating insights from the literature on disaster resilience, this study seeks to answer critical questions about the adequacy of current infrastructure and community readiness in the face of intensifying hurricane events. Ultimately, this research will contribute to the broader discourse on climate adaptation and disaster mitigation strategies, offering insights in physical impact, social vulnerability, and community preparedness.
Building on these objectives, prior research has examined various aspects of urban resilience and climate vulnerability. One study assessed urban resilience using a multidimensional index system, evaluating economic, social, institutional, ecological, and infrastructure dimensions in prefecture-level cities (Zhao et al., 2022). Vulnerability and risk assessments have also been conducted for climate change across nine sectors in Atlanta (Morsch, 2010). Additionally, Georgia Tech analyzed sustainability and climate vulnerability in Downtown Atlanta through spatial data indices encompassing economic, social, and environmental dimensions (Georgia Tech, 2019). Furthermore, a study highlighted the role of economic development in influencing investments in climate-related infrastructure, emphasizing the importance of various indices in evaluating an area’s capacity to adapt to climate change (Applegate and Tilt, 2023). These studies collectively offer valuable insights into how cities can evaluate their vulnerabilities and better understand their capacity for resilience against climate threats.
Key questions we aim to address:
How high have flood levels risen in Atlanta and Savannah, GA?
How significant have our studied hurricanes impacted the SVI index in Atlanta and Savannah?
Based on social media data, do individuals in Georgia feel equipped to handle hurricanes?
As hurricanes worsen due to climate change, can Georgia’s current infrastructure withstand them?
Data
To map the flooding levels of each hurricane in both Atlanta and Savannah, elevation and waterways data were sourced from the United States Geological Survey (USGS) in the form of TIFF files. Rainfall and flood data were obtained from National Weather Service reports.
The ATSDR Social Vulnerability Index is a widely used tool for assessing the vulnerability of communities in the context of disasters. This index uses 16 U.S. Census variables from the 5-year American Community Survey (ACS). The SVI considers factors such as socioeconomic status, household composition, disability, minority status, language barriers, housing type, and transportation. The data was fetched from a csv file in RStudio.
For example, this is the 2018 data for Georgia per each census tract after data cleaning:
## Warning in instance$preRenderHook(instance): It seems your data is too big for
## client-side DataTables. You may consider server-side processing:
## https://rstudio.github.io/DT/server.htmlThe community perceptions of the three hurricanes were extracted from social media data, specifically Reddit threads. The text from relevant subreddit threads were extracted using the RedditExtractoR package in R.
When assessing users’ perceptions of each hurricane, the data was fetched from all relevant threads based on specific keywords and subreddits. For example, this is the initial data frame for Hurricane Helene:
After data processing, the final data frame includes variables such as sentences extracted from subreddit threads and their sentiment score:
Methods
Physical Impact
We assessed flood levels by incorporating elevation and waterways data. By overlaying flood levels on the maps, we identified areas of Atlanta and Savannah impacted by rainfall from hurricanes. The flooding levels were then visualized using leaflet in R, where the bins were manually set to display varying flood intensities on the maps for each city and hurricane. Lastly, we compared these impacted areas with changes in the Social Vulnerability Index to analyze potential correlations between flood risk and socio-economic vulnerabilities.
Community Perceptions
To analyze perceptions of community preparedness and the aftermath of hurricanes that significantly impacted Georgia (Matthew, Irma, and Helene), Reddit data was mined and processed using text analysis techniques. Data was scraped from relevant subreddit threads using the RedditExtractoR and WordCloud2 packages in R. Subreddits such as “news”, “TropicalWeather”, “HurricaneHelene”, and “hurricane” were selected based on their relevance, with keywords like “hurricane helene” and “georgia” guiding data collection. The objective was to assess whether Reddit discussions expressed positive, negative, or neutral sentiments about these hurricanes’ effects on Georgia. Before conducting sentiment analysis, the text data underwent preprocessing (cleaning and tokenization) with stop words removed to enhance analysis accuracy. Sentiment analysis was performed at both the sentence and word levels using the sentimentr and syuzhet packages, respectively. Additionally, word clouds were generated to visually represent key themes and topics within the discussions.
Results
Physical Impact
The first hurricanes caused significant flooding in both cities. Savannah saw up to 15 feet in flooding, whereas Atlanta had increased water levels by five feet. The second hurricane caused Savannah to go under 12 feet of water. Lastly, Hurricane Helene did not impact Savannah as much, but caused significant flooding and damage in the Buckhead area, as well as more rural areas outside the scope of our investigation. We then compared the impacted areas with SVI levels.
Social Vulnerability
Atlanta: The analysis revealed that Hurricane Irma (2016–2018) had a more significant impact on SVI compared to Hurricane Matthew (2014–2016). The SVI changes observed between 2016 and 2018 were notably larger, with the range of SVI changes increasing from 0 to 68.6 in the 2014–2016 period to 0 to 120 in 2016–2018. This indicates that social vulnerability in Atlanta worsened more substantially during the later period, corresponding to the effects of Hurricane Irma.
Areas with consecutive reductions in SVI were identified, particularly those with reductions greater than the average change. These areas showed increases in all four components of the SVI, with Housing Type and Transportation, as well as Socioeconomic Status, experiencing the most significant rises.
Savannah: In Savannah, the SVI change was more pronounced between 2014 and 2016, following Hurricane Matthew, compared to the period between 2016 and 2018. While some areas showed consecutive reductions in SVI, the overall trend indicated that all four SVI components increased during both periods. Housing Type and Transportation were particularly affected, with these areas experiencing the most notable increases.
While the SVI change from 2014 to 2018 was evident, the difference was not statistically significant, suggesting that Savannah may have been more resilient in its recovery from hurricane impacts compared to Atlanta.
Comparison of Cities: The findings indicate distinct patterns of vulnerability changes between Atlanta and Savannah. In Atlanta, the impact of Hurricane Irma was more severe, resulting in higher SVI changes between 2016 and 2018. In contrast, Savannah experienced more significant changes in SVI following Hurricane Matthew, with the 2014–2016 period showing greater vulnerability shifts than the 2016–2018 period.
Although it is difficult to definitively link these changes to the hurricanes alone, the SVI shifts suggest a correlation between the events and increased vulnerability in certain areas, particularly in Atlanta following Hurricane Irma. Further analysis would be necessary to better understand the direct connection between the hurricanes and social vulnerability changes.
Community Perceptions
The word clouds (Figure #) and word-level sentiment analysis (Figure #) for Hurricanes Matthew, Irma, and Helene highlighted emotions such as negativity, positivity, trust, and fear in Reddit threads discussing their impact on Georgia. Among the three hurricanes, Helene, the most recent and comparatively weaker in category, displayed the highest levels of positive emotions. To enhance accuracy, sentence-level sentiment analysis was also performed; however, the predominance of neutral classifications heavily skewed the results. Using the sentimentr package to account for negations yielded similar mean sentiment values close to zero. Despite this, Helene demonstrated the most positive sentiment overall, as indicated by its highest maximum and minimum sentiment scores (Figure #).
Conclusion
In conclusion, disaster preparedness varies across the state of Georgia. While situated on the coast, Savannah has the infrastructure to handle flooding and dissipate the water. On the other hand, Atlanta does not, which is why we saw a significant change in SVI levels. However, the social media data tended to support the opposite notion. In order to further this research, we would need to analyze more cities in Georgia, with varying intensities of hurricanes. Additionally, we would need more social media or news data to assess social perceptions of hurricanes.
Citations
Analysis to Contextualize the Criteria for Urban Resiliency Planning from International and US Cities Perspectives.” Journal of Urban Resilience Planning, 2023.
Applegate, Joshua D., and Jenna H Tilt. Frontiers in Environmental Science. “Vulnerability Analysis for Coastal and Urban Areas.” Frontiers in Environmental Science, www.frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2023.1103115/full. Accessed 19 Nov. 2024.
Centers for Disease Control and Prevention. “Emergency Preparedness and Response View.” CDC Environmental Health Tracking, www.cdc.gov/environmental-health-tracking/php/communications-resources/emergency-preparedness-response-view.html. Accessed 19 Nov. 2024.
Centers for Disease Control and Prevention. “Preparedness and Response Data and Research.” CDC Environmental Health Tracking, www.cdc.gov/environmental-health-tracking/php/data-research/preparedness-response.html. Accessed 19 Nov. 2024.
Duke University. Hurricane Impact on Urban Vulnerability. DukeSpace Repository, dukespace.lib.duke.edu/server/api/core/bitstreams/e0c72369-4769-4d90-9019-b1eead634db2/content. Accessed 19 Nov. 2024.
Federal Emergency Management Agency. “OpenFEMA Data Sets.” FEMA, www.fema.gov/about/openfema/data-sets. Accessed 19 Nov. 2024.
Georgia Institute of Technology. Spatial Analysis of Sustainability and Climate Vulnerability in Downtown Atlanta. Georgia Tech, 2019. Morsch, Amy. A Climate Change Vulnerability and Risk Assessment for the City of Atlanta, Georgia. 2010.
National Hurricane Center. “Data Archives.” NOAA, www.nhc.noaa.gov/data/. Accessed 19 Nov. 2024.
NOAA Office for Coastal Management. “Coastal Flood Exposure Mapper.” NOAA, coast.noaa.gov/floodexposure/#-9289003,3852899,8z/eyJoIjoic3Rvcm1TdXJnZXwxfDUiLCJ0IjoiZW1wbG95ZWVzfDEifQ==. Accessed 19 Nov. 2024.
University of Connecticut. An Evaluation of Emergency Preparedness in Urban Areas. DigitalCommons@UConn, digitalcommons.lib.uconn.edu/cgi/viewcontent.cgi?article=1086&context=uchcgs_masters. Accessed 19 Nov. 2024.
Zhao, Xiao, et al. “The Evaluation and Obstacle Analysis of Urban Resilience from the Multidimensional Perspective in Chinese Cities.” Urban Resilience Studies, 2022.
Social Vulnerability
To evaluate the social impact of disasters, we used the Social Vulnerability Index (SVI), which is a widely used tool for determining the social vulnerability of counties and tracts. To assess the impact of disasters, we analyzed the change ratio of SVI before and after each event. Data from the Agency for Toxic Substances and Disease Registry (ATSDR), covering the years 2014 to 2018 and collected every two years, was used to calculate changes over two-year intervals.
We specifically analyzed the changes in SVI for three hurricanes: Hurricane Matthew (2016), Hurricane Irma (2017), and Hurricane Helene (2024), focusing on the counties and tracts in Atlanta and Savannah. For Hurricane Matthew, the change in SVI was assessed by comparing data from 2014 to 2016. The impact of Hurricane Irma was evaluated by comparing data from 2016 to 2018. Hurricane Helene was not analyzed due to the unavailability of data for 2024.
The SVI change was calculated by comparing the scores before and after each hurricane event. We also conducted a t-test to evaluate the statistical significance of the changes in SVI from 2014 to 2018. This allowed us to assess whether the observed changes in vulnerability were substantial.