New Orleans and Flooding

Background

Background

In 2005, Hurricane Katrina caused widespread flooding and became one of the most devastating natural disasters in the history of New Orleans. It exposed how vulnerable New Orleans is to recurring environmental hazards that threaten basic infrastructure, the environment, and communities. Table 1 highlights natural hazards with their annual probability of occurrence in New Orleans, as identified by the Hazard Mitigation Plan of the City of New Orleans1.

Table 1: Probability of natural hazard in New Orleans.

Table 1: Probability of natural hazard in New Orleans.

Based on the data, we see how many high-probability hazards are related to flooding, and conclude that focusing on flooding and its underlying causes, impacts, and mitigation strategies is a crucial research.

Flooding varies in type and severity and is influenced by a combination of natural and anthropogenic factors. Natural variables include precipitation, topography, vegetation, soil texture, and seasonality, which collectively shape the occurrence and impact of flood events. Human activities, referred as anthropogenic factors, further complicate flooding risks. Urbanization and land use changes disrupt natural drainage patterns, so flood-control structures, such as drainage pumps and levees, are constructed to mitigate flooding2.

Fig. 1: Levee systems in New Orleans.

Fig. 1: Levee systems in New Orleans.

Fig. 2: New Orleans groud elevation and levees.

Fig. 2: New Orleans groud elevation and levees.

Fig. 1, sourced from Wikimedia3, and Fig. 2, sourced from the US Army Corps of Engineers4, present the current levee systems in New Orleans and a cross-section of New Orleans with the elevation and the sea level, respectively. Although the coastal levee system has successfully mitigated coastal erosion and flooding from the Mississippi River and Lake Pontchartrain, significant vulnerabilities remain due to other factors and further attention to these flooding-related risks is essential.

Research Goals

Research Goals

The primary objective of this study is to assess the vulnerability of New Orleans, Louisiana, to flooding events, with a specific focus on the following aspects:

  1. Evaluate flood susceptibility: Analyze the geographical, environmental, and infrastructural factors that contribute to flooding in New Orleans.
  2. Assess current anti-flooding systems: Examine the effectiveness of existing flood mitigation systems, particularly the drainage networks and the distribution of grocery stores, to determine their adequacy in handling extreme weather events.
  3. Understand socioeconomic influences: Investigate how socioeconomic factors such as income levels, housing locations, population density, and accessibility influence the flood resilience of New Orleans.
  4. Prepare for future flooding events: Provide actionable insights and recommendations to help residents, policymakers, and urban planners prepare for and mitigate the impacts of future flooding.

Fig. 3 illustrates the location of Orleans Parish within Louisiana. Fig. 4 demonstrates the spatial context for this research. Both figures were captured from the official website of the City of New Orleans5, 6.

Fig. 3: Location of New Orleans.

Fig. 3: Location of New Orleans.

Fig. 4: Boundary and landmass of New Orleans.

Fig. 4: Boundary and landmass of New Orleans.

Logistic Regression Analysis

Logistic regression on factors that impact flooding risk

Data

Flooding zone

National Flood Hazard Layer – A digital database containing the flood hazard mapping information from the Federal Emergency Management Agency (FEMA)7.

  • High Risk Flood Areas have a 1 in 4 chance of flooding during a 30-year mortgage, otherwise known as a 1% annual flood risk. “A” areas are typically located near ponds, streams, and rivers. These features increase the likelihood of flood waters damaging the surrounding area. Homes and properties within these areas are subject to mandatory flood insurance.

  • High Risk Coastal Areas: Coastal areas are often the most hazardous of all flood zone areas.

In New Orleans, about 20.5% of the total area is classified as high risk for flooding, highlighting its vulnerability to flood disasters. The data highlight the need for targeted flood management strategies in these high-risk areas to safeguard infrastructure, minimize economic losses, and protect vulnerable populations.

Elevation and slope

DEM Data – USGS8 2021 Greater New Orleans 1 meter DEM

  • Areas with a low slope and minimal elevation changes are more vulnerable to flooding, as water tends to accumulate in these regions with limited natural drainage. The DEM data identifies these vulnerable areas, providing valuable insights for flood risk assessment and mitigation planning.

Land cover

Land Cover Data – 2017 NOAA9 OCM High Res Land Cover, 1m

  • Soil and vegetation-covered areas can absorb water, reducing the risk of flooding. In contrast, impervious surfaces such as paved roads and bare ground prevent water infiltration, increasing the likelihood of flooding in these areas.
  • We reclassified the land cover data on a scale of 0 to 5, based on the potential water absorption of each surface type.

Canals: Drainage system

Canal Network, with varying widths

  • Canals can help mitigate flooding risks by channeling excess water away from vulnerable areas.
  • Wider canals typically have a greater capacity for drainage and a higher ability to handle large volumes of water during heavy rainfall or storm surges.

Methods


Factors:

  • Elevation
  • Slope
  • Land cover absorption
  • Distance to canals


Steps:

  • Generate 1000 random sample points
    • Points were generated within the main urban areas of the city, focusing on regions with significant flood risk.
    • Approximately 90% of the urban area is covered by low-absorption land types, such as impervious surfaces, which increase the flooding risk. The sampled points ensure a comprehensive spatial analysis of flood-prone areas.

  • Extract multi values to points:
    • Elevation (-5 to 9)
    • Slope (0–74 degrees)
    • Land cover absorption (0–5)
    • Distance to the nearest canal (meters)
    • Flooding index (1 = Flooded, 0 = Not Flooded)
  • Logistic regression
    • Dependent variable: Flooding index
    • Independent variables:
      • Normalized values of elevation, slope, land cover absorption, and distance to the nearest canal. These variables are scaled between 0 and 100 to ensure consistency in the analysis.
    • We performed logistic regression to predict the likelihood of flooding based on the interaction of these variables and aim to identify key factors that contribute to flood vulnerability.

Analysis

Spatial coverage of canals
  • Weighted buffers
    • The weighted buffer approach shows that the canal system covers most of the area within the study region, highlighting its critical role in managing water flow and mitigating flood risks.
    • However, areas with low elevation and minimal slope changes remain at a high risk of flooding, even in proximity to large canals. This suggests that canal coverage alone is insufficient to address flood risks in these zones, and additional drainage infrastructure or flood management strategies are needed.
    • The map visualizes the extent of canal coverage, with varying canal widths represented in the legend. Wider canals are likely to have greater drainage capacity, but their effectiveness may be limited in low-lying areas where water tends to accumulate.

Logistic regression
Dependent Variable: Flooding
Variable Coef. Std. Error p-value Significance
(Intercept) -0.814 0.256 0.001 ***
Factor 1: Elevation -0.030 0.008 0.000 ***
Factor 2: Slope 0.010 0.008 0.206
Factor 3: Land cover absorption 0.001 0.003 0.696
Factor 4: Distance to canals 0.016 0.005 0.001 ***
Predictors of Flooding Risk
  • Elevation and Distance to canal are two significant predictors of flooding risk.
  • Land cover: Limited presence of high-absorption land cover types in the urban area.
  • Slope: Urban environments introduce abrupt changes in slope, such as curbs and building edges, creating dispersed slope values that do not correlate well with natural flooding patterns.
  • Flooding typically occurs in low-elevation, flat areas, where water accumulates due to limited drainage capacity.


Exponent Coefficients
Variable Exp(Coefficient)
Factor 1: Elevation 0.970
Factor 2: Slope 1.010
Factor 3: Land cover absorption 1.001
Factor 4: Distance to canals 1.016
  • Elevation: An odds ratio of 0.97 means a one-unit increase in normalized DEM decreases the odds of flooding by 3%.
  • Distance to canals: An odds ratio of 1.016 means a one-unit increase in distance from the canal increases the odds of flooding by 1.6%.

After logistic regression, we see that elevation and distance to canal are two of the most significant predictors of flooding risk. Areas with lower elevation and greater distances from canals are more prone to flooding due to reduced water flow and drainage capabilities. Furthermore, canals play a crucial role in mitigating flooding by providing pathways for water drainage. However, their effectiveness is often limited in low-elevation zones where drainage systems struggle to cope with water accumulation. Therefore, we recommend expanding the canal network, upgrading existing drainage infrastructure, and integrating green infrastructure such as permeable surfaces and retention basins. These essential strategies can address current flooding challenges.

Precipitation Analysis

Storm water and heavy rainfall are some of the main causes of urban flooding. Therefore, observing precipitation patterns is crucial for a flooding research.

Data

The precipitation data were sourced from National Oceanic and Atmospheric Administration10. The data consist of daily precipitation records from multiple weather stations in New Orleans over the past 10 years (12/1/2014 - 12/1/2024).

Methods

The monthly average precipitation was calculated for individual weather stations (with several obvious outliers/wrong data filtered out), followed by aggregating the values across all stations to derive an overall monthly trend. The data were further visualized to observe patterns and identify wetter periods.

Analysis

Fig. 5 demonstrates the monthly average precipitation by station. We see that while there were some minor deviations, the pattern is consistent across all stations and the reported data were similar. This makes sense because they are all in a small regional area. Hence, we chose to aggregate the data. When aggregated across stations, the monthly precipitation trend (Fig. 6) demonstrated evident seasonal patterns. A noticeable spike in precipitation occurred during the summer months, with July being the month with the highest precipitation on average, indicating the wettest season. You may use the mouse to hover over each data point in this interactive plot to see the calculated average value with the corresponding month in the tooltip.

## # A tibble: 1,074 × 3
## # Groups:   STATION [18]
##    STATION     YearMonth avg_prcp
##    <chr>       <chr>        <dbl>
##  1 US1LAJF0005 2020-05     0.66  
##  2 US1LAJF0005 2020-06     0.377 
##  3 US1LAJF0005 2020-07     0.467 
##  4 US1LAJF0005 2020-08     0.317 
##  5 US1LAJF0005 2020-09     0.14  
##  6 US1LAJF0005 2020-10     0.137 
##  7 US1LAJF0005 2020-11     0.0229
##  8 US1LAJF0005 2020-12     0.116 
##  9 US1LAJF0005 2021-01     0.0872
## 10 US1LAJF0005 2021-02     0.179 
## # ℹ 1,064 more rows

## # A tibble: 121 × 2
##    YearMonth avg_prcp
##    <chr>        <dbl>
##  1 2014-12     0.120 
##  2 2015-01     0.164 
##  3 2015-02     0.0694
##  4 2015-03     0.162 
##  5 2015-04     0.416 
##  6 2015-05     0.184 
##  7 2015-06     0.117 
##  8 2015-07     0.180 
##  9 2015-08     0.0930
## 10 2015-09     0.156 
## # ℹ 111 more rows
## # A tibble: 3 × 2
##   Month avg_prcp
##   <dbl>    <dbl>
## 1     7    0.290
## 2     6    0.260
## 3     8    0.256

Soil Texture Analysis

Certain soil types are more vulnerable to flooding, especially flooding caused by heavy rainfall. Therefore, we also analyzed the soil texture.

Data

The soil data were sourced from ArcGIS Online USA Soils Map Units11 to examine the soil types and characteristics in New Orleans. The data included shapefiles for all major soil types along with their spatial distribution across the region.

Methods

Since the shapefile of the soil data had a high resolution, and R was not adept at displaying that amount of data in the map, the data were processed and visualized using ArcGIS Pro to identify the distribution of specific soil types in New Orleans. The original data were clipped to the extent of the New Orleans boundary. By referring to past studies on soil type characteristics12, 13, 14, we used different colors to represent the water absorption of various soil types, ranging from dark green for the least capable to light green for the most capable.

Analysis

The spatial map shows that New Orleans is dominated by four soil types, namely histosols, vertisols, inceptisols and entisols. During the literature review, we see that histosols is the most flood-prone type of soil due to its high organic content and poor drainage capabilities, which would lead to surface flooding. Vertisols, which is rich in clay and exhibits low infiltration rates, prevents effective water absorption during heavy rainfall events. Inceptisols and entisols are primarily located in floodplain regions, which would naturally accumulate water and further amplify flooding risks. These demonstrate that the soil composition in New Orleans significantly contributes to its flood vulnerability and all regions in the city are highly susceptible to flooding during heavy rainfall.

Fig. 7: Soil type.

Fig. 7: Soil type.

Census Demographics Analysis

Socioeconomic and human influences are also vital aspects that should be considered. Therefore, we also analyzed the population and socioeconomic data to further understand how New Orleans is vulnerability to flooding.

Data

The 2020 Census data were acquired through Census API. The data consist of total population, race, median age, median income, and car ownership for residents in New Orleans.

Methods

From the data gathered with the Census API, we calculated the population density and dominant racial group in each census tract. We then plotted the census demographics and analyzed the individual maps. After getting some initial results, we proceeded to perform geospatial overlay analysis for the different variables.

Analysis

From the following five maps, we conclude that:

  1. Tracts with higher population or population density are particularly vulnerable to flooding due to the increased exposure of people and infrastructure. That is, we should pay special attention to the dark areas in the first two maps. These areas often face compounded risks, as denser developments can exacerbate drainage issues and limit the effectiveness of flood mitigation strategies.

  2. Tracts with lower median income, represented by the lighter green color in the map, are particularly vulnerable to flooding due to limited access to resources for recovery and resilience-building. Residents in these areas often face challenges such as inadequate infrastructure, poor housing conditions, and fewer financial means to implement mitigation measures or relocate from high-risk zones. Addressing flooding risks in lower-income tracts requires equitable strategies that prioritize resource allocation and community support to reduce disparities in flood resilience and recovery.

  3. Observe the northeast or north side of New Orleans, where the color is darker. These tracts have a higher median age, implying that they are likely to include more elderly residents, which can present unique challenges during flooding events. Elderly individuals may face challenges such as reduced mobility, chronic health issues, and greater dependence on support systems, making evacuation or relocation more complex. These findings highlight the importance of tailored resource allocation and emergency planning to address the needs of these populations. Policymakers should ensure accessible evacuation routes, provide medical facilities, and deploy special assistance teams, as these strategies can play a critical role in enhancing the safety and well-being of elderly residents in flood-prone areas.

  4. For the map that reflects the tracts with dominant racial group, we may not be able to retrieve any useful information by itself. Therefore, we perform the geospatial overlay analysis.

We derive our additional findings from the following four maps:

  1. The relationship between racial distribution and socioeconomic factors, such as income, highlights groups that may face heightened challenges when responding to or recovering from flooding. An overlap is observed between predominantly black tracts in the racial distribution map and lower-income tracts. This suggests a need for targeted support for these vulnerable groups. Minority groups often experience systemic disadvantages, such as limited access to flood-resistant housing and fewer resources, which increases their vulnerability.

  2. The overlay map of total population and population density can be very meaningful. While the southwest part of the map may not show a large total population in the previous map, it stands out with very high population density due to the division of the tracts. This may be ignored in analyses that focus solely on total population.

  3. Another way to represent the overlay is having the color of polygon to represent both of the two variables. Shown in the map, special attention should be given to the darker purple and blue areas, as they represent medium to high population, low to average income regions. These areas likely face greater challenges during flooding events due to limited financial resources and higher population density, which can strain infrastructure and recovery efforts.

  4. We also overlaid three variables, namely the population density (represented by shades of blue polygons), median income (represented by the red gradient), and percentage of carless households (represented by bubble sizes). The city should pay special attention to regions with darker blue polygons and larger bubbles in lighter red when recovering from flooding events, as these areas likely require targeted interventions to ensure accessible transportation and evacuation plans for vulnerable residents. From the map, we see that in the center of the city, several tracts exhibit high population density, low to moderate median income, and a significant percentage of carless households.

NOTE for this map: tmap_mode(“View”) would give warning: Legend for symbol sizes not available in view mode. Therefore, this map missed the size legend for carless household rate.

We also performed the overlay for the flooding map and census demographics maps. However, the resolution of the flooding data shapefile was too high for tmap view so it would crash the program. We tried to plot it the tmap plot mode or ggplot, but the returned maps are in very bad quality (refer to Fig. 8). Therefore, we chose to conduct these overlays in ArcGIS Pro.

This overlay of the median income map with the flooding map reveals the most economically vulnerable areas. The darker green areas, covered by pink overlay, highlights regions where residents are likely to be influenced by the flooding if it occurs within New Orleans. Such residents will face significant challenges when recovering from flood-related damages due to limited financial resources.

Fig. 8: Bad quality flooding overlay.

Fig. 8: Bad quality flooding overlay.

Fig. 9: Flooding and income overlay.

Fig. 9: Flooding and income overlay.

Grocery Store Distribution Analysis

Grocery stores are essential for local residents during natural disasters. Therefore, we looked at their locations to determine whether they are evenly distributed for residents.

Data

We acquired the locations of grocery stores from open data provided by New Orleans government15, which included the locations of all the grocery stores within New Orleans. To identify the flooded regions, we used the flood hazard mapping information from the FEMA. We also acquired census data through the Census API.

Methods

Unfortunately, the grocery store data collected from New Orleans government lacked the geometry attributes of each store, as demonstrated in Fig. 10. Therefore, we used the Nominatim API to acquire the geometry data of each store. We started with 316 stores from the data, but only got the point geometry of 305 stores, so we only used these stores in our analysis.

Fig. 10: Missing point geometry data.

Fig. 10: Missing point geometry data.

After we acquired the point geometry data, we overlaid the points with the flood zone and the census tracts to determine whether we can identify a spatial pattern.

Analysis

After spatial overlay, we discover that 261 of the 315 grocery stores are not within the flood zone. This figure is ideal, as it suggests that over 85% of grocery stores are theoretically accessible should floods occur.

Fig. 11: Location of unflooded grocery stores and flood zone.

Fig. 11: Location of unflooded grocery stores and flood zone.

Fig. 11 demonstrates the grocery stores that are not within the flood zone and the flood zone itself. We do see that they are clustered along certain regions, so we further investigate whether their location is related to the income of the census tract at which they are located in.

Fig. 12: Location of unflooded grocery stores and flood zone.

Fig. 12: Location of unflooded grocery stores and flood zone.

Since the Department of Health and Human Services16 defined poverty for a family of four in 2022 as having a lower household income than US$ 27750, we selected census tracts that had a lower average income than 27750 and overlaid them with the previous 261 grocery stores. We found that there were 50 grocery stores within a below-poverty line census tract (red dots), whereas the other 211 grocery stores were located in an above-poverty line census tract (gold dots). Although the numbers are not disproportionate between below-poverty line census tracts and above-poverty line census tracts, we do see that the southeast corner of New Orleans has evidently less grocery stores, so perhaps the distribution of grocery stores could still be improved.

Conclusion

Conclusion

  • The flood vulnerability of New Orleans arises from its low-lying geography, high precipitation in certain seasons, soil type prone to surface flooding, and infrastructural vulnerabilities in its drainage systems.

  • The flood resilience of New Orleans is deeply influenced by its socioeconomic factors, such as lower income levels, flood-prone housing, high population density areas, limited accessibility to resources, aging population in certain neighborhoods, and uneven car ownership distribution. All these factors disproportionately affect the ability of certain communities to prepare for, withstand, and recover from flooding events.

  • Therefore, we suggest strengthening and modernizing levee and drainage systems, promoting equitable urban planning and resource allocation to reduce vulnerability in at-risk communities. New Orleans should also increase public access to evacuation resources to mitigate long-term risks as access to flood mitigation resources remains uneven across communities and can still be improved.

Future work

  • Satellite image issues: The satellite images used in the project were not functioning as expected, potentially due to processing errors or resolution limitations. Addressing these technical issues will improve analysis as we would be able to identify real-time flood zones instead of using static flood zones.

  • Flooding data challenges: High-resolution flooding data from FEMA often caused software crashes during data visualization or processing with packages like tmap or ggplot. Lowering the data resolution or exploring alternative visualization tools may be necessary.

  • Incorporating additional factors: Future analyses should consider additional variables such as vegetation cover, land use, and flow accumulation to gain a more comprehensive understanding of flooding risks and improve prediction accuracy.