1. Introduction

1) Story of Research Purpose and Background

On August 24, 2005, a large-scale hurricane occurred in the United States. Hurricane Katrina. It started in the southern part of the United States and affected the central part, leaving severe damage. It was Hurricane Katrina, which caused unprecedented damage from 1953 when data existed in the United States to the present.

Hurricane Katrina still has a huge impact on American society, with “Katrina Kids” and “Katrina trauma”. Does change exist after a disaster? It is important to develop a system for disaster and recovery through lessons learned from the past. That is why we focused on Hurricane Katrina.

By analyzing the emergency declaration data applied by the U.S. federal government to these Katrina, the form of the Katrina was generated is analyzed by dividing it by state, and by date and analyzing it in detail. The U.S. federal government wants to study how it responded to Katrina.

2) Main question

What was Hurricane Katrina like and how did the U.S. Federal deal with it?

3) Data introduction

data source: “Federal Emergencies and Disasters, 1953-Present” by : FEMA (Federal Emergency Management Agency) data structure: 46185 rows & 14 cols

library(readr)
federal <- read_csv("~/Desktop/미데비/federal.csv")
## Rows: 46185 Columns: 14
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (14): Declaration Number, Declaration Type, Declaration Date, State, Cou...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

Data Analysis :

str(federal)
## spc_tbl_ [46,185 × 14] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
##  $ Declaration Number              : chr [1:46185] "DR-1" "DR-2" "DR-3" "DR-4" ...
##  $ Declaration Type                : chr [1:46185] "Disaster" "Disaster" "Disaster" "Disaster" ...
##  $ Declaration Date                : chr [1:46185] "05/02/1953" "05/15/1953" "05/29/1953" "06/02/1953" ...
##  $ State                           : chr [1:46185] "GA" "TX" "LA" "MI" ...
##  $ County                          : chr [1:46185] NA NA NA NA ...
##  $ Disaster Type                   : chr [1:46185] "Tornado" "Tornado" "Flood" "Tornado" ...
##  $ Disaster Title                  : chr [1:46185] "Tornado" "Tornado and Heavy Rainfall" "Flood" "Tornado" ...
##  $ Start Date                      : chr [1:46185] "05/02/1953" "05/15/1953" "05/29/1953" "06/02/1953" ...
##  $ End Date                        : chr [1:46185] "05/02/1953" "05/15/1953" "05/29/1953" "06/02/1953" ...
##  $ Close Date                      : chr [1:46185] "06/01/1954" "01/01/1958" "02/01/1960" "02/01/1956" ...
##  $ Individual Assistance Program   : chr [1:46185] "Yes" "Yes" "Yes" "Yes" ...
##  $ Individuals & Households Program: chr [1:46185] "No" "No" "No" "No" ...
##  $ Public Assistance Program       : chr [1:46185] "Yes" "Yes" "Yes" "Yes" ...
##  $ Hazard Mitigation Program       : chr [1:46185] "Yes" "Yes" "Yes" "Yes" ...
##  - attr(*, "spec")=
##   .. cols(
##   ..   `Declaration Number` = col_character(),
##   ..   `Declaration Type` = col_character(),
##   ..   `Declaration Date` = col_character(),
##   ..   State = col_character(),
##   ..   County = col_character(),
##   ..   `Disaster Type` = col_character(),
##   ..   `Disaster Title` = col_character(),
##   ..   `Start Date` = col_character(),
##   ..   `End Date` = col_character(),
##   ..   `Close Date` = col_character(),
##   ..   `Individual Assistance Program` = col_character(),
##   ..   `Individuals & Households Program` = col_character(),
##   ..   `Public Assistance Program` = col_character(),
##   ..   `Hazard Mitigation Program` = col_character()
##   .. )
##  - attr(*, "problems")=<externalptr>
What data shows :

FEMA’s Federal Emergencies and Disasters, 1953-Present data covers emergency and major disaster events that have occurred in the United States. The President may formally declare an emergency or major disaster when he or she determines that federal assistance is essential. Such declarations supplement response and recovery efforts by state, local, or tribal governments and are intended to protect life and property, ensure public health and safety, and reduce disaster risks.

  • Emergency Declarations Emergencies are declared to prevent major disasters in advance or to protect lives and minimize damage in urgent situations. They provide federal resources and personnel support and activate emergency services to respond immediately to the threat.

  • Major Disaster Declarations A major disaster declaration is issued when damage is more severe and widespread. The President may declare a major disaster for a variety of natural disasters and emergencies, including:

Natural disasters, such as hurricanes, tornadoes, storms, floods, earthquakes, volcanic eruptions, landslides, snowstorms, and droughts Severe damage caused by any cause, such as fires or explosions A major disaster declaration provides federal assistance when damage exceeds the ability of a local area to respond, including rebuilding and repairing private and public infrastructure. Federal assistance includes financial and technical assistance for short-term emergency response as well as long-term recovery and prevention of recurrence.

This data provides important information on the frequency, causes, and magnitude of disasters, which can be used as a basis for developing national disaster management policies, hazard assessment, and improving response planning.

Variables Description

  • ‘Disaster Type’ consists of a disaster and an emergency. A disaster is a declaration for long-term recovery and reconstruction support. An emergency is a declaration that provides limited support to respond immediately to urgent situations.

  • ‘start date’ & ‘end date’ means when the disaster occurred and when it ended

  • ‘Declaration Date’ & ‘close Date’ means when the declaration occurred and when it ended

  • programs

  1. Individual Assistance Program The Individual Assistance Program provides direct financial and practical help to individuals and families impacted by a disaster. The program typically covers costs such as temporary housing, home repairs, and personal property losses that are not covered by insurance. It aims to help disaster survivors recover and rebuild their lives after an emergency.

  2. Individuals & Households Program The Individuals & Households Program is a specific component of the Individual Assistance Program. It offers financial assistance to eligible individuals and households affected by disasters.

  3. Public Assistance Program The Public Assistance Program is designed to support governments and certain nonprofits in the aftermath of a disaster.

  4. Hazard Mitigation Program The Hazard Mitigation Program focuses on reducing long-term risks to life and property from future disasters.

2. Body

1) Background on selecting Hurricane Katrina

In order to find the relative position of Hurricane Katrina compared to other disasters, the data was analyzed as follows. In the entire federal data, the variable called Disaster Title was put into the count function, sorted in descending order, and only the top 5 were extracted and saved under a new name called ‘frequency’. As a result, we could see that ‘Hurricane Katrina’ was ranked after ‘Severe Storms and Flooding’ and ‘Severe Storms, Tornadoes, and Flooding’.

frequency <- federal %>% 
    count(`Disaster Title`) %>% 
    arrange(desc(n)) %>% 
    head(5)

frequency
## # A tibble: 5 × 2
##   `Disaster Title`                           n
##   <chr>                                  <int>
## 1 Severe Storms and Flooding              6855
## 2 Severe Storms, Tornadoes, and Flooding  3241
## 3 Hurricane Katrina Evacuation            2602
## 4 Severe Winter Storm                     1911
## 5 Flooding                                1304

In order to see the relative size of Hurricane Katrina, geom_col was used. x is Disaster Title, y is the count number n, and x is specified in different colors for better visualization. In addition, geom_text was used to better show the relative frequency of Hurricane Katrina through the exact numbers on the bars.

1-1) Figure1

frequency %>% 
    ggplot(mapping = aes(x=reorder(`Disaster Title`,-n), y=n, fill = `Disaster Title`)) +
    geom_col() +
    geom_text(aes(label = n), vjust = -0.3) +
    theme(axis.text.x = element_text(size=3,face='bold')) +
    labs(title = "Background on selecting Hurricane Katrina",
       x = "Disaster Title 5th",
       y = "Number of disasters")

As a result, Hurricane Katrina was the only variable to which a specific disaster name was mentioned and ranked third.

Data were extracted after changing the date to a form that could be calculated for use with Katrina and the data thereafter. And data that correspond only to Hurricane Katrina were extracted.

Data cleaning

federal$`Start Date` <- as.Date(federal$`Start Date`, "%m/%d/%Y")

fed <- federal %>% #2005년 이후 데이터, 카트리나
  filter(`Start Date` >= '2005-01-01' &
         `Disaster Title` == "Hurricane Katrina")

2) Hurricane Katrina Damage Rate by State

To find out which states Hurricane Katrina struck and what percentage of the total damage each state had, we analyzed the data as follows. We counted the states in fed, which is limited to Hurricane Katrina, and introduced a new variable called pct to calculate the percentage of the total damage each state had, and saved it under a new name called ‘rate.’

rate <- fed %>% 
    count(State) %>% 
    mutate(`pct` = round(n/sum(n)*100, 2)) 
rate
## # A tibble: 7 × 3
##   State     n   pct
##   <chr> <int> <dbl>
## 1 AL       28  5.71
## 2 FL       11  2.24
## 3 ID       44  8.98
## 4 KY      120 24.5 
## 5 LA      127 25.9 
## 6 MS       93 19.0 
## 7 PA       67 13.7

In order to visualize this state-by-state ratio data, a pie chart was needed. Therefore, we first separated the colors by state, then created geom_bar, and inserted a white line between the states to clearly separate them. we created a pie chart using coord_polar, and in the same way, we used geom_text to directly express the numbers on the graph so that pct could be intuitively viewed.

2-1) Figure2

rate %>% 
    ggplot(mapping = aes(x='', y=pct, fill=State)) +
    geom_bar(stat='identity', color = "white") +
    theme_void() +
    coord_polar('y', start=0) +
    geom_text(aes(label=paste0(round(pct,1), '%')),
            position=position_stack(vjust=0.5)) +
    labs(title = "Hurricane Katrina Damage Rate by State")

As a result, it was confirmed that more than 50% of the total damage occurred in Louisiana and Kentucky, and that five other areas suffered significant damage.

3) Hurricane Katrina damage figures by day of occurrence by State

The graph above shows that LA and KY were damaged by Hurricane Katrina at similar rates. However, since natural disasters do not occur all at once but rather over time, we wanted to check the damage patterns by Hurricane Katrina start date and state in detail. we selected State and Start Date from the Fed data, counted State and Start Date, and created data, and saved the data in state_of_fed.

#날짜별 주별 카트리나 시작 및 크기
state_of_fed <- fed %>% 
    select(`State`, `Start Date`) %>% 
    count(State, `Start Date`) %>% 
    arrange(desc(n))
state_of_fed 
## # A tibble: 10 × 3
##    State `Start Date`     n
##    <chr> <date>       <int>
##  1 KY    2005-08-29     120
##  2 MS    2005-08-29      82
##  3 PA    2005-08-29      67
##  4 LA    2005-08-29      64
##  5 LA    2005-08-26      63
##  6 ID    2005-08-29      44
##  7 AL    2005-08-29      22
##  8 FL    2005-08-24      11
##  9 MS    2005-08-27      11
## 10 AL    2005-08-28       6

When we looked at the table where the data was analyzed, the damage patterns by date and state were not clearly visible, so we used geom_point and geom_line graphs to identify the damage patterns. we used the scale_x_date function so that all dates could be mentioned on the X-axis.

3-1) Figure3

state_of_fed %>% 
    ggplot(mapping = aes(x=`Start Date`, y= n, color = `State`))+
    geom_point(size=3) +
    geom_line(size=1.5) +
    scale_x_date(date_breaks = "1 days") +
    labs(title = "Hurricane Katrina damage figures by day of occurrence by State")
## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.

Data was visualized by date to compare the two worst-hit states. As a result, Louisiana and Mississippi suffered almost the same number of damages, but the pattern was different. Louisiana suffered from a constant intensity for four days, but Mississippi experienced an exponential increase over three days.

4) Application status of each of the four support programs

If so, did the affected areas receive support at the national level proportional to it?

After analyzing up to this point, we wondered whether the affected area was supported by the reasonable assistance program of the US federal government. Therefore, we grouped the cases of the four assistance programs in the data, “Individual Assistance Program”, “Individuals & Households Program”, “Public Assistance Program”, and “Hazard Mitigation Program”, and identified them by state.

In order to visualize the four analyzed data at once, we first visualized them individually. we created bars using geom_col and assigned different colors to each state. we made the exact size of the bars known using geom_text and saved them in pro1~4 respectively.

4-1) Figure4

pro1 <- fed %>%
    filter(`Individual Assistance Program` == "Yes") %>% 
    group_by(State) %>% 
    count() %>% 
    ggplot(aes(x=State, y=n, fill=State)) +
    geom_col() +
    geom_text(aes(label = n), vjust = 1.5) +
    labs(title = "Individual Assistance Program")

pro2 <-fed %>%
    filter(`Individuals & Households Program` == "Yes") %>% 
    group_by(State) %>% 
    count() %>% 
    ggplot(aes(x=State, y=n, fill=State)) +
    geom_col() +
    geom_text(aes(label = n), vjust = 1.5) +
    labs(title = "Individuals & Households Program")

pro3 <- fed %>%
    filter(`Public Assistance Program` == "Yes") %>% 
    group_by(State) %>% 
    count() %>% 
    ggplot(aes(x=State, y=n, fill=State)) +
    geom_col() +
    geom_text(aes(label = n), vjust = 1.5) +
    labs(title = "Public Assistance Program")
    
pro4 <- fed %>%
    filter(`Hazard Mitigation Program` == "Yes") %>% 
    group_by(State) %>% 
    count() %>% 
    ggplot(aes(x=State, y=n, fill=State)) +
    geom_col() + 
    geom_text(aes(label = n), vjust = 1.5) +
    labs(title = "Hazard Mitigation Program")

After downloading the package called patchwork, we saved four graphs at once under the name combined_plot and visualized them at once.

install.packages("patchwork", repos = "https://cloud.r-project.org")
## 
## The downloaded binary packages are in
##  /var/folders/h4/rxfgjm816cjbh5s5lhtyl0_h0000gn/T//RtmpiR1HtX/downloaded_packages
library(patchwork)

combined_plot <- (pro1 + pro2) / (pro3 + pro4)
combined_plot

As a result, it was confirmed that all three programs except the public assistance program provided more support to Mississippi, which is relatively less affected than Louisiana, which is the worst affected.

3. Results

Hurricane Katrina was recorded as one of the greatest disasters in American history, and its impact and aftermath are still being recalled in various aspects to this day. This disaster caused widespread damage, mainly in the central and southern regions of the United States, and took away countless lives and properties across seven states. Louisiana and Kentucky were particularly severely hit, accounting for nearly half of the damage declared by Katrina. In addition, five states, including Mississippi, Alabama, and Florida, were directly or indirectly affected.

However, the federal assistance programs implemented for disaster recovery had several limitations. A relatively large number of assistance programs were concentrated in Mississippi, rather than Louisiana, which suffered the most severe damage. This unbalanced support showed the inadequacy of the U.S. federal government in the disaster response and recovery process. This can be seen as a case that clearly revealed the structural flaws of FEMA (the U.S. Federal Emergency Management Agency).

This experience became an important turning point in the U.S. disaster response system. After Hurricane Katrina, the federal government and FEMA began to comprehensively reorganize their disaster response systems and operations. In particular, major improvements were made to ensure rapid and systematic support in the event of a disaster.

The change was confirmed by Hurricane Rita, which occurred right after Katrina. Hurricane Rita hit the United States on a larger scale than Katrina, but the scale of damage was relatively greatly reduced thanks to FEMA’s improved response system. This shows that disaster management capabilities have greatly improved.

As a result, Hurricane Katrina left a painful mark in the history of disaster response in the United States, but it provided an opportunity for the U.S. federal government to further develop its disaster management system. This change played an important role in minimizing damage from subsequent disasters. We hope that the lessons learned from Hurricane Katrina will serve as a foundation for creating a better disaster response system for future domestic and international disasters.