Analyzing the US Disasters with the Highest Economic and Societal Impact
Teodor Nedevski
2025-01-28
Synopsis
This analysis focuses on identifying and comparing the most impactful natural disasters in the U.S. based on two main criteria: their consequences on human health and their economic damages. Using data from the NOAA Storm Database, we explore event types associated with the highest fatalities, injuries, and property damage. By examining both average and total figures, this report provides insights into the leading contributors to societal harm and economic loss.
Data Processing
Data Description: The data set being used is derived from the NOAA Storm Database ranging from the 1950s to 2011; for the purposes of this project the version uploaded to the course platform is used to ensure consistency. Download Storm Data
Loading the data
storm_data <- read.csv("./repdata-data-StormData.csv")
names(storm_data)## [1] "STATE__" "BGN_DATE" "BGN_TIME" "TIME_ZONE" "COUNTY"
## [6] "COUNTYNAME" "STATE" "EVTYPE" "BGN_RANGE" "BGN_AZI"
## [11] "BGN_LOCATI" "END_DATE" "END_TIME" "COUNTY_END" "COUNTYENDN"
## [16] "END_RANGE" "END_AZI" "END_LOCATI" "LENGTH" "WIDTH"
## [21] "F" "MAG" "FATALITIES" "INJURIES" "PROPDMG"
## [26] "PROPDMGEXP" "CROPDMG" "CROPDMGEXP" "WFO" "STATEOFFIC"
## [31] "ZONENAMES" "LATITUDE" "LONGITUDE" "LATITUDE_E" "LONGITUDE_"
## [36] "REMARKS" "REFNUM"From the documentation from NOAA, the following descriptions pertinent to the analysis are available: “Damage. Property damage estimates should be entered as actual dollar amounts” the variables i have identified that correspond to damage are:
PROPDMG (amount). and
PROPDMGEXP(multiplier, Alphabetical characters used to
signify magnitude: include “K” for thousands, “M” for millions, and “B”
for billions) CROPDMG (amount) and
CROPDMGEXP(multiplier, Alphabetical characters used to
signify magnitude same as PROPDMGEXP)
variables relating to the demographic health:
FATALITIES indicates fatalities attributed to the
event
INJURIES indicates injuries attributed to the event
and the category of the event EVTYPE
simplifying the data frame to only include data relevant to population health and economic consequences:
relevant_df <- storm_data[, c("EVTYPE", "INJURIES", "FATALITIES", "CROPDMG", "CROPDMGEXP", "PROPDMG", "PROPDMGEXP")]
names(relevant_df)## [1] "EVTYPE" "INJURIES" "FATALITIES" "CROPDMG" "CROPDMGEXP"
## [6] "PROPDMG" "PROPDMGEXP"Looking for the largest offenders for fatalities and injuries
Creating a table that summarizes the average and cumulative mortality for each event
# Calculate average fatalities and injuries for each EVTYPE
avg_fatalities <- aggregate(FATALITIES ~ EVTYPE, data = relevant_df, FUN = mean, na.rm = TRUE)
avg_injuries <- aggregate(INJURIES ~ EVTYPE, data = relevant_df, FUN = mean, na.rm = TRUE)
# Total fatalities and injuries by EVTYPE
total_fatalities <- aggregate(FATALITIES ~ EVTYPE, data = relevant_df, FUN = sum, na.rm = TRUE)
total_injuries <- aggregate(INJURIES ~ EVTYPE, data = relevant_df, FUN = sum, na.rm = TRUE)
population_df <- merge(avg_fatalities, avg_injuries, by = "EVTYPE")
population_df <- merge(population_df, total_fatalities, by = "EVTYPE")
population_df <- merge(population_df, total_injuries, by = "EVTYPE")sorting the population data frame by highest A) average fatality rate B) average injury rate C) total fatalities D) total injuries
# Load necessary libraries
library(tidyr)
# Sort the dataframe for each category: average and total fatalities, injuries
sorted_by_avg_fatalities <- population_df[order(-population_df$FATALITIES.x), ]
sorted_by_avg_injuries <- population_df[order(-population_df$INJURIES.x), ]
sorted_by_total_fatalities <- population_df[order(-population_df$FATALITIES.y), ]
sorted_by_total_injuries <- population_df[order(-population_df$INJURIES.y), ]
# Get the top 5 for each category
top5_avg_fatalities <- sorted_by_avg_fatalities[1:5, c("EVTYPE", "FATALITIES.x", "INJURIES.x")]
top5_avg_injuries <- sorted_by_avg_injuries[1:5, c("EVTYPE", "FATALITIES.x", "INJURIES.x")]
top5_total_fatalities <- sorted_by_total_fatalities[1:5, c("EVTYPE", "FATALITIES.y", "INJURIES.y")]
top5_total_injuries <- sorted_by_total_injuries[1:5, c("EVTYPE", "FATALITIES.y", "INJURIES.y")]
# Reshape data for plotting using gather
top5_avg_fatalities_long <- gather(top5_avg_fatalities, key = "Type", value = "Value", -EVTYPE)
top5_avg_fatalities_long$Type <- factor(top5_avg_fatalities_long$Type, levels = c("FATALITIES.x", "INJURIES.x"), labels = c("Fatalities", "Injuries"))
top5_avg_injuries_long <- gather(top5_avg_injuries, key = "Type", value = "Value", -EVTYPE)
top5_avg_injuries_long$Type <- factor(top5_avg_injuries_long$Type, levels = c("FATALITIES.x", "INJURIES.x"), labels = c("Fatalities", "Injuries"))
top5_total_fatalities_long <- gather(top5_total_fatalities, key = "Type", value = "Value", -EVTYPE)
top5_total_fatalities_long$Type <- factor(top5_total_fatalities_long$Type, levels = c("FATALITIES.y", "INJURIES.y"), labels = c("Fatalities", "Injuries"))
top5_total_injuries_long <- gather(top5_total_injuries, key = "Type", value = "Value", -EVTYPE)
top5_total_injuries_long$Type <- factor(top5_total_injuries_long$Type, levels = c("FATALITIES.y", "INJURIES.y"), labels = c("Fatalities", "Injuries"))Calcualting the average and total economic cost of each event type
Converting Damage Amounts Based on Multipliers
Natural disasters use specific multipliers for property
(PROPDMGEXP) and crop (CROPDMGEXP) damage
values, like “K” for thousands, “M” for millions, and “B” for billions.
function to convert these multipliers into actual numeric values.
# Function to convert damage amounts based on multiplier
convert_damage <- function(damage, exponent) {
if (exponent == "K") {
return(damage * 1e3)
} else if (exponent == "M") {
return(damage * 1e6)
} else if (exponent == "B") {
return(damage * 1e9)
} else {
return(damage)
}
}This function will be applied to each row in the dataset to ensure property and crop damage values are converted into their actual amounts.
Applying the Conversion to Damage Columns
Now, we will apply the convert_damage function to the
PROPDMG and CROPDMG columns using the
respective multipliers in PROPDMGEXP and
CROPDMGEXP. We will create new columns to hold the actual
damage values.
# Apply the conversion to property and crop damage columns without using mutate
PROPDMG_ACTUAL <- mapply(convert_damage, relevant_df$PROPDMG, relevant_df$PROPDMGEXP)
CROPDMG_ACTUAL <- mapply(convert_damage, relevant_df$CROPDMG, relevant_df$CROPDMGEXP)
# Add the converted damage columns to the original data frame
relevant_df$PROPDMG_ACTUAL <- PROPDMG_ACTUAL
relevant_df$CROPDMG_ACTUAL <- CROPDMG_ACTUALNow, the relevant_df dataset will contain two new
columns: PROPDMG_ACTUAL for property damage and
CROPDMG_ACTUAL for crop damage in actual dollar
amounts.
Calculating Total Damage for Each Event Type
Next, we will calculate the total damage for each event type by summing both the property and crop damage amounts. This will help us compare the overall economic consequences of each event type.
# Calculate the total economic damage (property + crop damage) for each event type without using mutate
TOTAL_DAMAGE <- relevant_df$PROPDMG_ACTUAL + relevant_df$CROPDMG_ACTUAL
# Add the calculated total damage to the original data frame
relevant_df$TOTAL_DAMAGE <- TOTAL_DAMAGEThis creates a new column TOTAL_DAMAGE that contains the
combined property and crop damage for each event type.
Summarizing the Damage by Event Type
We will now summarize the total damage by event type
(EVTYPE). This will allow us to identify which events
caused the most economic damage overall, including both property and
crop damage.
# Summarize the total damage by event type
damage_summary <- aggregate(
cbind(PROPDMG_ACTUAL, CROPDMG_ACTUAL, TOTAL_DAMAGE) ~ EVTYPE,
data = relevant_df,
FUN = sum,
na.rm = TRUE
)
# Sort the damage_summary manually by TOTAL_DAMAGE in descending order
damage_summary_sorted <- damage_summary[order(-damage_summary$TOTAL_DAMAGE), ]
# View the top 10 events with the greatest economic consequences
head(damage_summary_sorted, 10) ## EVTYPE PROPDMG_ACTUAL CROPDMG_ACTUAL TOTAL_DAMAGE
## 170 FLOOD 144657709807 5661968450 150319678257
## 411 HURRICANE/TYPHOON 69305840000 2607872800 71913712800
## 834 TORNADO 56925660790 414953270 57340614060
## 670 STORM SURGE 43323536000 5000 43323541000
## 244 HAIL 15727367053 3025537890 18752904943
## 153 FLASH FLOOD 16140812067 1421317100 17562129167
## 95 DROUGHT 1046106000 13972566000 15018672000
## 402 HURRICANE 11868319010 2741910000 14610229010
## 590 RIVER FLOOD 5118945500 5029459000 10148404500
## 427 ICE STORM 3944927860 5022113500 8967041360# Summarize the average damage by event type
damage_summary_avg <- aggregate(
cbind(PROPDMG_ACTUAL, CROPDMG_ACTUAL, TOTAL_DAMAGE) ~ EVTYPE,
data = relevant_df,
FUN = mean, # Use mean to calculate averages
na.rm = TRUE # Remove NA values
)
# Sort the damage_summary_avg manually by TOTAL_DAMAGE in descending order
damage_summary_sorted_avg <- damage_summary_avg[order(-damage_summary_avg$TOTAL_DAMAGE), ]
# View the top 10 events with the greatest average economic consequences
head(damage_summary_sorted_avg, 10)## EVTYPE PROPDMG_ACTUAL CROPDMG_ACTUAL TOTAL_DAMAGE
## 842 TORNADOES, TSTM WIND, HAIL 1600000000 2.500000e+06 1602500000
## 299 HEAVY RAIN/SEVERE WEATHER 1250000000 0.000000e+00 1250000000
## 411 HURRICANE/TYPHOON 787566364 2.963492e+07 817201282
## 408 HURRICANE OPAL 350316224 1.000001e+06 351316226
## 670 STORM SURGE 165990559 1.915709e+01 165990579
## 954 WILD FIRES 156025000 0.000000e+00 156025000
## 136 EXCESSIVE WETNESS 0 1.420000e+08 142000000
## 409 HURRICANE OPAL/HIGH WINDS 100000000 1.000000e+07 110000000
## 604 SEVERE THUNDERSTORM 92720000 1.538462e+04 92735385
## 402 HURRICANE 68208730 1.575810e+07 83966833The damage_summary data frame will show the total damage
for each event type, sorted by the highest total damage. This will help
us identify which event types have the greatest economic impact.
Results
Across the United States, which types of events are most harmful with respect to population health?
Creating a plot for the top contributors to population health disturbances:
# Load necessary libraries
library(gridExtra)
library(grid)
# Set up plotting area for 2x2 grid (base R)
par(mfrow = c(2, 2), mar = c(11, 6, 3, 2))
# Plot 1: Top 5 by average fatalities
barplot(top5_avg_fatalities$FATALITIES.x,
names.arg = top5_avg_fatalities$EVTYPE,
col = "steelblue",
main = "A) Top 5 Events by Average Fatalities",
xlab = "",
ylab = "Count",
las = 2, # Rotate x-axis labels
cex.names = 0.8) # Adjust the size of event type labels
# Plot 2: Top 5 by average injuries (Use the correct data)
barplot(top5_avg_injuries$INJURIES.x,
names.arg = top5_avg_injuries$EVTYPE,
col = "darkorange",
main = "B) Top 5 Events by Average Injuries",
xlab = "",
ylab = "Count",
las = 2,
cex.names = 0.8)
# Plot 3: Top 5 by total fatalities
barplot(top5_total_fatalities$FATALITIES.y,
names.arg = top5_total_fatalities$EVTYPE,
col = "darkred",
main = "C) Top 5 Events by Total Fatalities",
xlab = "",
ylab = "Count",
las = 2,
cex.names = 0.8)
# Plot 4: Top 5 by total injuries (Use the correct data)
barplot(top5_total_injuries$INJURIES.y,
names.arg = top5_total_injuries$EVTYPE,
col = "green",
main = "D) Top 5 Events by Total Injuries",
xlab = "",
ylab = "Count",
las = 2,
cex.names = 0.8)
Figure 1 Comparing the averages and totals of
fatalities and injuries for the top 5 event types
From the data, we can see that while the average rates for both fatalities and injuries ted to have varying causation, that include hurricanes, heat/cold waves and tornadoes, the Tornado category really takes a commanding presence in the cumulative fatality and injury rates.
Looking into the data collection standards, one weather pattern can cause many tornado touchdowns which means that in one period many different tornado entries for the same weather event. This leads to a dilution of the average fatality and injury rates, while we can see that tornados are clearly the largest contributors to the totals.
Nonetheless, tornadoes are still one of the contributing factors in the highest average fatality rate, so we can safely conclude that tornadoes have historically caused the most deaths in the US.
Looking into other events such as excessive heat I expect the sahare of fatalities and injuries to increase for this type of event as climate change causes an increase in sudden extreme temperature swings.
Across the United States, which types of events have the greatest economic consequences?
# Prepare the data for the bar plot (Total Damage)
top10_data_total <- head(damage_summary_sorted, 10)
damage_matrix_total <- t(as.matrix(top10_data_total[, c("TOTAL_DAMAGE", "CROPDMG_ACTUAL", "PROPDMG_ACTUAL")]))
# Prepare the data for the bar plot (Average Damage)
top10_data_avg <- head(damage_summary_sorted_avg, 10)
damage_matrix_avg <- t(as.matrix(top10_data_avg[, c("TOTAL_DAMAGE", "CROPDMG_ACTUAL", "PROPDMG_ACTUAL")]))
# Set up a 2x1 layout for the stacked panel (one plot on top of the other)
par(mfrow = c(2, 1), mar = c(7, 6, 3, 2))
# Plot 1: Total Damage by Event Type
barplot(
damage_matrix_total,
beside = TRUE,
col = c("darkred", "darkgreen", "steelblue"),
names.arg = top10_data_total$EVTYPE,
main = "A) Total Damage by Event Type",
xlab = "",
ylab = "Damage (in billions)",
las = 2,
cex.names = 0.6,
legend.text = c("Total Damage", "Crop Damage", "Property Damage"),
args.legend = list(x = "topright", bty = "n"),
axes = FALSE
)
# Customize Y-axis for Total Damage to show values in billions (B)
yticks <- axTicks(2)
yticks_labels <- sapply(yticks, function(x) {
paste0(round(x / 1e9, 1), "B")
})
axis(2, at = yticks, labels = yticks_labels)
# Plot 2: Average Damage by Event Type
barplot(
damage_matrix_avg,
beside = TRUE,
col = c("darkred", "darkgreen", "steelblue"),
names.arg = top10_data_avg$EVTYPE,
main = "B) Average Damage by Event Type",
xlab = "",
ylab = "Damage (in billions)",
las = 2,
cex.names = 0.6,
legend.text = c("Total Damage", "Crop Damage", "Property Damage"),
args.legend = list(x = "topright", bty = "n"),
axes = FALSE
)
# Customize Y-axis for Average Damage to show values in billions (B)
yticks <- axTicks(2)
yticks_labels <- sapply(yticks, function(x) {
paste0(round(x / 1e9, 1), "B")
})
axis(2, at = yticks, labels = yticks_labels)
Figure 2 Comparing the averages and totals of
damages for the top 10 event types When evaluating the
economic costs of natural disasters, Tornadoes and Inclement/excessive
rain/ extreme weather emerge as the leading contributors to both average
property and crop damage, while when breaking down by category Droughts
and Excessive wetness do the most damage in agriculture, as one might
expect.
However the totals are decisively indicating that flooding is the single largest source of inclement weather damage, followed by a cluster of hurricane, tornado and storm surge damages which have a similar total. Rising sea levels will only increase the totals of flooding and storm surge.