Health and Economic Impacts of Severe Weather Events in the United States

Synopsis

This analysis examines severe weather events recorded in the NOAA Storm Database from 1950 through November 2011.

Population health impacts are measured using the combined number of fatalities and injuries associated with each event type. Economic impacts are measured using reported property and crop damage after converting the magnitude codes into dollar values. Event types are aggregated across the United States and ranked by their total impacts.

Through this analysis I hope to identify the event categories responsible for the greatest health and economic consequences in the United States (U.S.A).

Data Processing

Load dataset

storm_data <- read.csv(
  "repdata_data_StormData.csv",
  stringsAsFactors = FALSE
)

dim(storm_data)

## [1] 902297     37

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"

For the analysis, the key variables are:

EVTYPE: identifies the type of severe weather event. It is used to group the observations and compare the effects of different event categories.
FATALITIES: records the number of deaths associated with each event. It is used to measure harm to population health.
INJURIES: records the number of injuries associated with each event. It is combined with fatalities to calculate total health impact.
PROPDMG: records the reported amount of property damage caused by an event.
PROPDMGEXP: indicates the scale of the property damage value, such as thousands (K), millions (M), or billions (B). It is used to convert PROPDMG into dollars.
CROPDMG: records the reported amount of crop damage caused by an event.
CROPDMGEXP: indicates the scale of the crop damage value and is used to convert CROPDMG into dollars.

Data Cleaning

The raw NOAA Storm Database contains inconsistencies in event names and damage-value codes. Several transformations were therefore required before calculating health and economic impacts.

First, event names were converted to uppercase and surrounding spaces were removed. This reduces differences caused only by capitalization or accidental spacing. However, the original event categories were otherwise retained because extensive manual recoding could introduce subjective classification decisions.

storm_clean <- storm_data[
  ,
  c(
    "EVTYPE",
    "FATALITIES",
    "INJURIES",
    "PROPDMG",
    "PROPDMGEXP",
    "CROPDMG",
    "CROPDMGEXP"
  )
]

storm_clean$EVTYPE <- toupper(
  trimws(storm_clean$EVTYPE)
)

storm_clean$PROPDMGEXP <- toupper(
  trimws(storm_clean$PROPDMGEXP)
)

storm_clean$CROPDMGEXP <- toupper(
  trimws(storm_clean$CROPDMGEXP)
)

Measuring Population-Health Impact

The database reports fatalities and injuries separately. Total health impact was calculated as their sum for each observation.

storm_clean$HEALTH_IMPACT <- (
  storm_clean$FATALITIES +
  storm_clean$INJURIES
)

health_totals <- aggregate(
  cbind(FATALITIES, INJURIES, HEALTH_IMPACT) ~ EVTYPE,
  data = storm_clean,
  FUN = sum,
  na.rm = TRUE
)

health_totals <- health_totals[
  order(
    health_totals$HEALTH_IMPACT,
    decreasing = TRUE
  ),
]

top_health <- head(health_totals, 10)

top_health

##                EVTYPE FATALITIES INJURIES HEALTH_IMPACT
## 750           TORNADO       5633    91346         96979
## 108    EXCESSIVE HEAT       1903     6525          8428
## 771         TSTM WIND        504     6957          7461
## 146             FLOOD        470     6789          7259
## 410         LIGHTNING        816     5230          6046
## 235              HEAT        937     2100          3037
## 130       FLASH FLOOD        978     1777          2755
## 379         ICE STORM         89     1975          2064
## 677 THUNDERSTORM WIND        133     1488          1621
## 880      WINTER STORM        206     1321          1527

Converting Economic Damage into Dollars

The variables PROPDMG and CROPDMG are accompanied by exponent codes that indicate their magnitude. The principal codes are:

K: thousands of dollars
M: millions of dollars
B: billions of dollars
H: hundreds of dollars
Numeric codes: powers of ten

The property- and crop-damage variables are not recorded directly in dollars. Instead, PROPDMG and CROPDMG contain numerical values, while PROPDMGEXP and CROPDMGEXP indicate the corresponding magnitude. For example, K, M, and B represent thousands, millions, and billions.

A function was created to convert the exponent codes into numerical multipliers.

damage_multiplier <- function(code) {
  code <- toupper(trimws(as.character(code)))

  multiplier <- rep(1, length(code))

  multiplier[code == "H"] <- 1e2
  multiplier[code == "K"] <- 1e3
  multiplier[code == "M"] <- 1e6
  multiplier[code == "B"] <- 1e9

  numeric_code <- code %in% as.character(0:9)

  multiplier[numeric_code] <- 10^as.numeric(
    code[numeric_code]
  )

  multiplier
}

The multipliers were then applied to the reported property and crop-damage values. Total economic damage was calculated as the sum of property and crop losses.

storm_clean$PROP_MULTIPLIER <- damage_multiplier(
  storm_clean$PROPDMGEXP
)

storm_clean$CROP_MULTIPLIER <- damage_multiplier(
  storm_clean$CROPDMGEXP
)

storm_clean$PROPERTY_DAMAGE <- (
  storm_clean$PROPDMG *
  storm_clean$PROP_MULTIPLIER
)

storm_clean$CROP_DAMAGE <- (
  storm_clean$CROPDMG *
  storm_clean$CROP_MULTIPLIER
)

storm_clean$ECONOMIC_DAMAGE <- (
  storm_clean$PROPERTY_DAMAGE +
  storm_clean$CROP_DAMAGE
)

economic_totals <- aggregate(
  cbind(
    PROPERTY_DAMAGE,
    CROP_DAMAGE,
    ECONOMIC_DAMAGE
  ) ~ EVTYPE,
  data = storm_clean,
  FUN = sum,
  na.rm = TRUE
)

economic_totals <- economic_totals[
  order(
    economic_totals$ECONOMIC_DAMAGE,
    decreasing = TRUE
  ),
]

top_economic <- head(economic_totals, 10)

top_economic

##                EVTYPE PROPERTY_DAMAGE CROP_DAMAGE ECONOMIC_DAMAGE
## 146             FLOOD    144657709807  5661968450    150319678257
## 364 HURRICANE/TYPHOON     69305840000  2607872800     71913712800
## 750           TORNADO     56947380677   414953270     57362333947
## 591       STORM SURGE     43323536000        5000     43323541000
## 204              HAIL     15735267513  3025954473     18761221986
## 130       FLASH FLOOD     16822723979  1421317100     18244041079
## 76            DROUGHT      1046106000 13972566000     15018672000
## 355         HURRICANE     11868319010  2741910000     14610229010
## 521       RIVER FLOOD      5118945500  5029459000     10148404500
## 379         ICE STORM      3944927860  5022113500      8967041360

Results

Events Most Harmful to Population Health

The following table presents the ten event types with the largest combined number of fatalities and injuries.

knitr::kable(
  top_health,
  col.names = c(
    "Event type",
    "Fatalities",
    "Injuries",
    "Total health impact"
  ),
  caption = "Ten weather-event types with the greatest population-health impact."
)

Ten weather-event types with the greatest population-health impact.
	Event type	Fatalities	Injuries	Total health impact
750	TORNADO	5633	91346	96979
108	EXCESSIVE HEAT	1903	6525	8428
771	TSTM WIND	504	6957	7461
146	FLOOD	470	6789	7259
410	LIGHTNING	816	5230	6046
235	HEAT	937	2100	3037
130	FLASH FLOOD	978	1777	2755
379	ICE STORM	89	1975	2064
677	THUNDERSTORM WIND	133	1488	1621
880	WINTER STORM	206	1321	1527

par(mar = c(5, 12, 4, 2))

barplot(
  rev(top_health$HEALTH_IMPACT),
  names.arg = rev(top_health$EVTYPE),
  horiz = TRUE,
  las = 1,
  main = "Weather Events Most Harmful to Population Health",
  xlab = "Combined fatalities and injuries",
  col = "steelblue",
  border = NA
)

The ten weather-event types associated with the greatest combined number of fatalities and injuries. Longer bars indicate greater total harm to population health.

The event type associated with the greatest combined health impact was TORNADO, with 96,979 recorded fatalities and injuries. It caused 5,633 fatalities and 91,346 injuries.

The ranking indicates that the health burden of severe weather is concentrated among a relatively small number of event categories. Events may rank highly because they occur frequently, because individual events are especially dangerous, or because of a combination of both factors.

Events with the Greatest Economic Consequences

For easier interpretation, economic damage is displayed in billions of dollars.

top_economic$PROPERTY_BILLIONS <- (
  top_economic$PROPERTY_DAMAGE / 1e9
)

top_economic$CROP_BILLIONS <- (
  top_economic$CROP_DAMAGE / 1e9
)

top_economic$TOTAL_BILLIONS <- (
  top_economic$ECONOMIC_DAMAGE / 1e9
)

knitr::kable(
  top_economic[
    ,
    c(
      "EVTYPE",
      "PROPERTY_BILLIONS",
      "CROP_BILLIONS",
      "TOTAL_BILLIONS"
    )
  ],
  digits = 2,
  col.names = c(
    "Event type",
    "Property damage ($ billions)",
    "Crop damage ($ billions)",
    "Total damage ($ billions)"
  ),
  caption = "Ten weather-event types with the greatest economic consequences."
)

Ten weather-event types with the greatest economic consequences.
	Event type	Property damage ($ billions)	Crop damage ($ billions)	Total damage ($ billions)
146	FLOOD	144.66	5.66	150.32
364	HURRICANE/TYPHOON	69.31	2.61	71.91
750	TORNADO	56.95	0.41	57.36
591	STORM SURGE	43.32	0.00	43.32
204	HAIL	15.74	3.03	18.76
130	FLASH FLOOD	16.82	1.42	18.24
76	DROUGHT	1.05	13.97	15.02
355	HURRICANE	11.87	2.74	14.61
521	RIVER FLOOD	5.12	5.03	10.15
379	ICE STORM	3.94	5.02	8.97

par(mar = c(5, 12, 4, 2))

barplot(
  rev(top_economic$TOTAL_BILLIONS),
  names.arg = rev(top_economic$EVTYPE),
  horiz = TRUE,
  las = 1,
  main = "Weather Events with the Greatest Economic Consequences",
  xlab = "Combined property and crop damage ($ billions)",
  col = "tomato",
  border = NA
)

The ten weather-event types associated with the greatest combined property and crop damage. Economic losses are expressed in billions of dollars.

The event type associated with the greatest economic loss was FLOOD, producing approximately $150.32 billion in combined property and crop damage.

Of this amount, approximately $144.66 billion came from property damage and $5.66 billion came from crop damage.

Summary

The analysis shows that the event types creating the greatest population-health burden are not necessarily the same as those producing the greatest economic losses. TORNADO had the highest combined number of fatalities and injuries, whereas FLOOD caused the greatest combined property and crop damage. This distinction is important because health impacts and economic impacts represent different dimensions of severe-weather risk. Municipal and emergency-management officials may therefore need to consider both rankings when prioritizing preparedness and response resources.