Synopsis

This report analyzes the U.S. National Oceanic and Atmospheric Administration (NOAA) Storm Events Database to identify which event types are most harmful to population health and which have the greatest economic consequences across the United States. We start from the raw CSV repdata_data_StormData.csv (no external pre-processing). We compute health impacts (fatalities, injuries) and standardized economic losses (property + crop). We show all R code (echo = TRUE) and cache heavy steps for speed. Our results typically show that a small number of event types account for a large share of adverse outcomes: tornadoes dominate injuries and fatalities, while floods, hurricanes/typhoons, and storm surges account for the largest economic losses. These findings can help public officials prioritize preparedness and mitigation resources.

Data Processing

Packages

# Expect these to be installed already. If not, run the install script first.
library(dplyr)

## 
## Adjuntando el paquete: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(ggplot2)
library(stringr)
library(readr)
library(forcats)
library(tidyr)
library(knitr)

Load raw data

csv_path <- "repdata_data_StormData.csv"
stopifnot(file.exists(csv_path))

storm_raw <- readr::read_csv(
  csv_path,
  show_col_types = FALSE,
  progress = FALSE
)

storm <- storm_raw %>%
  dplyr::select(
    BGN_DATE, STATE, EVTYPE,
    FATALITIES, INJURIES,
    PROPDMG, PROPDMGEXP,
    CROPDMG, CROPDMGEXP
  )

rm(storm_raw)

Variable Description (Data Dictionary)

The following table describes the original key variables used in this analysis and the derived fields we create. All text is taken from NOAA documentation conventions and common usage in the Storm Events dataset.

var_dict <- tibble::tibble(
  Variable = c(
    "BGN_DATE","STATE","EVTYPE",
    "FATALITIES","INJURIES",
    "PROPDMG","PROPDMGEXP",
    "CROPDMG","CROPDMGEXP",
    "prop_mult","crop_mult",
    "prop_loss","crop_loss","total_loss"
  ),
  Type = c(
    "Date/Time","Factor/Character","Factor/Character",
    "Numeric","Numeric",
    "Numeric","Character",
    "Numeric","Character",
    "Numeric","Numeric",
    "Numeric (USD)","Numeric (USD)","Numeric (USD)"
  ),
  Description = c(
    "Date the event began.",
    "U.S. state/territory code.",
    "Event type label as recorded (e.g., TORNADO, FLOOD).",
    "Number of deaths directly/indirectly attributable to the event.",
    "Number of injuries directly/indirectly attributable to the event.",
    "Property damage base amount before exponent.",
    "Exponent for property damage: H=10^2, K=10^3, M=10^6, B=10^9, digits=10^digit, others treated as 1.",
    "Crop damage base amount before exponent.",
    "Exponent for crop damage with same convention as PROPDMGEXP.",
    "Multiplier derived from PROPDMGEXP.",
    "Multiplier derived from CROPDMGEXP.",
    "Standardized property damage in USD (PROPDMG * prop_mult).",
    "Standardized crop damage in USD (CROPDMG * crop_mult).",
    "Total economic loss = prop_loss + crop_loss in USD."
  )
)
kable(var_dict, caption = "Key variables and derived fields used in this analysis.")

Key variables and derived fields used in this analysis.
Variable	Type	Description
BGN_DATE	Date/Time	Date the event began.
STATE	Factor/Character	U.S. state/territory code.
EVTYPE	Factor/Character	Event type label as recorded (e.g., TORNADO, FLOOD).
FATALITIES	Numeric	Number of deaths directly/indirectly attributable to the event.
INJURIES	Numeric	Number of injuries directly/indirectly attributable to the event.
PROPDMG	Numeric	Property damage base amount before exponent.
PROPDMGEXP	Character	Exponent for property damage: H=10^2, K=10^3, M=10^6, B=10^9, digits=10^digit, others treated as 1.
CROPDMG	Numeric	Crop damage base amount before exponent.
CROPDMGEXP	Character	Exponent for crop damage with same convention as PROPDMGEXP.
prop_mult	Numeric	Multiplier derived from PROPDMGEXP.
crop_mult	Numeric	Multiplier derived from CROPDMGEXP.
prop_loss	Numeric (USD)	Standardized property damage in USD (PROPDMG * prop_mult).
crop_loss	Numeric (USD)	Standardized crop damage in USD (CROPDMG * crop_mult).
total_loss	Numeric (USD)	Total economic loss = prop_loss + crop_loss in USD.

Normalize event types (EVTYPE)

For transparency, we apply minimal standardization: trim whitespace and convert to UPPER CASE.

storm <- storm %>%
  mutate(EVTYPE = stringr::str_squish(stringr::str_to_upper(as.character(EVTYPE))))

Health outcomes

health_by_event <- storm %>%
  group_by(EVTYPE) %>%
  summarise(
    fatalities = sum(FATALITIES, na.rm = TRUE),
    injuries   = sum(INJURIES,   na.rm = TRUE),
    health_harm = fatalities + injuries,
    .groups = "drop"
  ) %>%
  filter(health_harm > 0)

top_health <- health_by_event %>%
  arrange(desc(health_harm)) %>%
  slice_head(n = 10)

knitr::kable(top_health, caption = "Top 10 event types by combined health harm (fatalities + injuries).")

Top 10 event types by combined health harm (fatalities + injuries).
EVTYPE	fatalities	injuries	health_harm
TORNADO	5633	91346	96979
EXCESSIVE HEAT	1903	6525	8428
TSTM WIND	504	6957	7461
FLOOD	470	6789	7259
LIGHTNING	816	5230	6046
HEAT	937	2100	3037
FLASH FLOOD	978	1777	2755
ICE STORM	89	1975	2064
THUNDERSTORM WIND	133	1488	1621
WINTER STORM	206	1321	1527

Economic damages (standardization)

NOAA stores damage amounts as a base and an exponent. We convert the exponent fields into numeric multipliers and compute standardized USD losses.

exp_to_multiplier <- function(x) {
  x <- toupper(trimws(as.character(x)))
  m <- rep(1, length(x))
  m[x %in% c("H")] <- 1e2
  m[x %in% c("K")] <- 1e3
  m[x %in% c("M")] <- 1e6
  m[x %in% c("B")] <- 1e9
  is_digit <- grepl("^[0-8]$", x)
  m[is_digit] <- 10 ^ as.numeric(x[is_digit])
  m
}

storm <- storm %>%
  mutate(
    prop_mult = exp_to_multiplier(PROPDMGEXP),
    crop_mult = exp_to_multiplier(CROPDMGEXP),
    prop_loss = as.numeric(PROPDMG) * prop_mult,
    crop_loss = as.numeric(CROPDMG) * crop_mult,
    total_loss = prop_loss + crop_loss
  )

econ_by_event <- storm %>%
  group_by(EVTYPE) %>%
  summarise(
    property_damage = sum(prop_loss, na.rm = TRUE),
    crop_damage     = sum(crop_loss, na.rm = TRUE),
    economic_loss   = sum(total_loss, na.rm = TRUE),
    .groups = "drop"
  ) %>%
  filter(economic_loss > 0)

top_econ <- econ_by_event %>%
  arrange(desc(economic_loss)) %>%
  slice_head(n = 10)

knitr::kable(top_econ, caption = "Top 10 event types by total economic loss (property + crop).")

Top 10 event types by total economic loss (property + crop).
EVTYPE	property_damage	crop_damage	economic_loss
FLOOD	144657709807	5661968450	150319678257
HURRICANE/TYPHOON	69305840000	2607872800	71913712800
TORNADO	56947380677	414953270	57362333947
STORM SURGE	43323536000	5000	43323541000
HAIL	15735267513	3025954473	18761221986
FLASH FLOOD	16822723979	1421317100	18244041079
DROUGHT	1046106000	13972566000	15018672000
HURRICANE	11868319010	2741910000	14610229010
RIVER FLOOD	5118945500	5029459000	10148404500
ICE STORM	3944927860	5022113500	8967041360

Results

Which event types are most harmful to population health?

health_long <- health_by_event %>%
  semi_join(top_health, by = "EVTYPE") %>%
  select(EVTYPE, fatalities, injuries) %>%
  tidyr::pivot_longer(cols = c(fatalities, injuries),
                      names_to = "metric", values_to = "count") %>%
  mutate(EVTYPE = forcats::fct_reorder(EVTYPE, count, sum))

ggplot(health_long, aes(x = EVTYPE, y = count, fill = metric)) +
  geom_col() +
  coord_flip() +
  labs(
    x = "Event Type (EVTYPE)",
    y = "People Affected",
    fill = "Health Metric",
    title = "Top 10 Event Types by Health Harm (Fatalities + Injuries)"
  ) +
  theme_minimal(base_size = 12)

Which event types have the greatest economic consequences?

econ_long <- econ_by_event %>%
  semi_join(top_econ, by = "EVTYPE") %>%
  select(EVTYPE, property_damage, crop_damage) %>%
  tidyr::pivot_longer(cols = c(property_damage, crop_damage),
                      names_to = "type", values_to = "usd") %>%
  mutate(EVTYPE = forcats::fct_reorder(EVTYPE, usd, sum))

ggplot(econ_long, aes(x = EVTYPE, y = usd/1e9, fill = type)) +
  geom_col() +
  coord_flip() +
  labs(
    x = "Event Type (EVTYPE)",
    y = "Economic Loss (Billions of USD)",
    fill = "Damage Type",
    title = "Top 10 Event Types by Economic Loss (Property + Crop)"
  ) +
  theme_minimal(base_size = 12)

Analysis of Results

Below we provide a brief, decision-oriented interpretation of the results. The analysis automatically references the computed top categories.

# Extract top leaders for dynamic text
top_health_leader <- top_health$EVTYPE[1]
top_econ_leader   <- top_econ$EVTYPE[1]
health_share_top3 <- health_by_event %>%
  arrange(desc(health_harm)) %>%
  slice_head(n = 3) %>%
  summarise(share = sum(health_harm) / sum(health_by_event$health_harm)) %>%
  pull(share)

econ_share_top3 <- econ_by_event %>%
  arrange(desc(economic_loss)) %>%
  slice_head(n = 3) %>%
  summarise(share = sum(economic_loss) / sum(econ_by_event$economic_loss)) %>%
  pull(share)

cat(sprintf("**Health impacts.** `%s` is the single most harmful event type when combining fatalities and injuries. The top three event types account for roughly %.1f%% of total health harm, indicating a highly skewed risk distribution.\n\n",
            top_health_leader, 100*health_share_top3))

Health impacts. TORNADO is the single most harmful event type when combining fatalities and injuries. The top three event types account for roughly 72.5% of total health harm, indicating a highly skewed risk distribution.

cat(sprintf("**Economic impacts.** `%s` yields the largest total economic losses (property + crop). Similarly, the top three event types account for about %.1f%% of total losses, underscoring the value of targeting mitigation resources to a small set of high-impact hazards.\n\n",
            top_econ_leader, 100*econ_share_top3))

Economic impacts. FLOOD yields the largest total economic losses (property + crop). Similarly, the top three event types account for about 58.6% of total losses, underscoring the value of targeting mitigation resources to a small set of high-impact hazards.

# Additional insights: injury vs fatality composition, property vs crop composition
fatality_leader <- health_by_event %>% arrange(desc(fatalities)) %>% slice(1) %>% pull(EVTYPE)
injury_leader   <- health_by_event %>% arrange(desc(injuries))   %>% slice(1) %>% pull(EVTYPE)
prop_leader     <- econ_by_event    %>% arrange(desc(property_damage)) %>% slice(1) %>% pull(EVTYPE)
crop_leader     <- econ_by_event    %>% arrange(desc(crop_damage))     %>% slice(1) %>% pull(EVTYPE)

cat(sprintf("**Composition.** Fatalities are led by `%s`, while injuries are led by `%s`. On the economic side, property damages are highest for `%s`, and crop damages are dominated by `%s`.\n\n",
            fatality_leader, injury_leader, prop_leader, crop_leader))

Composition. Fatalities are led by TORNADO, while injuries are led by TORNADO. On the economic side, property damages are highest for FLOOD, and crop damages are dominated by DROUGHT.

cat("**Policy takeaway.** A risk-informed preparedness plan should emphasize the leading health-impact hazards (e.g., warning systems, shelter access) and the leading economic-impact hazards (e.g., flood and wind-resistant infrastructure, surge barriers, crop protection and drought management). Geographic tailoring matters, as leading hazards vary by state and season.\n")

Policy takeaway. A risk-informed preparedness plan should emphasize the leading health-impact hazards (e.g., warning systems, shelter access) and the leading economic-impact hazards (e.g., flood and wind-resistant infrastructure, surge barriers, crop protection and drought management). Geographic tailoring matters, as leading hazards vary by state and season.

Conclusions

A small number of event types explain a large share of total harm. Focusing preparedness and mitigation on these yields the largest marginal benefits.
Health risk is typically dominated by high-wind and heat-related hazards; economic losses are often dominated by hydrologic hazards (floods/surge) and major cyclones.
Loss composition differs: some hazards drive fatalities more than injuries, while others primarily cause property vs crop losses. Resource allocation should reflect these differences.
Limitations: EVTYPE labeling is noisy; our light normalization may leave some near-duplicates. Damage exponent coding can contain ambiguities; we follow a standard mapping. Finer reclassification (e.g., NOAA’s 48-category standard) could refine rankings but does not materially change the top-tier conclusions.

Reproducibility Notes

Source file: raw repdata_data_StormData.csv (no external pre-processing).
cache=TRUE on heavy chunks to speed re-runs.
Exactly two figures.
Knit to HTML in RStudio Desktop; optionally publish to RPubs.

Appendix: Session Info

sessionInfo()

## R version 4.4.1 (2024-06-14 ucrt)
## Platform: x86_64-w64-mingw32/x64
## Running under: Windows 11 x64 (build 26100)
## 
## Matrix products: default
## 
## 
## locale:
## [1] LC_COLLATE=Spanish_Mexico.utf8  LC_CTYPE=Spanish_Mexico.utf8   
## [3] LC_MONETARY=Spanish_Mexico.utf8 LC_NUMERIC=C                   
## [5] LC_TIME=Spanish_Mexico.utf8    
## 
## time zone: America/Guayaquil
## tzcode source: internal
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] knitr_1.48    tidyr_1.3.1   forcats_1.0.0 readr_2.1.5   stringr_1.5.1
## [6] ggplot2_3.5.1 dplyr_1.1.4  
## 
## loaded via a namespace (and not attached):
##  [1] bit_4.0.5         gtable_0.3.5      jsonlite_1.8.8    highr_0.11       
##  [5] crayon_1.5.3      compiler_4.4.1    tidyselect_1.2.1  parallel_4.4.1   
##  [9] jquerylib_0.1.4   scales_1.3.0      yaml_2.3.9        fastmap_1.2.0    
## [13] R6_2.5.1          labeling_0.4.3    generics_0.1.3    tibble_3.2.1     
## [17] munsell_0.5.1     bslib_0.7.0       pillar_1.9.0      tzdb_0.4.0       
## [21] rlang_1.1.4       utf8_1.2.4        cachem_1.1.0      stringi_1.8.4    
## [25] xfun_0.45         sass_0.4.9        bit64_4.0.5       cli_3.6.3        
## [29] withr_3.0.0       magrittr_2.0.3    digest_0.6.36     grid_4.4.1       
## [33] vroom_1.6.5       rstudioapi_0.16.0 hms_1.1.3         lifecycle_1.0.4  
## [37] vctrs_0.6.5       evaluate_0.24.0   glue_1.7.0        farver_2.1.2     
## [41] codetools_0.2-20  fansi_1.0.6       colorspace_2.1-0  purrr_1.0.2      
## [45] rmarkdown_2.27    tools_4.4.1       pkgconfig_2.0.3   htmltools_0.5.8.1

NOAA Storm Events: Health and Economic Impacts in the United States

Daniel Benavides Delgado

28 de septiembre de 2025