NOAA Storm Events: Health Impacts and Economic Consequences (1996–2011)
Roger Sala Fradera
2025-08-15
knitr::opts_chunk$set(echo = TRUE, message = FALSE, warning = FALSE, cache = TRUE)1 Synopsis
This analysis uses the U.S. NOAA Storm Events database to identify
which severe weather events are (1) most harmful to population health
and (2) most costly economically. To improve comparability across event
types, we restrict to 1996–2011, when NOAA standardized
the official event list. We clean variable names, parse dates, convert
property/crop damage exponent codes (H, K, M, B and digits)
to USD, and standardize event labels to NOAA-like categories using
robust regular expressions.
Findings: For health impacts, Tornado produces the
highest combined injuries+fatalities (~22.2k), while Excessive Heat
leads in fatalities (1.8k). For economic losses, Flood accounts for
$149.5B, followed by Hurricane (Typhoon) ($87.1B) and Storm Surge/Tide
($47.8B)). We present two figures (≤3 allowed) and top-10 tables to
support these conclusions.
2 Data Processing (what we did and why)
Load & clean column names. We read the
compressed CSV directly and applied janitor::clean_names()
to obtain consistent, snake_case names. This avoids downstream naming
issues and makes code clearer.
Parse dates. bgn_date was parsed with
lubridate::parse_date_time() and converted to
Date.
Restrict to 1996+. We filtered to events with
bgn_date >= 1996-01-01. NOAA’s event definitions and
reporting coverage stabilize from 1996 onward, reducing name noise and
biases.
Convert damage exponents to USD. We mapped
propdmgexp/cropdmgexp to numeric multipliers:
H=1e2, K=1e3, M=1e6,
B=1e9, digits 0–8 as 10^digit.
Ambiguous codes (+/?/- or blank) were conservatively
treated as multiplier 1. We then computed
prop_dmg_usd, crop_dmg_usd, and
total_dmg_usd.
Standardize event types. Free-text
evtype values were normalized (uppercased, punctuation
removed, whitespace squished) and matched with regex to NOAA-like
categories (e.g., TSTM WIND →
THUNDERSTORM WIND, “HURRICANE”/“TYPHOON” →
HURRICANE (TYPHOON)). This collapses common variants and
typographical noise.
Diagnostics. We compared the number of distinct raw
vs. standardized event types and listed the most frequent labels that
remained unchanged (top_unmapped) to check coverage.
Aggregate metrics
Health: per standardized event, we summed
fatalities, injuries, and
health_total = fatalities + injuries.
Economy: per event, we summed
prop_dmg_usd, crop_dmg_usd, and
econ_total = prop + crop.
Output tables & figures. We created top-10 tables and two bar charts (stacked) that show component contributions (injuries vs. fatalities; property vs. crop), meeting the assignment’s ≤3-figure rule.
3 Methods & Assumptions
Scope: U.S. events, 1996–2011, nominal USD as recorded (no inflation adjustment).
Exponent handling: Ambiguous/blank exponents treated
as multiplier 1 (conservative).
Labeling: Event standardization uses explicit regex rules covering the vast majority of common variants.
Missing values: Damage and count fields are
aggregated with na.rm = TRUE.
# Core tidyverse + dates + tables
library(tidyverse)
library(lubridate)
library(knitr)
library(scales) # nice axis formatting
library(ggrepel)
library(lubridate)
library(janitor)
# Repro note
set.seed(123)4 glimpse(df_raw)
What it shows.
A structural snapshot of the raw NOAA file: 902,297
rows and 37 columns spanning 1950–2011. Key
variables include:
evtype (event label), fatalities,
injuries (health impacts),
propdmg, propdmgexp, cropdmg,
cropdmgexp (economic impacts with exponent codes),
bgn_date (event start timestamp), location fields
(state_2, countyname, lat/long), and an ID
refnum.
Why it matters.
Confirms the file loaded correctly, the columns we need exist, and types are plausible before we transform anything.
# Path to local file (edit if needed)
local_file <- "repdata_data_StormData.csv.bz2"
# Coursera dataset URL (same file, compressed)
data_url <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
if (!file.exists(local_file)) {
message("Local file not found; downloading...")
download.file(data_url, destfile = local_file, mode = "wb", quiet = TRUE)
}
# Read raw (compressed) CSV
df_raw <- readr::read_csv(local_file, show_col_types = FALSE) %>%
janitor::clean_names()
glimpse(df_raw)## Rows: 902,297
## Columns: 37
## $ state <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ bgn_date <chr> "4/18/1950 0:00:00", "4/18/1950 0:00:00", "2/20/1951 0:00:…
## $ bgn_time <chr> "0130", "0145", "1600", "0900", "1500", "2000", "0100", "0…
## $ time_zone <chr> "CST", "CST", "CST", "CST", "CST", "CST", "CST", "CST", "C…
## $ county <dbl> 97, 3, 57, 89, 43, 77, 9, 123, 125, 57, 43, 9, 73, 49, 107…
## $ countyname <chr> "MOBILE", "BALDWIN", "FAYETTE", "MADISON", "CULLMAN", "LAU…
## $ state_2 <chr> "AL", "AL", "AL", "AL", "AL", "AL", "AL", "AL", "AL", "AL"…
## $ evtype <chr> "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TO…
## $ bgn_range <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ bgn_azi <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ bgn_locati <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ end_date <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ end_time <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ county_end <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ countyendn <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ end_range <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ end_azi <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ end_locati <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ length <dbl> 14.0, 2.0, 0.1, 0.0, 0.0, 1.5, 1.5, 0.0, 3.3, 2.3, 1.3, 4.…
## $ width <dbl> 100, 150, 123, 100, 150, 177, 33, 33, 100, 100, 400, 400, …
## $ f <dbl> 3, 2, 2, 2, 2, 2, 2, 1, 3, 3, 1, 1, 3, 3, 3, 4, 1, 1, 1, 1…
## $ mag <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ fatalities <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 4, 0, 0, 0, 0…
## $ injuries <dbl> 15, 0, 2, 2, 2, 6, 1, 0, 14, 0, 3, 3, 26, 12, 6, 50, 2, 0,…
## $ propdmg <dbl> 25.0, 2.5, 25.0, 2.5, 2.5, 2.5, 2.5, 2.5, 25.0, 25.0, 2.5,…
## $ propdmgexp <chr> "K", "K", "K", "K", "K", "K", "K", "K", "K", "K", "M", "M"…
## $ cropdmg <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ cropdmgexp <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ wfo <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ stateoffic <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ zonenames <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ latitude <dbl> 3040, 3042, 3340, 3458, 3412, 3450, 3405, 3255, 3334, 3336…
## $ longitude <dbl> 8812, 8755, 8742, 8626, 8642, 8748, 8631, 8558, 8740, 8738…
## $ latitude_e <dbl> 3051, 0, 0, 0, 0, 0, 0, 0, 3336, 3337, 3402, 3404, 0, 3432…
## $ longitude_2 <dbl> 8806, 0, 0, 0, 0, 0, 0, 0, 8738, 8737, 8644, 8640, 0, 8540…
## $ remarks <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
## $ refnum <dbl> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,…5 Diagnostics: distinct event labels before vs. after standardization
Output:
Our regex normalization reduces 508 distinct
free-text EVTYPE strings to 230
standardized categories.
Why it matters.
This consolidation removes typos/variants (e.g., “TSTM WIND” vs
“THUNDERSTORM WIND”), making downstream ranking more reliable. We didn’t
force to the official 48 NOAA categories, but we eliminated a
lot of noise while keeping detail.
Conclusion.
Standardization substantially cleans the label space (≈55% reduction), improving comparability across events.
# 1) Parse start date (BGN_DATE) to Date
df_dates <- df_raw %>%
mutate(
bgn_date = suppressWarnings(parse_date_time(bgn_date, orders = c("mdY HMS", "mdY HM", "mdY")))
) %>%
mutate(bgn_date = as_date(bgn_date))
# 2) Restrict to NOAA standardized period (1996+)
df <- df_dates %>%
filter(bgn_date >= as.Date("1996-01-01"))
# 3) Map exponent codes to numeric multipliers
exp_to_multiplier <- function(x) {
x <- toupper(trimws(as.character(x)))
dplyr::case_when(
x %in% c("", NA) ~ 1,
x %in% c("+", "-", "?") ~ 1, # ambiguous -> neutral
x == "H" ~ 1e2,
x == "K" ~ 1e3,
x == "M" ~ 1e6,
x == "B" ~ 1e9,
x %in% as.character(0:8) ~ 10^(as.numeric(x)),
TRUE ~ 1
)
}
# 4) Standardize EVTYPE to NOAA-like categories with robust regex rules
normalize_evtype <- function(x) {
x %>%
stringr::str_to_upper() %>%
stringr::str_replace_all("[^A-Z ]", " ") %>%
stringr::str_squish()
}
standardize_evtype <- function(x) {
nx <- normalize_evtype(x)
dplyr::case_when(
stringr::str_detect(nx, "EXCESSIVE HEAT") ~ "EXCESSIVE HEAT",
stringr::str_detect(nx, "\\bHEAT\\b|RECORD HIGH|WARM") ~ "HEAT",
stringr::str_detect(nx, "TORN") ~ "TORNADO",
stringr::str_detect(nx, "TSTM.*WIND|THUNDER.*WIND") ~ "THUNDERSTORM WIND",
stringr::str_detect(nx, "\\bHAIL\\b|HAIL STORM") ~ "HAIL",
stringr::str_detect(nx, "FLASH FLOOD") ~ "FLASH FLOOD",
stringr::str_detect(nx, "\\bFLOOD\\b|RIVER FLOOD|URBAN.*FLOOD|COASTAL FLOOD") ~ "FLOOD",
stringr::str_detect(nx, "HURRICANE|TYPHOON") ~ "HURRICANE (TYPHOON)",
stringr::str_detect(nx, "TROPICAL STORM") ~ "TROPICAL STORM",
stringr::str_detect(nx, "WINTER STORM") ~ "WINTER STORM",
stringr::str_detect(nx, "WINTER WEATHER") ~ "WINTER WEATHER",
stringr::str_detect(nx, "LIGHTNING") ~ "LIGHTNING",
stringr::str_detect(nx, "HIGH WIND|STRONG WIND") ~ "HIGH WIND",
stringr::str_detect(nx, "WILDFIRE|WILD FIRE|FOREST FIRE") ~ "WILDFIRE",
stringr::str_detect(nx, "DROUGHT") ~ "DROUGHT",
stringr::str_detect(nx, "STORM SURGE|TIDE") ~ "STORM SURGE/TIDE",
stringr::str_detect(nx, "HEAVY RAIN|TORRENTIAL RAIN|RAIN STORM") ~ "HEAVY RAIN",
stringr::str_detect(nx, "HEAVY SNOW|LAKE.*SNOW") ~ "HEAVY SNOW",
stringr::str_detect(nx, "BLIZZARD") ~ "BLIZZARD",
stringr::str_detect(nx, "ICE STORM") ~ "ICE STORM",
stringr::str_detect(nx, "COLD|RECORD COLD|FREEZ|LOW TEMP|WIND CHILL") ~ "COLD/WIND CHILL",
stringr::str_detect(nx, "RIP CURRENT") ~ "RIP CURRENT",
stringr::str_detect(nx, "AVALAN") ~ "AVALANCHE",
stringr::str_detect(nx, "\\bSLEET\\b") ~ "SLEET",
stringr::str_detect(nx, "FUNNEL") ~ "FUNNEL CLOUD",
stringr::str_detect(nx, "WATERSPOUT") ~ "WATERSPOUT",
stringr::str_detect(nx, "SURF|SWELL") ~ "HIGH SURF",
stringr::str_detect(nx, "TSUNAMI") ~ "TSUNAMI",
stringr::str_detect(nx, "VOLCAN") ~ "VOLCANIC ASH",
TRUE ~ nx
)
}6 top_unmapped (most frequent labels that were already “clean”)
Top entries (counts of rows that didn’t change under our
mapping):
HAIL (207,716), THUNDERSTORM WIND (81,403),
FLASH FLOOD (51,000), FLOOD (24,248),
TORNADO (23,154), HIGH WIND (19,909),
HEAVY SNOW (14,000), LIGHTNING (13,204),
HEAVY RAIN (11,528), WINTER STORM (11,317),
WINTER WEATHER (6,987), FUNNEL CLOUD (6,063),
URBAN SML STREAM FLD (3,392),
WATERSPOUT (3,391), WILDFIRE (2,732).
What it shows.
The largest share of records already used standard labels (e.g.,
HAIL, TORNADO), so our mapping didn’t need to alter them. One notable
leftover is “URBAN SML STREAM FLD” (abbrev. “FLD”),
which our regex intentionally didn’t force to FLOOD.
Why it matters.
Confirms we’re not over-rewriting good labels and highlights any edge
cases. If desired, we could map “URBAN SML STREAM FLD” into
FLOOD to further reduce fragmentation (results won’t
materially change the top ranks).
Conclusion.
Most high-volume event labels were already standardized; only a few
niche/abbreviated labels remain candidates for further
consolidation.
df2 <- df %>%
mutate(
prop_mult = exp_to_multiplier(propdmgexp),
crop_mult = exp_to_multiplier(cropdmgexp),
prop_dmg_usd = coalesce(propdmg, 0) * prop_mult,
crop_dmg_usd = coalesce(cropdmg, 0) * crop_mult,
total_dmg_usd = prop_dmg_usd + crop_dmg_usd,
evtype_std = standardize_evtype(evtype)
)
# Diagnostics (optional to inspect standardization coverage)
df2 %>%
summarise(
events_raw = n_distinct(evtype),
events_std = n_distinct(evtype_std)
)## # A tibble: 1 × 2
## events_raw events_std
## <int> <int>
## 1 508 230top_unmapped <- df2 %>%
filter(evtype_std == normalize_evtype(evtype)) %>%
count(evtype_std, sort = TRUE) %>%
slice_head(n = 15)
top_unmapped## # A tibble: 15 × 2
## evtype_std n
## <chr> <int>
## 1 HAIL 207716
## 2 THUNDERSTORM WIND 81403
## 3 FLASH FLOOD 51000
## 4 FLOOD 24248
## 5 TORNADO 23154
## 6 HIGH WIND 19909
## 7 HEAVY SNOW 14000
## 8 LIGHTNING 13204
## 9 HEAVY RAIN 11528
## 10 WINTER STORM 11317
## 11 WINTER WEATHER 6987
## 12 FUNNEL CLOUD 6063
## 13 URBAN SML STREAM FLD 3392
## 14 WATERSPOUT 3391
## 15 WILDFIRE 2732# Health
health_by_event <- df2 %>%
group_by(evtype_std) %>%
summarise(
fatalities = sum(fatalities, na.rm = TRUE),
injuries = sum(injuries, na.rm = TRUE),
health_total = fatalities + injuries,
.groups = "drop"
) %>%
arrange(desc(health_total))
health_top10 <- health_by_event %>% slice_head(n = 10)
# Economic
econ_by_event <- df2 %>%
group_by(evtype_std) %>%
summarise(
prop_usd = sum(prop_dmg_usd, na.rm = TRUE),
crop_usd = sum(crop_dmg_usd, na.rm = TRUE),
econ_total = prop_usd + crop_usd,
.groups = "drop"
) %>%
arrange(desc(econ_total))
econ_top10 <- econ_by_event %>% slice_head(n = 10)
# Quick tables
health_top10## # A tibble: 10 × 4
## evtype_std fatalities injuries health_total
## <chr> <dbl> <dbl> <dbl>
## 1 TORNADO 1511 20667 22178
## 2 EXCESSIVE HEAT 1797 6391 8188
## 3 FLOOD 422 6766 7188
## 4 THUNDERSTORM WIND 397 5163 5560
## 5 LIGHTNING 651 4141 4792
## 6 FLASH FLOOD 887 1674 2561
## 7 HIGH WIND 360 1405 1765
## 8 HEAT 239 1311 1550
## 9 WILDFIRE 87 1456 1543
## 10 WINTER STORM 191 1292 1483econ_top10## # A tibble: 10 × 4
## evtype_std prop_usd crop_usd econ_total
## <chr> <dbl> <dbl> <dbl>
## 1 FLOOD 144485251110 5004673400 149489924510
## 2 HURRICANE (TYPHOON) 81718889010 5350107800 87068996810
## 3 STORM SURGE/TIDE 47844469000 855000 47845324000
## 4 TORNADO 24616945710 283425010 24900370720
## 5 HAIL 14595237420 2496822450 17092059870
## 6 FLASH FLOOD 15222268910 1334901700 16557170610
## 7 DROUGHT 1046101000 13367566000 14413667000
## 8 THUNDERSTORM WIND 7919453280 1016992600 8936445880
## 9 TROPICAL STORM 7642475550 677711000 8320186550
## 10 WILDFIRE 7760449500 402255130 81627046307
health_top10 (Top 10 event types by injuries +
fatalities, 1996–2011)
What it shows.
Tornado is the top driver of combined health impacts (~22.2k people), dominated by injuries.
Excessive Heat has the highest fatalities (~1,797) even though its injury count is lower than Tornado.
Flood, Thunderstorm Wind, and Lightning are also major contributors.
Why it matters.
For preparedness planning, the relative burden differs by mechanism:
wind/tornado events create many injuries at once; heat events produce a
sustained fatality burden.
Conclusions.
Most harmful overall: Tornado (injuries-heavy).
Most lethal per se: Excessive Heat.
Other high-impact hazards: Flood, Thunderstorm Wind, Lightning.
health_long <- health_top10 %>%
select(evtype_std, fatalities, injuries) %>%
pivot_longer(-evtype_std, names_to = "type", values_to = "count")
health_long %>%
mutate(evtype_std = fct_reorder(evtype_std, count, .fun = sum)) %>%
ggplot(aes(x = evtype_std, y = count, fill = type)) +
geom_col() +
coord_flip() +
labs(
title = "Top 10 Event Types by Health Impact (1996–2011)",
x = NULL, y = "People Affected (injuries + fatalities)",
fill = NULL,
caption = "Source: NOAA Storm Events (restricted to 1996+)"
) +
theme_minimal(base_size = 12)
health_by_event %>%
slice_head(n = 10)## # A tibble: 10 × 4
## evtype_std fatalities injuries health_total
## <chr> <dbl> <dbl> <dbl>
## 1 TORNADO 1511 20667 22178
## 2 EXCESSIVE HEAT 1797 6391 8188
## 3 FLOOD 422 6766 7188
## 4 THUNDERSTORM WIND 397 5163 5560
## 5 LIGHTNING 651 4141 4792
## 6 FLASH FLOOD 887 1674 2561
## 7 HIGH WIND 360 1405 1765
## 8 HEAT 239 1311 1550
## 9 WILDFIRE 87 1456 1543
## 10 WINTER STORM 191 1292 14838 econ_top10
(Top 10 event types by property + crop damage,
1996–2011)
What it shows.
Flood is the dominant economic hazard (~$149.5B), mainly property damage.
Hurricane (Typhoon) (~$87.1B) and Storm Surge/Tide (~$47.8B) are also property-heavy.
Drought is the standout crop-driven hazard (~$13.4B crop vs. $1.0B property).
Tornado, Hail, and Flash Flood round out the top with sizable totals.
Why it matters.
Different hazards threaten different asset classes: surge/hurricanes
devastate infrastructure and housing, while drought predominantly
impacts agriculture.
Conclusions.
Greatest total economic losses: Flood, then Hurricane (Typhoon), then Storm Surge/Tide.
Crop-centric loss pattern: Drought.
Wind/hail hazards: material but below the flood/hurricane tier.
econ_long <- econ_top10 %>%
select(evtype_std, prop_usd, crop_usd) %>%
pivot_longer(-evtype_std, names_to = "type", values_to = "usd")
econ_long %>%
mutate(evtype_std = fct_reorder(evtype_std, usd, .fun = sum)) %>%
ggplot(aes(x = evtype_std, y = usd, fill = type)) +
geom_col() +
coord_flip() +
scale_y_continuous(labels = label_dollar()) +
labs(
title = "Top 10 Event Types by Economic Damage (1996–2011)",
x = NULL, y = "USD (property + crop)",
fill = NULL,
caption = "Damage exponents converted: H=1e2, K=1e3, M=1e6, B=1e9"
) +
theme_minimal(base_size = 12)
econ_by_event %>%
slice_head(n = 10) %>%
mutate(across(ends_with("_usd") | econ_total, ~round(.x)))## # A tibble: 10 × 4
## evtype_std prop_usd crop_usd econ_total
## <chr> <dbl> <dbl> <dbl>
## 1 FLOOD 144485251110 5004673400 149489924510
## 2 HURRICANE (TYPHOON) 81718889010 5350107800 87068996810
## 3 STORM SURGE/TIDE 47844469000 855000 47845324000
## 4 TORNADO 24616945710 283425010 24900370720
## 5 HAIL 14595237420 2496822450 17092059870
## 6 FLASH FLOOD 15222268910 1334901700 16557170610
## 7 DROUGHT 1046101000 13367566000 14413667000
## 8 THUNDERSTORM WIND 7919453280 1016992600 8936445880
## 9 TROPICAL STORM 7642475550 677711000 8320186550
## 10 WILDFIRE 7760449500 402255130 8162704630sessionInfo()## R version 4.5.0 (2025-04-11 ucrt)
## Platform: x86_64-w64-mingw32/x64
## Running under: Windows 11 x64 (build 26100)
##
## Matrix products: default
## LAPACK version 3.12.1
##
## locale:
## [1] LC_COLLATE=Spanish_Spain.utf8 LC_CTYPE=Spanish_Spain.utf8
## [3] LC_MONETARY=Spanish_Spain.utf8 LC_NUMERIC=C
## [5] LC_TIME=Spanish_Spain.utf8
##
## time zone: Europe/Madrid
## tzcode source: internal
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] janitor_2.2.1 ggrepel_0.9.6 scales_1.4.0 knitr_1.50
## [5] lubridate_1.9.4 forcats_1.0.0 stringr_1.5.1 dplyr_1.1.4
## [9] purrr_1.1.0 readr_2.1.5 tidyr_1.3.1 tibble_3.3.0
## [13] ggplot2_3.5.2 tidyverse_2.0.0
##
## loaded via a namespace (and not attached):
## [1] bit_4.6.0 gtable_0.3.6 jsonlite_2.0.0 crayon_1.5.3
## [5] compiler_4.5.0 Rcpp_1.1.0 tidyselect_1.2.1 parallel_4.5.0
## [9] snakecase_0.11.1 jquerylib_0.1.4 yaml_2.3.10 fastmap_1.2.0
## [13] R6_2.6.1 generics_0.1.4 bslib_0.9.0 pillar_1.11.0
## [17] RColorBrewer_1.1-3 tzdb_0.5.0 rlang_1.1.6 cachem_1.1.0
## [21] stringi_1.8.7 xfun_0.52 sass_0.4.10 bit64_4.6.0-1
## [25] timechange_0.3.0 cli_3.6.5 withr_3.0.2 magrittr_2.0.3
## [29] digest_0.6.37 grid_4.5.0 vroom_1.6.5 rstudioapi_0.17.1
## [33] hms_1.1.3 lifecycle_1.0.4 vctrs_0.6.5 evaluate_1.0.4
## [37] glue_1.8.0 farver_2.1.2 codetools_0.2-20 rmarkdown_2.29
## [41] tools_4.5.0 pkgconfig_2.0.3 htmltools_0.5.8.19 Results (what the outputs show)
Population health (1996–2011). The stacked bar chart of the top 10 event types by total injuries+fatalities shows:
Tornado has the highest combined toll (~22,178), driven mostly by injuries.
Excessive Heat has the highest fatalities (~1,797) despite fewer injuries than Tornado.
Flood, Thunderstorm Wind, and Lightning also contribute materially to health impacts.
Economic consequences (1996–2011). The stacked bar chart of top 10 event types by total property+crop damage indicates:
Flood causes the greatest losses (~$149.5B), mostly property damage.
(Typhoon) ($87.1B) and Storm Surge/Tide ($47.8B) are next, largely property-driven.
Drought stands out for crop losses (~$13.4B crop vs. ~$1.0B property).
Tornado ($24.9B) and Hail ($17.1B) also rank highly.
These tables in your output support the statements above:
Health top-10 shows Tornado (≈22,178), Excessive Heat (≈8,188 total; fatalities ≈1,797), Flood (≈7,188), etc.
Economic top-10 shows Flood ($149.49B), Hurricane (Typhoon) ($87.07B), Storm Surge/Tide ($47.85B), Tornado ($24.90B), Hail ($17.09B), Flash Flood ($16.56B), Drought (~$14.41B), etc.
10 Limitations
Event coding noise: While the regex mapping is robust, free-text EVTYPE introduces residual heterogeneity.
Reporting changes over time: Even post-1996, reporting practices can evolve; restricting to standardized years mitigates but doesn’t eliminate this.
No inflation adjustment: Dollar amounts are nominal; adjusting to real terms would shift magnitudes but not usually the ranking of top categories.
Outliers dominate: A few historic floods/hurricanes heavily influence totals.
11 Reproducibility
All steps are performed inside the document from the raw compressed
CSV; code chunks use echo=TRUE and cache=TRUE.
The final chunk shows sessionInfo() for package
versions.