Risk Analysis of Storm Events Across the United States
Summary
This report presents an examination of the dangers posed by storm occurrences throughout the United States. Information for this examination is sourced from the Storm Data database by NOAA. Risk arising from storm events is interpreted as the correlation between the seriousness of the outcomes resulting from the event and the probability of the event taking place. Results of an event are defined as the impact on public health (deaths and injuries) and as financial repercussions (damage to property and crops).
Data
The study utilizes the Storm Data database provided by the National Oceanic & Atmospheric Administration, NOAA. The primary database is accessible here, with supporting documentation available here.The data spans from January 1950 to November 2011 and includes details such as the date and time of the event, duration, location specifics (state, county, latitude, and longitude), event classification, severity, counts of casualties, property damage costs, crop damage costs, responsible state office, and additional notes. Storm events are classified into 48 distinct groups, as outlined in table 1 on page 5 of the documentation. Detailed descriptions and classification criteria for each storm event type can be found in section 7 of the documentation.
Data Processing
Initial cleanup
This section illustrates the complete process of data cleaning along with the corresponding R code, which is provided in its entirety and can be executed to replicate the analysis
First, load all necessary libraries for the analysis.
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(tidyr)
library(lubridate)##
## Attaching package: 'lubridate'## The following objects are masked from 'package:base':
##
## date, intersect, setdiff, unionlibrary(stringr)
#library(ggplot2)Load the data into R
setwd("C:/Users/Somto/Documents/Git/data")
stormdata <- read.csv("repdata_data_StormData.csv.bz2", stringsAsFactors = FALSE)An initial investigation of the data
names(stormdata)## [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"glimpse(stormdata)## 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 0:00:0…
## $ BGN_TIME <chr> "0130", "0145", "1600", "0900", "1500", "2000", "0100", "09…
## $ TIME_ZONE <chr> "CST", "CST", "CST", "CST", "CST", "CST", "CST", "CST", "CS…
## $ COUNTY <dbl> 97, 3, 57, 89, 43, 77, 9, 123, 125, 57, 43, 9, 73, 49, 107,…
## $ COUNTYNAME <chr> "MOBILE", "BALDWIN", "FAYETTE", "MADISON", "CULLMAN", "LAUD…
## $ STATE <chr> "AL", "AL", "AL", "AL", "AL", "AL", "AL", "AL", "AL", "AL",…
## $ EVTYPE <chr> "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TOR…
## $ 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> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ BGN_LOCATI <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ END_DATE <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ END_TIME <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ 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> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ END_LOCATI <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ LENGTH <dbl> 14.0, 2.0, 0.1, 0.0, 0.0, 1.5, 1.5, 0.0, 3.3, 2.3, 1.3, 4.7…
## $ WIDTH <dbl> 100, 150, 123, 100, 150, 177, 33, 33, 100, 100, 400, 400, 2…
## $ F <int> 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> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ WFO <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ STATEOFFIC <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ ZONENAMES <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ 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_ <dbl> 8806, 0, 0, 0, 0, 0, 0, 0, 8738, 8737, 8644, 8640, 0, 8540,…
## $ REMARKS <chr> "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",…
## $ REFNUM <dbl> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, …# Calculate the number of unique event types in the 'stormdata' dataset
num_unique_event_types <- length(unique(stormdata$EVTYPE))
num_unique_event_types## [1] 985# Summarize the 'FATALITIES' column in the 'stormdata' dataset
summary(stormdata$FATALITIES)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0000 0.0000 0.0000 0.0168 0.0000 583.0000# Summarize the 'PROPDMG' column in the 'stormdata' dataset
summary(stormdata$PROPDMG)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.00 0.00 0.00 12.06 0.50 5000.00# Summarize the 'PROPDMGEXP' column in the 'stormdata' dataset
summary(stormdata$PROPDMGEXP)## Length Class Mode
## 902297 character character# Summarize the 'CROPDMG' column in the 'stormdata' dataset
summary(stormdata$CROPDMG)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.000 0.000 0.000 1.527 0.000 990.000# Summarize the 'CROPDMGEXP' column in the 'stormdata' dataset
summary(stormdata$CROPDMGEXP)## Length Class Mode
## 902297 character characterInitial investigation revealed that fatalities and injuries are presented as whole numbers, while property and crop damage variables are represented using exponents indicated by the _EXP columns. This issue will be resolved at a later stage
Start with an initial subsetting of the data, and some basic cleaning.
stormdata2 <- stormdata %>%
select(BGN_DATE, STATE, EVTYPE, FATALITIES, INJURIES, PROPDMG, PROPDMGEXP, CROPDMG, CROPDMGEXP)
# Dates into POSIX
stormdata2$BGN_DATE <- mdy_hms(stormdata2$BGN_DATE)
# Variable names in lowercase
names(stormdata2) <- tolower(names(stormdata2))
str(stormdata2)## 'data.frame': 902297 obs. of 9 variables:
## $ bgn_date : POSIXct, format: "1950-04-18" "1950-04-18" ...
## $ state : chr "AL" "AL" "AL" "AL" ...
## $ evtype : chr "TORNADO" "TORNADO" "TORNADO" "TORNADO" ...
## $ fatalities: num 0 0 0 0 0 0 0 0 1 0 ...
## $ injuries : num 15 0 2 2 2 6 1 0 14 0 ...
## $ propdmg : num 25 2.5 25 2.5 2.5 2.5 2.5 2.5 25 25 ...
## $ propdmgexp: chr "K" "K" "K" "K" ...
## $ cropdmg : num 0 0 0 0 0 0 0 0 0 0 ...
## $ cropdmgexp: chr "" "" "" "" ...Convert property and crop damage values into standard numerical format. Retain only the encoded exponents: ““,”H”, “T”, “M”, and “B”. Treat “0”, “-”, and “” as equal. Disregard any other values as their significance is unclear and not defined in the documentation.
stormdata2$propdmgexp <- stormdata2$propdmgexp %>%
toupper %>%
str_replace_all("-","") %>%
str_replace_all("0","")
stormdata2$cropdmgexp <- stormdata2$cropdmgexp %>%
toupper %>%
str_replace_all("0","")
validdmgexp <- c("", "H", "K", "M", "B")
stormdata3 <- stormdata2 %>%
filter(propdmgexp %in% validdmgexp) %>%
filter(cropdmgexp %in% validdmgexp) %>%
mutate(propdmgtr = 0, cropdmgtr = 0)
# Now, multiply the values in the dmg columns by the appropriate magnitude
expmagnitude = c(1, 100, 1000, 1000000, 1000000000)
for(i in seq_along(validdmgexp)) {
pexpind <- which(stormdata3$propdmgexp == validdmgexp[i])
cexpind <- which(stormdata3$cropdmgexp == validdmgexp[i])
stormdata3$propdmgtr[pexpind] <- stormdata3$propdmg[pexpind] * expmagnitude[i]
stormdata3$cropdmgtr[cexpind] <- stormdata3$cropdmg[cexpind] * expmagnitude[i]
}
# Number of entries removed
dim(stormdata2)[1] - dim(stormdata3)[1]## [1] 105(dim(stormdata2)[1] - dim(stormdata3)[1])/dim(stormdata2)[1] *100## [1] 0.01163697Cleaning the event type variable
stormdata3$evtype <- str_trim(tolower(stormdata3$evtype))
stormdata4 <- mutate(stormdata3, cleanevtype = "")Clean obvious misspellings.
stormdata4$cleanevtype <- stormdata4$evtype
stormdata4$cleanevtype <- stormdata4$cleanevtype %>%
str_replace_all("\\s+", " ") %>% # clear all double spaces
str_replace_all("\\bav[a-z]*e\\b", "avalanche") %>% # avalanches
str_replace_all("cstl", "coastal") %>%
str_replace_all("erosin", "erosion") %>%
str_replace_all("coastal[a-z]", "coastal ") %>%
str_replace_all("^dust[a-z]", "dust ") %>%
str_replace_all("\\slood", " flood") %>%
str_replace_all("vog", "fog") %>% # fog
str_replace_all("\\bd[a-z]*l\\b", "devil") %>% # Dust devil
str_replace_all("\\bch[a-z]*l\\b", "chill") %>% # Chill
str_replace_all("fld", "flood") %>%
str_replace_all("\\bfloo[a-z]*\\b", "flood") %>% # floods
str_replace_all("hvy", "heavy") %>%
str_replace_all("\\bw[a-z]*d\\b", "wind") %>%
str_replace_all("\\bt[a-z]*m\\b", "thunderstorm") %>%
str_replace_all("tstmw", "thunderstorm") %>%
str_replace_all("^thunderstorm[a-z]", "thunderstorm") %>%
str_replace_all("^wayter", "water")Remove observations where event type does not describe any official event type.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype
stormdata4 <- stormdata4 %>%
filter(!str_detect(cleanevtype2, "^\\?")) %>%
filter(!str_detect(cleanevtype2, "^apache")) %>%
filter(!str_detect(cleanevtype2, "^dam break")) %>%
filter(!str_detect(cleanevtype2, "^drowning")) %>%
filter(!str_detect(cleanevtype2, "^hyperthermia")) %>%
filter(!str_detect(cleanevtype2, "^hypothermia")) %>%
filter(!str_detect(cleanevtype2, "^marine accident")) %>%
filter(!str_detect(cleanevtype2, "^marine mishap")) %>%
filter(!str_detect(cleanevtype2, "^mild pattern")) %>%
filter(!str_detect(cleanevtype2, "^no severe weather")) %>%
filter(!str_detect(cleanevtype2, "^none")) %>%
filter(!str_detect(cleanevtype2, "^other")) %>%
filter(!str_detect(cleanevtype2, "^red flag criteria")) %>%
filter(!str_detect(cleanevtype2, "^southeast")) %>%
filter(!str_detect(cleanevtype2, "^summary"))With misspellings corrected and nonsense observations removed, the manual process of cleaning event types can start. Events with more than one type are assumed to be the first event listed. events synonyms are assigned to the appropriate event type. Name abbreviations are expanded.
Astronomical Low Tide Also includes references to blow-out tides.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^blow-out.*$", "astronomical low tide")Avalanche No synonyms or needed changes.
Blizzard Also includes references to snowstorms and blowing snow.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^blizzard.*$", "blizzard") %>%
str_replace_all("^snowstorm", "blizzard") %>%
str_replace_all("ground blizzard", "blizzard") %>%
str_replace_all("^blowing\\ssnow.*$", "blizzard")Coastal Flood Also includes references to erosion caused floods, beach floods and tidal floods.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^coastal(\\s[fs]|\\/).*$", "coastal flood") %>%
str_replace_all("^erosion.*$", "coastal flood") %>%
str_replace_all("^beach.*flood$", "coastal flood") %>%
str_replace_all("^tidal flood$", "coastal flood")Cold/Wind Chill Also includes references to low temperature, unseasonably cold, bitter wind chill and prolonged cold.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
#cold/wind chill
str_replace_all("^cold.*$", "cold/wind chill") %>%
str_replace_all("^wind\\schill.*$", "cold/wind chill") %>%
str_replace_all("^bitter wind chill.*$", "cold/wind chill") %>%
str_replace_all("^low\\stemp.*$", "cold/wind chill") %>%
str_replace_all("^low wind chill", "cold/wind chill") %>%
str_replace_all("^low wind chill", "cold/wind chill") %>%
str_replace_all("^unseasonably co.*$", "cold/wind chill") %>%
str_replace_all("^prolong co.*$", "cold/wind chill")Debris Flow Also includes references to mud flows, landslides and rock flows.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
#debris flow
str_replace_all("^mud.*$", "debris flow") %>%
str_replace_all("^landsl.*$", "debris flow") %>%
str_replace_all("^rock\\ss.*$", "debris flow")Dense Smoke Also includes references to smoke.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
#dense smoke
str_replace_all("^smoke.*$","dense smoke")Dense Fog Also includes references to fog.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^fog.*$", "dense fog") Drought Also includes references to dry weather, extreme dry, excessive dry, abnormally dry, driest period, and unseasonably dry.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^dry.*$", "drought") %>%
str_replace_all("^ex[a-z]*\\b\\sdr.*$", "drought") %>%
str_replace_all("^very dry$", "drought") %>%
str_replace_all("^drought.*$", "drought") %>%
str_replace_all("^snow drought", "drought") %>%
str_replace_all("^abnormally dry", "drought") %>%
str_replace_all("^driest month", "drought") %>%
str_replace_all("^unseasonably dry", "drought")Dust Storm Also includes references to blowing dust and Saharan dust.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^dust\\sst.*$", "dust storm") %>%
str_replace_all("blowing dust", "dust storm") %>%
str_replace_all("saharan dust", "dust storm")Dust devil No synonyms.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^dust\\sde.*$", "dust devil") Excessive Heat Also includes references to very warm, abnormal warmth, record heat, record warmth and record temperature.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^ex[a-z]*\\b\\sheat.*$", "excessive heat") %>%
str_replace_all("^very warm$", "excessive heat") %>%
str_replace_all("^abnormal warmth", "excessive heat") %>%
str_replace_all("^record\\sheat.*$", "excessive heat") %>%
str_replace_all("^record\\swar.*$", "excessive heat") %>%
str_replace_all("^record\\stemp.*$", "excessive heat")Extreme Cold/Wind Chill Also includes references to excessive cold, extreme wind chill, record cold and severe cold.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^ex[a-z]*\\b\\scold.*$", "extreme cold/wind chill") %>%
str_replace_all("^ex[a-z]*\\b\\swind.*$", "extreme cold/wind chill") %>%
str_replace_all("^ex[a-z]*\\b\\/.*$", "extreme cold/wind chill") %>%
str_replace_all("^record\\scold.*$", "extreme cold/wind chill") %>%
str_replace_all("^severe\\scold.*$", "extreme cold/wind chill") Flash Flood Also includes references to local flash flood.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^flash.*$", "flash flood") %>%
str_replace_all("^local flash.*$", "flash flood") Flood Also includes references to any flood event that does not explicitly fall into the coastal, flash or lakeshore category.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^flood.*$", "flood") %>%
str_replace_all("^local flood", "flood") %>%
str_replace_all("^major flood", "flood") %>%
str_replace_all("^snowmelt flood", "flood") Freezing Fog Also includes references to glaze and freezing drizzle.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^glaze.*$", "freezing fog") %>%
str_replace_all("^freezing dri.*$", "freezing fog")Frost/Freeze Also includes references to agricultural freeze, hard freeze, damaging freeze, icy roads and black ice.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^frost.*$", "frost/freeze") %>%
str_replace_all("^agric.*$", "frost/freeze") %>%
str_replace_all("^freeze.*$", "frost/freeze") %>%
str_replace_all("^hard freeze.*$", "frost/freeze") %>%
str_replace_all("^damaging freeze.*$", "frost/freeze") %>%
str_replace_all("^icy roads.*$", "frost/freeze") %>%
str_replace_all("^black ice.*$", "frost/freeze") Funnel Cloud Also includes references to wall clouds.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^funnel.*$", "funnel cloud") %>%
str_replace_all("^wall\\sc.*$", "funnel cloud")Hail Also includes references to small hail.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^hail.*$","hail") %>%
str_replace_all("^small hail.*$","hail")Heat Also includes references to hot weather, warm weather and unseasonably warm weather.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^heat.*$", "heat") %>%
str_replace_all("^hot.*$", "heat") %>%
str_replace_all("^warm.*$", "heat") %>%
str_replace_all("^unseasonably wa.*$", "heat")Heavy Rain Also includes references to
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^heavy ra.*$", "heavy rain") %>%
str_replace_all("^torrential.*$", "heavy rain") %>%
str_replace_all("^ex[a-z]*\\b\\srain.*$", "heavy rain") %>%
str_replace_all("^ex[a-z]*\\b\\spreci.*$", "heavy rain") %>%
str_replace_all("^ex[a-z]*\\b\\swe.*$", "heavy rain") %>%
str_replace_all("^heavy\\spre.*$", "heavy rain") %>%
str_replace_all("^heavy\\ssh.*$", "heavy rain") %>%
str_replace_all("^locally heavy rain", "heavy rain") %>%
str_replace_all("^rain.*$", "heavy rain") %>%
str_replace_all("^record rain.*$", "heavy rain") %>%
str_replace_all("^monthly\\s[rp].*$", "heavy rain") %>%
str_replace_all("^unseasonably wet", "heavy rain") Heavy Snow Also includes references to other snow events not explicitly lake-effect related.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^heavy sno.*$", "heavy snow") %>%
str_replace_all("^heavy sno.*$", "heavy snow") %>%
str_replace_all("^ex[a-z]*\\b\\ssnow.*$", "heavy snow") %>%
str_replace_all("^snow.*$", "heavy snow") %>%
str_replace_all("^moderate snowfall", "heavy snow") High Surf Also includes references to heavy surf, heavy swell, high swell, rough surf and hazardous surf.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^heavy\\ssu.*$", "high surf") %>%
str_replace_all("^heavy\\ssw.*$", "high surf") %>%
str_replace_all("^high\\ssu.*$", "high surf") %>%
str_replace_all("^high\\ssw.*$", "high surf") %>%
str_replace_all("^rough\\ssu.*$", "high surf") %>%
str_replace_all("^hazardous surf", "high surf") High Wind Also includes references to heavy wind and gusty wind.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^high wi.*$", "high wind") %>%
str_replace_all("^heavy wi.*$", "high wind") %>%
str_replace_all("^gusty\\sw.*$", "high wind")Hurricane/Typhoon Includes all references to hurricanes and typhoons. For the purpose of this report, all events are aggregated, i.e., hurricanes and typhoons are not named.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^hurricane.*$", "hurricane/typhoon") %>%
str_replace_all("^typhoon$", "hurricane/typhoon") Ice Storm No synonyms in the database.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^ice.*$", "ice storm")Lake-effect Snow Also includes references to heavy lake snow.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^lake.*snow$", "lake-effect snow") %>%
str_replace_all("^heavy lake snow$", "lake-effect snow")Lakeshore Flood No synonyms in the database.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^lake.*flood$", "lakeshore flood") Lightning No synonyms in the database
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^light[a-z]*.*$", "lightning")Marine Hail No synonyms in the database.
Marine High Wind No synonyms in the database.
Marine Strong Wind No synonyms in the database.
Marine Thunderstorm Winds Also includes references to coastal thunderstorms.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("coastal thunderstorm", "marine thunderstorm wind") Rip Current No synonyms in the database.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^rip.*$", "rip current")Seiche No synonyms in the database.
Sleet Also includes references to freezing rain and freezing spray.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^sleet.*$", "sleet") %>%
str_replace_all("^freezing\\srain.*$", "sleet") %>%
str_replace_all("^freezing\\sspra.*$", "sleet")Storm/Surge Tide Also includes references to astronomical high tide.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^storm.*$", "storm surge/tide") %>%
str_replace_all("^astronomical high tide", "storm surge/tide")Strong Wind Also includes references to wind not included in other wind event types.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^strong wi.*$", "strong wind") %>%
str_replace_all("^wind.*$", "strong wind")Thunderstorm Wind Also includes references to thunderstorms, gusty thunderstorm wind and heavy thunderstorm wind.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^thunders[a-z]*.*$", "thunderstorm wind") %>%
str_replace_all("^gusty\\st.*$", "thunderstorm wind") %>%
str_replace_all("^severe\\sth.*$", "thunderstorm wind") Tornado Also includes references to gustnado and landspout.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^torn.*$", "tornado") %>%
str_replace_all("^gustnado.*$", "tornado") %>%
str_replace_all("^landspout.*$", "tornado") Tropical Depression No sysnonyms in the database.
Tropical Storm No synonyms in the database.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^tropical\\ss.*$", "tropical storm")Tsunami No synonyms in the database.
Volcanic Ash Also include references to other volcanic events.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^volca.*$", "volcanic ash")Waterspout No synonyms in the database.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^water.*$","waterspout") Wildfire Also includes references to other fire events.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^.*fire.*$", "wildfire")Winter Storm No synonyms in the database.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
#winter storm
str_replace_all("^winter\\ss.*$", "winter storm")Winter Weather Also includes references to other winter events not included in previous event types.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^winter\\sw.*$", "winter weather") %>%
str_replace_all("^wint[a-z]*\\sm.*$", "winter weather")Finish cleaning the event types Include remaining references to floods into the Flood type.
stormdata4$cleanevtype2 <- stormdata4$cleanevtype2 %>%
str_replace_all("^[^cfl].*flood", "flood")And remove observations of the remaining event types that could not be reclassified.
officialevtypes <- c("astronomical low tide",
"avalanche",
"blizzard",
"coastal flood",
"cold/wind chill",
"debris flow",
"dense fog",
"dense smoke",
"drought",
"dust devil",
"dust storm",
"excessive heat",
"extreme cold/wind chill",
"flash flood",
"flood",
"freezing fog",
"frost/freeze",
"funnel cloud",
"hail",
"heat",
"heavy rain",
"heavy snow",
"high surf",
"high wind",
"hurricane/typhoon",
"ice storm",
"lake-effect snow",
"lakeshore flood",
"lightning",
"marine hail",
"marine high wind",
"marine strong wind",
"marine thunderstorm wind",
"rip current",
"seiche",
"sleet",
"storm surge/tide",
"strong wind",
"thunderstorm wind",
"tornado",
"tropical depression",
"tropical storm",
"tsunami",
"volcanic ash",
"waterspout",
"wildfire",
"winter storm",
"winter weather")
#filter the data to only include official types
finalstormdata <- stormdata4 %>%
filter(cleanevtype2 %in% officialevtypes) %>%
group_by(cleanevtype2) %>%
select(bgn_date, state, cleanevtype2, fatalities, injuries, propdmgtr, cropdmgtr)Analysis
To assess the risk associated with storm events, specific measures of severity were chosen to signify the impact of storm occurrences on public health and financial losses. Indicators for the effect on the population include aggregate, mean, median, and maximum values for both fatalities and injuries associated with each type of storm event. Indicators for economic impact include aggregate, mean, median, and maximum values for property damage and crop damage.
To evaluate the probability of storm events, the average frequency of events per category per year was computed. Recognizing the sparsity of data at the beginning of the database, which might influence the findings, the analysis was conducted for both the complete dataset and the most recent 20 years of data
#make a summary table for the whole database
fulldbsummary <- finalstormdata %>%
summarize(sumfatalities = sum(fatalities), meanfatalities = mean(fatalities),
medianfatalities = median(fatalities), maxfatalities = max(fatalities),
suminjuries = sum(injuries), meaninjuries = mean(injuries),
medianinjuries = median(injuries), maxinjuries = max(injuries),
sumpropdmg = sum(propdmgtr), meanpropdmg = mean(propdmgtr),
medianpropdmg = median(propdmgtr), maxpropdmg = max(propdmgtr),
sumcropdmg = sum(cropdmgtr), meancropdmg = mean(cropdmgtr),
mediancropdmg = median(cropdmgtr), maxcropdmg = max(cropdmgtr)) %>%
mutate(nevents = count(finalstormdata, cleanevtype2)$n)
# and another for the last 20 years (2011 is incomplete so 1991-2010)
last20ydbsummary <- finalstormdata %>%
filter(year(bgn_date) > 1990 & year(bgn_date) < 2011) %>%
summarize(sumfatalities = sum(fatalities), meanfatalities = mean(fatalities),
medianfatalities = median(fatalities), maxfatalities = max(fatalities),
suminjuries = sum(injuries), meaninjuries = mean(injuries),
medianinjuries = median(injuries), maxinjuries = max(injuries),
sumpropdmg = sum(propdmgtr), meanpropdmg = mean(propdmgtr),
medianpropdmg = median(propdmgtr), maxpropdmg = max(propdmgtr),
sumcropdmg = sum(cropdmgtr), meancropdmg = mean(cropdmgtr),
mediancropdmg = median(cropdmgtr), maxcropdmg = max(cropdmgtr)) %>%
mutate(nevents = count(filter(finalstormdata, year(bgn_date) > 1990 & year(bgn_date) < 2011),
cleanevtype2)$n)Results
Most damaging event types
The analysis identifies the event types with the greatest cumulative impact on either public health or economic cost throughout the entire database period. The findings highlight the top five event types responsible for the most fatalities, injuries, and property or crop damage.
library(ggplot2)
library(gridExtra)##
## Attaching package: 'gridExtra'## The following object is masked from 'package:dplyr':
##
## combineoptions(scipen = 4, digits = 2)
psum1 <- fulldbsummary %>%
filter(sumfatalities > quantile(sumfatalities,0.9)) %>%
ggplot(aes(x=reorder(cleanevtype2,-sumfatalities), y=sumfatalities, fill=cleanevtype2)) +
geom_bar(stat = "identity") +
scale_fill_brewer(palette = "Set3") +
theme(axis.text = element_text(angle = 45, hjust=1, vjust=1)) +
ylab("Total fatalities") +
xlab("") +
ggtitle("Top Event Types Causing the Most Fatalities")
psum2 <- fulldbsummary %>%
filter(suminjuries > quantile(suminjuries,0.9)) %>%
ggplot(aes(x=reorder(cleanevtype2,-suminjuries), y=suminjuries, fill=cleanevtype2)) +
geom_bar(stat = "identity") +
scale_fill_brewer(palette = "Set3") +
theme(axis.text = element_text(angle = 45, hjust=1, vjust=1)) +
ylab("Total injuries") +
xlab("") +
ggtitle("Top Event Types Causing the Most Injuries")
psum3 <- fulldbsummary %>%
filter(sumpropdmg > quantile(sumpropdmg,0.9)) %>%
ggplot(aes(x=reorder(cleanevtype2,-sumpropdmg), y=sumpropdmg, fill=cleanevtype2)) +
geom_bar(stat = "identity") +
scale_fill_brewer(palette = "Set3") +
theme(axis.text = element_text(angle = 45, hjust=1, vjust=1)) +
ylab("Total property damage (USD)") +
xlab("Event type") +
ggtitle("Top Event Types Causing the Most Property Damage")
psum4 <- fulldbsummary %>%
filter(sumcropdmg > quantile(sumcropdmg,0.9)) %>%
ggplot(aes(x=reorder(cleanevtype2,-sumcropdmg), y=sumcropdmg, fill=cleanevtype2)) +
geom_bar(stat = "identity") +
scale_fill_brewer(palette = "Set3") +
theme(axis.text = element_text(angle = 45, hjust=1, vjust=1)) +
ylab("Total crop damage (USD)") +
xlab("Event type") +
ggtitle("Top Event Types Causing the Most Crop Damage")
gridExtra::grid.arrange(psum1, psum2, psum3, psum4, nrow=2)
Figure 1 illustrates the top five event types that have caused the most
significant impact on population health (in terms of fatalities and
injuries) and economic damages (including property and crop damage) over
the entire period covered by the database.
Most potentially damaging event types
The study emphasizes the significance of individual catastrophic events that might not be captured by average damage analysis. It introduces the concept of the most potentially damaging event type, highlighting heat, tornadoes, excessive heat, tsunamis, and tropical storms as the most damaging in terms of fatalities. Similarly, tornadoes, ice storms, floods, hurricane/typhoons, and excessive heat are noted as the most damaging in terms of injuries. Additionally, floods, storm surge/tides, hurricane/typhoons, tropical storms, and winter storms emerge as the most damaging for property damage, while floods, ice storms, hurricane/typhoons, droughts, and extreme cold/wind chill are identified as the most damaging for crop damage.
options(scipen = 4, digits = 2)
# Color palette
my_colors <- scale_fill_brewer(palette = "Set3")
pmax1 <- fulldbsummary %>%
filter(maxfatalities > quantile(maxfatalities, 0.9)) %>%
ggplot(aes(x = reorder(cleanevtype2, -maxfatalities), y = maxfatalities, fill = cleanevtype2)) +
geom_bar(stat = "identity") +
my_colors +
theme(axis.text = element_text(angle = 45, hjust = 1, vjust = 1)) +
ylab("Max fatalities") +
xlab("") +
ggtitle("Top Event Types Causing Max Fatalities")
pmax2 <- fulldbsummary %>%
filter(maxinjuries > quantile(maxinjuries, 0.9)) %>%
ggplot(aes(x = reorder(cleanevtype2, -maxinjuries), y = maxinjuries, fill = cleanevtype2)) +
geom_bar(stat = "identity") +
my_colors +
theme(axis.text = element_text(angle = 45, hjust = 1, vjust = 1)) +
ylab("Max injuries") +
xlab("") +
ggtitle("Top Event Types Causing Max Injuries")
pmax3 <- fulldbsummary %>%
filter(maxpropdmg > quantile(maxpropdmg, 0.9)) %>%
ggplot(aes(x = reorder(cleanevtype2, -maxpropdmg), y = maxpropdmg, fill = cleanevtype2)) +
geom_bar(stat = "identity") +
my_colors +
theme(axis.text = element_text(angle = 45, hjust = 1, vjust = 1)) +
ylab("Max property damage (USD)") +
xlab("Event type") +
ggtitle("Top Event Types Causing Max Property Damage")
pmax4 <- fulldbsummary %>%
filter(maxcropdmg > quantile(maxcropdmg, 0.9)) %>%
ggplot(aes(x = reorder(cleanevtype2, -maxcropdmg), y = maxcropdmg, fill = cleanevtype2)) +
geom_bar(stat = "identity") +
my_colors +
theme(axis.text = element_text(angle = 45, hjust = 1, vjust = 1)) +
ylab("Max crop damage (USD)") +
xlab("Event type") +
ggtitle("Top Event Types Causing Max Crop Damage")
grid.arrange(pmax1, pmax2, pmax3, pmax4, nrow = 2)