The Impacts of Severe Weather Events on Public Health and Welfare

Synopsis

The NOAA Storm Database was explored to answer some basic questions about severe weather events and other natural occurrences, specifically which ones have the most negative impact on public health, and the economy (in dollars expended to recover from property and agricultural damage). Natural disasters, extreme weather events, and adverse conditions can have a significant effect on public health, safety, and agriculture. Understanding which events are most likely to cause negative effects are key to planning and preparation. Tornadoes have the greatest impact on public safety, with over 150 combined injuries and fatalities resulting per tornado rated as Severe or Catastrophic (F4 or F5 on the Fujita Intensity Scale); nearly twice as many casualties as the next most frequent cause (excessive heat/hyperthermia). Floods of various types have the greatest collective economic impacts, with most of the damage being to property (but significant agricultural effects can be seen when occurring in farm areas), followed by storm surges/tides, and drought being the most negatively impactful to crop growth.

Data Processing

First, the data, obtained from the Department of Commerce, National Oceanic & Atmospheric Administration (NOAA), National Weather Service (NWS), was downloaded from the “Reproducible Research” course website, Week 4, Course Project 2.
The packages likely needed for this analysis were loaded (dplyr, lubridate, data.table)
For readability and processing later, the column named “F” was renamed “FUJITA”, and the “BGN_DATE” column was transformed into a workable date format.
Additionally, the support documentation states that, generally, the newer data will be more plentiful and more accurate. The database goes all the way back to 1950. The desire is to still provide a good sample of history, however, so data from the past 30 years, roughly (since 1 January 1990) was used as the sample.
The “MAG” column means two different things, depending on whether the event is discussing the size of hailstones (expressed in inches * 100) or the speed of the wind (in knots). So, these need to be examined and separated into two different variables, one discussing each (and eventually only, once the tables are further separated and worked with below) of the aforementioned references.

if (!file.exists("storm_data.csv")){
  download.file("https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2", destfile = "storm_data.csv")}
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, union

library(lubridate)

## Warning: package 'lubridate' was built under R version 3.6.2

## 
## Attaching package: 'lubridate'

## The following object is masked from 'package:base':
## 
##     date

library(data.table)

## Warning: package 'data.table' was built under R version 3.6.2

## 
## Attaching package: 'data.table'

## The following objects are masked from 'package:lubridate':
## 
##     hour, isoweek, mday, minute, month, quarter, second, wday, week,
##     yday, year

## The following objects are masked from 'package:dplyr':
## 
##     between, first, last

library(ggplot2)
storm_data <- read.csv("storm_data.csv", stringsAsFactors = FALSE)
colnames(storm_data)[21] <- "FUJITA"
storm_data$BGN_DATE <- as.Date(strptime(storm_data$BGN_DATE, "%m/%d/%Y"))
storm_data <- storm_data %>% filter(BGN_DATE >= "1990-01-01")
storm_data <- mutate(storm_data, HAIL.SIZE = MAG/100)
storm_data <- data.table(filter(storm_data, EVTYPE != "?"))
colnames(storm_data)[22] <- "WIND.SPEED(KNOTS)"

Unfortunately, the resultant data set, even when parsed out, restricted to the past 30 years, and cleaned up a bit, was still in no way tidy, nor were all the rows of data of interest for the study.
The event types (“EVTYPE”) were particularly inconsistent in their definitions. The NWS Guide defines only 48 different valid event types. Naturally, in the database, there are 985 different event types. So, tidying and binning of those was key to providing meaningful feedback (NOTE: FOR THIS CODE SECTION ONLY, have set echo = FALSE. But, in short, had to use “mutate” to change the EVTYPE 428 times in the larger “storm_data” set before even proceeding to the analysis of the individual health and economic events:
Since the two main outputs of this report are intended to discuss the most detrimental weather events to public health and the economy, respectively. So, two separate data table branches from the main “storm_data” table, “health_effects” and “economic_effects”, were created.

health_effects <- storm_data %>%
select(BGN_DATE, BGN_TIME, EVTYPE, FUJITA, `WIND.SPEED(KNOTS)`, HAIL.SIZE, FATALITIES, INJURIES, WFO, STATEOFFIC, REMARKS)

economic_effects <- storm_data %>%
select(BGN_DATE, BGN_TIME, EVTYPE, FUJITA, `WIND.SPEED(KNOTS)`, HAIL.SIZE, PROPDMG, PROPDMGEXP, CROPDMG, CROPDMGEXP, WFO, STATEOFFIC, REMARKS)

Further preparing the “health effects” table

The total number of casualties (defined here as fatalities + injuries) may be of interest, so that column was created.
- Additionally, when reporting on the overall effect of weather events on public health, only significant events should be considered. Therefore, this data set was restricted to only events resulting in at least 1 fatality OR 5 total casualties.
Furthermore, even some of the valid event types, as cleaned up before, are still redundant in the information they give (e.g. “EXCESSIVE HEAT and”HEAT"), so more consolidation is needed.
Finally, the data can be grouped, summarized, and sorted to get the total of each casualty type for each event type. And the Top 15 event types should paint a stark enough picture to get the point across

health_effects$TOTAL.CASUALTIES <- health_effects$FATALITIES + health_effects$INJURIES
health_effects <- filter(health_effects, FATALITIES >=1|TOTAL.CASUALTIES >=5)

health_effects <- health_effects %>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "EXCESSIVE HEAT/DROUGHT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "EXTREME HEAT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "UNSEASONABLY WARM","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAT WAVES","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "UNSEASONABLY WARM AND DRY","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "RECORD HEAT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "RECORD/EXCESSIVE HEAT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE THUNDERSTORM WIND","THUNDERSTORM WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "THUNDERSTORM WIND","DAMAGING WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "STROND WIND","DAMAGING WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "STRONG WIND","DAMAGING WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE STRONG WIND","DAMAGING WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HIGH WIND","DAMAGING WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "TORNADO F2","TORNADO"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "Heat Wave","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "FLASH FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "COASTAL FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "TROPICAL STORM GORDON","TROPICAL STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "TORNADOES, TSTM WIND, HAIL","TORNADO"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "STORM SURGE","STORM SURGE/TIDE"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAVY SNOW","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "SLEET","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "FROST/FREEZE","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "COLD/WIND CHILL","EXTREME COLD/WIND CHILL"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HYPERTHERMIA/EXPOSURE","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "WINTER WEATHER MIX","WINTER WEATHER"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAT WAVE DROUGHT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE HIGH WIND","DAMAGING WIND"))

combined_health_effects <- group_by(health_effects,EVTYPE)
combined_health_effects <- combined_health_effects %>% summarize(FATALITIES = sum(FATALITIES), INJURIES = sum(INJURIES), TOTAL.CASUALTIES = sum(TOTAL.CASUALTIES))

combined_health_effects <- combined_health_effects%>%
  arrange(desc(combined_health_effects$TOTAL.CASUALTIES))

top_15_health_effects <- data.table(combined_health_effects[1:15,])
EVTYPE <- rep(top_15_health_effects$EVTYPE, 3)
CASUALTY.TYPE <- c(rep("TOTAL.CASUALTIES",15),rep("INJURIES",15),rep("FATALITIES",15))
VICTIMS <- c(top_15_health_effects$TOTAL.CASUALTIES,top_15_health_effects$INJURIES,top_15_health_effects$FATALITIES)
top_15_health_effects <- data.table(cbind(EVTYPE, CASUALTY.TYPE, VICTIMS))
top_15_health_effects$CASUALTY.TYPE <- as.factor(top_15_health_effects$CASUALTY.TYPE)
top_15_health_effects$EVTYPE <- as.factor(top_15_health_effects$EVTYPE)
top_15_health_effects$VICTIMS <- as.numeric(top_15_health_effects$VICTIMS)

An additional examination of the leading threat to public health and safety (tornadoes, specifically) was done, showing the relationship between tornado intensity and the casualties they cause.

tornado_effects <- filter(health_effects, EVTYPE == "TORNADO")
fujitacount <- count(tornado_effects, FUJITA)
tornado_effects <- group_by(tornado_effects,FUJITA)
tornado_effects <- tornado_effects %>% summarize(FATALITIES = sum(FATALITIES), INJURIES = sum(INJURIES), TOTAL.CASUALTIES = sum(TOTAL.CASUALTIES))
tornado_effects <- data.table(cbind(fujitacount$n,tornado_effects))
tornado_effects <- data.table(tornado_effects[1:6,])
tornado_effects$AVG.FATALITIES <- tornado_effects$FATALITIES/tornado_effects$`fujitacount$n`
tornado_effects$AVG.INJURIES <- tornado_effects$INJURIES/tornado_effects$`fujitacount$n`
tornado_effects$AVG.TOTAL.CASUALTIES <- tornado_effects$TOTAL.CASUALTIES/tornado_effects$`fujitacount$n`
FUJITA <- rep(tornado_effects$FUJITA,3)
CASUALTY.TYPE <- c(rep("AVG.TOTAL.CASUALTIES",6),rep("AVG.INJURIES",6),rep("AVG.FATALITIES",6))
AVG.VICTIMS <- c(tornado_effects$AVG.TOTAL.CASUALTIES,tornado_effects$AVG.INJURIES,tornado_effects$AVG.FATALITIES)
tornado_effects <- data.table(cbind(FUJITA, CASUALTY.TYPE, AVG.VICTIMS))
tornado_effects$CASUALTY.TYPE <- as.factor(tornado_effects$CASUALTY.TYPE)
tornado_effects$FUJITA <- as.factor(tornado_effects$FUJITA)
tornado_effects$AVG.VICTIMS <- as.numeric(tornado_effects$AVG.VICTIMS)

tornado_effects <- tornado_effects%>%
  arrange(desc(CASUALTY.TYPE),desc(FUJITA))

Finally, a separate subset of the “storm_data” database was prepared to convey the adverse economic impacts of different natural events and disasters, specifically for dmages to property and crops. As was the case with the health impacts dataset, additional cleaning and binning (specifically, of the “EVTYPE”, “PROPDMG”, “PROPDMGEXP”,“CROPDMG”, and “CROPDMGEXP” columns) was necessary here with the economic impacts dataset.
From there, three tables branching from the “economic effects” branch were created, to more easily handle property and agricultural damage on their own, as well as one to easily show the total damage. They will all be combined into one data set for easier graphing.

economic_effects <- filter(economic_effects, economic_effects$PROPDMG != 0 | economic_effects$CROPDMG != 0)
economic_effects <- filter(economic_effects, economic_effects$PROPDMGEXP == "M"| economic_effects$PROPDMGEXP == "K" | economic_effects$PROPDMGEXP == "B"| economic_effects$PROPDMGEXP == "m"| economic_effects$CROPDMG != 0)

property_table <- data.table(select(economic_effects,EVTYPE,FUJITA,'WIND.SPEED(KNOTS)',HAIL.SIZE,PROPDMG,PROPDMGEXP,REMARKS))
prop_billions <- property_table %>%
filter(PROPDMGEXP == "B")%>%
mutate(DAMAGE.DOLLARS = PROPDMG * 1000000000)

prop_millions <- property_table %>%
filter(PROPDMGEXP == "M" | PROPDMGEXP == "m")%>%
mutate(DAMAGE.DOLLARS = PROPDMG * 1000000)

prop_thousands <- property_table %>%
filter(PROPDMGEXP == "K")%>%
mutate(DAMAGE.DOLLARS = PROPDMG * 1000)

property_table <- data.table(rbind(prop_billions,prop_millions,prop_thousands))
property_table <- arrange(property_table, property_table$DAMAGE.DOLLARS)
property_table <- data.table(filter(property_table, DAMAGE.DOLLARS != 0))

property_table <- property_table %>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "THUNDERSTORM WIND","HIGH WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "FLASH FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE STRONG WIND","STRONG WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE HIGH WIND","STRONG WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAVY SNOW","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "COASTAL FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "ICE STORM","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "TROPICAL DEPRESSION","TROPICAL STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE HAIL","HAIL"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "FREEZING FOG","WINTER WEATHER CONDITIONS"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "LAKE-EFFECT SNOW","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "LAKESHORE FLOOD","FLOOD"))

property_effects <- group_by(property_table,EVTYPE)
property_effects <- property_effects %>% summarize(DAMAGE.DOLLARS = sum(DAMAGE.DOLLARS))
property_effects <- arrange(property_effects,desc(DAMAGE.DOLLARS))
property_effects <- mutate(property_effects,DAMAGE.BILLIONS = DAMAGE.DOLLARS/1000000000)
property_effects[30,1] <- "FROST/FREEZE"
property_effects <- group_by(property_effects,EVTYPE)
property_effects <- property_effects %>% summarize(DAMAGE.DOLLARS = sum(DAMAGE.DOLLARS))
property_effects <- mutate(property_effects,TYPE = rep("PROPERTY",length(EVTYPE)))
property_effects <- arrange(property_effects,desc(DAMAGE.DOLLARS))
property_effects <- mutate(property_effects,DAMAGE.BILLIONS = DAMAGE.DOLLARS/1000000000)

crop_table <- data.table(select(economic_effects,EVTYPE,FUJITA,'WIND.SPEED(KNOTS)',HAIL.SIZE,CROPDMG,CROPDMGEXP,REMARKS))

crop_billions <- crop_table %>%
filter(CROPDMGEXP == "B")%>%
mutate(DAMAGE.DOLLARS = CROPDMG * 1000000000)

crop_millions <- crop_table %>%
filter(CROPDMGEXP == "M" | CROPDMGEXP == "m")%>%
mutate(DAMAGE.DOLLARS = CROPDMG * 1000000)

crop_thousands <- crop_table %>%
filter(CROPDMGEXP == "K" | CROPDMGEXP == "k")%>%
mutate(DAMAGE.DOLLARS = CROPDMG * 1000)

crop_table <- data.table(rbind(crop_billions,crop_millions,crop_thousands))
crop_table <- arrange(crop_table, crop_table$DAMAGE.DOLLARS)
crop_table <- data.table(filter(crop_table, DAMAGE.DOLLARS != 0))

crop_table <- crop_table %>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "THUNDERSTORM WIND","HIGH WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "FLASH FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE STRONG WIND","STRONG WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE HIGH WIND","STRONG WIND"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAVY SNOW","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "COASTAL FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "ICE STORM","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "TROPICAL DEPRESSION","TROPICAL STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "MARINE HAIL","HAIL"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "FREEZING FOG","WINTER WEATHER CONDITIONS"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "HEAT","EXCESSIVE HEAT"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "LAKE-EFFECT SNOW","WINTER STORM"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "LAKESHORE FLOOD","FLOOD"))%>%
mutate(EVTYPE = replace(EVTYPE, EVTYPE == "EXTREME COLD/WIND CHILL","COLD/WIND CHILL"))
  
crop_effects <- group_by(crop_table,EVTYPE)
crop_effects <- crop_effects %>% summarize(DAMAGE.DOLLARS = sum(DAMAGE.DOLLARS))
crop_effects <- arrange(crop_effects,desc(DAMAGE.DOLLARS))
crop_effects <- mutate (crop_effects,DAMAGE.BILLIONS = DAMAGE.DOLLARS/1000000000)
crop_effects <- mutate(crop_effects,TYPE = rep("CROP",length(EVTYPE)))

combined_effects <- data.table(rbind(property_effects,crop_effects))
combined_effects <- group_by(combined_effects, EVTYPE)
combined_effects <- combined_effects %>% summarize(DAMAGE.DOLLARS = sum(DAMAGE.DOLLARS))
combined_effects <- arrange(combined_effects,desc(DAMAGE.DOLLARS))
combined_effects <- mutate(combined_effects,DAMAGE.BILLIONS = DAMAGE.DOLLARS/1000000000)
combined_effects <- mutate(combined_effects,TYPE = rep("COMBINED",length(EVTYPE)))

all_effects <- data.frame(rbind(property_effects,crop_effects,combined_effects))
all_effects <- arrange(all_effects, desc(DAMAGE.BILLIONS))
all_effects <-  filter(all_effects, EVTYPE == "FLOOD" | EVTYPE == "HURRICANE(TYPHOON)"|EVTYPE == "STORM SURGE/TIDE" |EVTYPE == "TORNADO" |EVTYPE == "HAIL" |EVTYPE == "HIGH WIND" |EVTYPE == "WINTER STORM" |EVTYPE == "DROUGHT" |EVTYPE == "WILDFIRE" |EVTYPE == "TROPICAL STORM" |EVTYPE == "HEAVY RAIN" |EVTYPE == "FROST/FREEZE" |EVTYPE == "COLD/WIND CHILL")

Results

Natural disasters, extreme weather events, and adverse conditions can have a significant effect on public health and safety, resulting in various injuries and fatalities. Understanding which of these events are most likely to cause these negative effects are key to planning and preparation. Figure 1 below shows the 15 types of events over the past 30 years that resulted in the most human casualties.

dodge <- position_dodge(width = 0.4)

healthplot1 <- ggplot(data = top_15_health_effects, aes(x = reorder(EVTYPE,-VICTIMS), y = VICTIMS, fill = CASUALTY.TYPE))+
  geom_bar(stat = "identity", position = dodge, width = 0.4)+
    labs(title = "Figure 1 - Injuries and Fatalities Caused By Various\nNatural Disasters or Weather Events - Since 1990", x = "Natural Disaster/Weather Event Type", y = "Number of Casualties, By Type")+
  theme(axis.text.x=element_text(face = "bold", size = 4.25,angle = 0,vjust = 0.5),axis.text.y=element_text(size = 8,angle = 45))+
  scale_fill_manual(values=c("#009900", "#FF3300", "#003366"),
                    name = "Casualty Type",
                    labels = c("Fatalities", "Injuries", "Total Casualties\n(Injuries + Fatalities)"))+
  scale_x_discrete(labels = c("Tornadoes\n(All)","Extreme\nCold","Flood","Damaging\nWinds","Lightning","Winter\nStorm","Ice\nStorm","Hurricane\n(Typhoon)","Wildfire","Dense\nFog","Hail","Rip\nCurrent","Blizzard","Winter\nWeather","Excess.\nHeat"))
print(healthplot1)

As one can see, tornadoes result in nearly twice as many casualties as the next most frequent cause (excessive heat/hyperthermia). Tornadoes, however, are not created equally. They range in severity from F0 to F5 on the Fujita Scale, with the former more likely to cause minimal damages and the latter most often resulting in catastrophic loss of life and well-being. Figure 2 below shows a breakdown of the over 25,000 casualties caused by tornadoes by Fujita Scale severity:

fit <- lm(log(tornado_effects$AVG.VICTIMS[1:6]) ~ tornado_effects$FUJITA[1:6], data = tornado_effects[1:6,])

tornadoplot <- ggplot(data = tornado_effects, aes(x = reorder(FUJITA,AVG.VICTIMS), y = AVG.VICTIMS, fill = CASUALTY.TYPE))+
  geom_bar(stat = "identity", position = "dodge")+
  stat_function(fun = function(x) exp(fit$coefficients[1] + x*fit$coefficients[2]))+
  labs(title = "Injuries and Fatalities Caused By Tornadoes\nBy Fujita Scale Intensity, Per Event - Since 1990", x = "Fujita Scale Intensity", y = "Average Casualties Per Tornado Occurrence, By Type")+
  scale_fill_discrete(name = "Average...",labels = c("Fatalities", "Injuries", "Total Casualties"))+
  theme(axis.text.x=element_text(face = "bold", size=8,angle = 0,vjust = 0.65))
print(tornadoplot)

The growth in casualties is exponential in nature as the event intensity moves up the Fujita Scale. Therefore, extensive data about not only the direction and expected path of tornadoes, but also the expected intensity, is needed to help citizens best generally prepare for anm actively shelter during an imminent event.
In additional to significant impacts on public health and well-being, adverse natural events can also have economic impacts, which may or may not be related, directly or indirectly, to the health impacts.

dodge2 <- position_dodge(width = 0.3)

prop_ag_plot <- ggplot(data = all_effects, aes(x = reorder(EVTYPE,-DAMAGE.BILLIONS), y = DAMAGE.BILLIONS, fill = TYPE))+
  geom_bar(stat = "identity", position = dodge2, width = 0.4)+
  labs(title = "Figure 3 - Property and Agricultural Damage Caused\nNatural Disasters or Weather Events - Since 1990", x = "Natural Disaster/Weather Event Type", y = "Damages(Billions of U.S.Dollars)")+
  scale_fill_manual(values=c("#009900", "#FF3300", "#003366"),
                  name = "Type of Damage",
                  labels = c("Combined", "Crop", "Property"))+
  theme(axis.text.x=element_text(size = 4,angle = 0,vjust = 0.5),axis.text.y=element_text(size = 8,angle = 45))+
  scale_x_discrete(labels = c("Flood","Storm Surge/\nTide","Tornado","Hail","High\nWind","Winter\nStorm","Drought","Wildfire","Tropical\nStorm","Heavy\nRain","Frost\nFreeze","Extreme\nCold"))
  print(prop_ag_plot)

Floods of various types have the greatest collective economic impacts, with most of the damage being to property (but significant agricultural effects can be seen when occurring in farm areas). Storm surges and storm-related tides are next on the list, with all of the damage by those events being to property (as agricultiral areas tend to not be coastal regions). The greatest overall negative impact to agriculture comes from droughts, with few revenue-generating crops being able to survive extended periods of dryness.

The Impacts of Severe Weather Events on Public Health and Welfare

Sean Dobbs

18 Mar 2020

Synopsis

Data Processing

Further preparing the “health effects” table

Results