Effects of Severe Weather Events on Population and Economy
Matt Dancho
May 30, 2016
Synopsis
This report reviews the effects of severe weather events on both the population and the economy. This is an investigative report in that we are seeking to determine which event types have historically caused the most fatalities (a measure of population impact) and property damage (a measure of economic cost). Data was obtained from the U.S. National Oceanic and Atmospheric Administration’s (NOAA) storm database. The events start in the year 1950 and end in November 2011. The conclusion is that tornados cause the most harm to the population in terms of fatalties and floods cause the most economic cost.
Data Processing
The first step is to load the packages that will be used for data manipulation/wrangling and plotting.
# Data wrangling
library(dplyr)
library(tidyr)
# Plotting
library(ggplot2)
# Tables
library(pander)Next, the data is loaded using the read.table function.
stormDataRaw <- read.table("repdata-data-StormData.csv.bz2", header = TRUE, sep = ",")We call the head function to make a quick check of the dataset.
head(stormDataRaw)## STATE__ BGN_DATE BGN_TIME TIME_ZONE COUNTY COUNTYNAME STATE
## 1 1 4/18/1950 0:00:00 0130 CST 97 MOBILE AL
## 2 1 4/18/1950 0:00:00 0145 CST 3 BALDWIN AL
## 3 1 2/20/1951 0:00:00 1600 CST 57 FAYETTE AL
## 4 1 6/8/1951 0:00:00 0900 CST 89 MADISON AL
## 5 1 11/15/1951 0:00:00 1500 CST 43 CULLMAN AL
## 6 1 11/15/1951 0:00:00 2000 CST 77 LAUDERDALE AL
## EVTYPE BGN_RANGE BGN_AZI BGN_LOCATI END_DATE END_TIME COUNTY_END
## 1 TORNADO 0 0
## 2 TORNADO 0 0
## 3 TORNADO 0 0
## 4 TORNADO 0 0
## 5 TORNADO 0 0
## 6 TORNADO 0 0
## COUNTYENDN END_RANGE END_AZI END_LOCATI LENGTH WIDTH F MAG FATALITIES
## 1 NA 0 14.0 100 3 0 0
## 2 NA 0 2.0 150 2 0 0
## 3 NA 0 0.1 123 2 0 0
## 4 NA 0 0.0 100 2 0 0
## 5 NA 0 0.0 150 2 0 0
## 6 NA 0 1.5 177 2 0 0
## INJURIES PROPDMG PROPDMGEXP CROPDMG CROPDMGEXP WFO STATEOFFIC ZONENAMES
## 1 15 25.0 K 0
## 2 0 2.5 K 0
## 3 2 25.0 K 0
## 4 2 2.5 K 0
## 5 2 2.5 K 0
## 6 6 2.5 K 0
## LATITUDE LONGITUDE LATITUDE_E LONGITUDE_ REMARKS REFNUM
## 1 3040 8812 3051 8806 1
## 2 3042 8755 0 0 2
## 3 3340 8742 0 0 3
## 4 3458 8626 0 0 4
## 5 3412 8642 0 0 5
## 6 3450 8748 0 0 6The data was loaded correctly, but we have many columns that are not necessary for the purpose of this analysis. Let’s select only the columns needed for the analysis, which relate to the event type, population damage (fatalities) and economic damage (property damage). I use pander for better table formatting.
stormDataSelections <- stormDataRaw %>%
select(EVTYPE, FATALITIES, PROPDMG, PROPDMGEXP)
pander(head(stormDataSelections), caption="Storm Dataset: Pre-Processing, First Six Rows")| EVTYPE | FATALITIES | PROPDMG | PROPDMGEXP |
|---|---|---|---|
| TORNADO | 0 | 25 | K |
| TORNADO | 0 | 2.5 | K |
| TORNADO | 0 | 25 | K |
| TORNADO | 0 | 2.5 | K |
| TORNADO | 0 | 2.5 | K |
| TORNADO | 0 | 2.5 | K |
Upon inspection, further data processing is necessary because the property damage values are stored in various units depeding on the respective “exp” code, where:
- K = Thousands
- M = Millions
- B = Billions
Specifically, the issue with PROPDMG is that the value can be in measured in thousands, millions, or billions depending on the code K, M, or B in PROPDMGEXP. To account for this, PROPDMG.MULT is added as a multiplier depending on the PROPDMGEXP code. The total value is the product of the PROPDMG and PROPDMG.MULT, which is stored as PROPDMG.TOTAL.
stormData <- stormDataSelections %>%
mutate(PROPDMG.MULT = ifelse(PROPDMGEXP == "K", 1000,
ifelse(PROPDMGEXP == "M", 1000000,
ifelse(PROPDMGEXP == "B", 1000000000, 0)))) %>%
mutate(PROPDMG.TOTAL = PROPDMG * PROPDMG.MULT) %>%
select(EVTYPE, FATALITIES, PROPDMG.TOTAL)
pander(head(stormData), caption="Storm Dataset: Post-Processing, First Six Rows")| EVTYPE | FATALITIES | PROPDMG.TOTAL |
|---|---|---|
| TORNADO | 0 | 25000 |
| TORNADO | 0 | 2500 |
| TORNADO | 0 | 25000 |
| TORNADO | 0 | 2500 |
| TORNADO | 0 | 2500 |
| TORNADO | 0 | 2500 |
Finally, we can look at a summary of the final dataset to get a sense of what the dataset looks like.
summary(stormData)## EVTYPE FATALITIES PROPDMG.TOTAL
## HAIL :288661 Min. : 0.0000 Min. :0.000e+00
## TSTM WIND :219940 1st Qu.: 0.0000 1st Qu.:0.000e+00
## THUNDERSTORM WIND: 82563 Median : 0.0000 Median :0.000e+00
## TORNADO : 60652 Mean : 0.0168 Mean :4.735e+05
## FLASH FLOOD : 54277 3rd Qu.: 0.0000 3rd Qu.:5.000e+02
## FLOOD : 25326 Max. :583.0000 Max. :1.150e+11
## (Other) :170878Results
Which types of events are most harmful to population health?
Fatalities are the most serious impact to population health, and as such this was the measure selected for analysis of which events are most harmful to the population. The dataset was summarized on fatalities by event type, which resulted in 985 events. However, the bulk of the fatalties are accumulated from a small fraction of the 985 events. The top 10 were examined, which comprises 80% of the fatalities. We can see from the plot below that Tornados resulted in 5633 fatalities over the time period analyzed, which is significantly more than any other event. Therefore, Tornados are the most harmful to population health.
# Organize data
fatalities <- stormData %>%
group_by(EVTYPE) %>%
summarize(FATALITIES.TOTAL = sum(FATALITIES)) %>%
arrange(desc(FATALITIES.TOTAL))
fatalities.top10 <- fatalities[1:10, ]
# Need to order by value otherwise the graph will be alphabetically ordered
fatalities.top10.ordered <- fatalities.top10
fatalities.top10.ordered$EVTYPE <- factor(fatalities.top10.ordered$EVTYPE,
levels = arrange(fatalities.top10.ordered, FATALITIES.TOTAL)$EVTYPE)
# Plot
ggplot(data=fatalities.top10.ordered, aes(x=EVTYPE, y=FATALITIES.TOTAL)) +
geom_bar(stat="identity") +
geom_text(aes(ymax=FATALITIES.TOTAL, label=FATALITIES.TOTAL),
hjust= -0.25,
vjust= 0.5,
color="black",
size=4) +
coord_flip() +
labs(title="Top 10 Weather Events Causing Fatalities",
x="Weather Event Type",
y="Total Fatalities")+
scale_y_continuous(limits = c(0,6500))
Top 10 Weather Events Causing Fatalities
Which types of events have the greatest economic consequences?
Property damage is the best measurement of economic consequences. Similar to fatalities, the data can be summarized by event type. Following the same process, the top 10 events causing property damage are shown in the graph below, as these events comprise 88% of the total property damage. As shown below, the majority of property damage is caused by floods at $145T over the time period analyzed. The next closest is Hurricanes/Typhoons at less than half the property damage. Therefore, floods are the most damaging weather event when measured by economic cost.
# Organize data
propertyDamage <- stormData %>%
group_by(EVTYPE) %>%
summarize(PROPDMG.TOTAL = sum(PROPDMG.TOTAL)) %>%
arrange(desc(PROPDMG.TOTAL))
propertyDamage.top10 <- propertyDamage[1:10, ]
# Need to order by value otherwise the graph will be alphabetically ordered
propertyDamage.top10.ordered <- propertyDamage.top10
propertyDamage.top10.ordered$EVTYPE <- factor(propertyDamage.top10.ordered$EVTYPE,
levels = arrange(propertyDamage.top10.ordered, PROPDMG.TOTAL)$EVTYPE)
# Plot
ggplot(data=propertyDamage.top10.ordered, aes(x=EVTYPE, y=PROPDMG.TOTAL)) +
geom_bar(stat="identity") +
geom_text(aes(ymax=PROPDMG.TOTAL, label=scales::dollar(PROPDMG.TOTAL)),
hjust= -0.25,
vjust= 0.5,
color="black",
size=4) +
coord_flip() +
labs(title="Top 10 Weather Events Causing Property Damage",
x="Weather Event",
y="Total Property Damage Value")+
scale_y_continuous(labels = scales::dollar, limits = c(0,250000000000))
Top 10 Weather Events Causing Property Damage