Synopsis

In this report we aim to describe the type of weather events that are 1) most harmful with respect to population health in the United States and 2) most costly to the United States with respect to economic consequences. Using storm data from the U.S. National Oceanic and Atmospheric Administration (NOAA), we measured population health by the number of fatalities and injuries caused and economic consequences by the dollar amount damage done to property and crops. As observed from 1950-2011, we determined the following:

Data Processing

Getting & Loading Data

In order to conduct this analysis, we used data from the U.S. National Oceanic and Atomspheric Administration’s (NOAA) storm database. This database tracks characteristics of major storms and weather events in the United States. To manually download the file into the parent directory, you can access the file here.

The file containing the storm data is compressed using a bz2 algorithm. The data is a comma-delimited file. We use header = TRUE to read in the header data. Be forewarned that this code chunk can take a few minutes to process.

if (!file.exists("storm_data.csv.bz2")){
        url <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
        download.file(url, "storm_data.csv.bz2", method = "curl")
}
stormData <- read.csv(bzfile("storm_data.csv.bz2"),sep=",", header = TRUE)

After reading in the storm data, we check the number of observations (rows) and variables (columns) in this dataset. There are 902,297 rows and 37 columns.

dim(stormData)
## [1] 902297     37
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"

Determining Harm to Population Health

We are interested in the EVTYPE variable which gives us the weather event responsible for the damage. To assess harm to population health, we will be using the FATALITIES and INJURIES variables. Missing values are dropped from the analysis by default.

popHealth <- aggregate(FATALITIES + INJURIES ~ EVTYPE, data = stormData, sum)
names(popHealth) = c("EVENT.TYPE", "FATALITIES.INJURIES")
popHealth <- popHealth[order(popHealth$FATALITIES.INJURIES, decreasing = TRUE),]

Determining Consequences to the Economy

To assess economic damage, we will be using the PROPDMG and CROPDMG variables. Note that PROPDMG and CROPDMG have associated variables that denote its exponential value: PROPDMGEXP and CROPDMGEXP, respectively. We make this cost information more use for comparison and do this only for values with meaningful exponential symbols: K, M, B, and "" (for none).

# Subset stormData for values ending in meaningful exponent symbols 
exponentData <- stormData[stormData$PROPDMGEXP %in% c("", "K", "M", "B") & stormData$CROPDMGEXP %in% 
    c("", "K", "M", "B"), ]
# Function to apply exponent values to base values
applyExponent <- function(DMG, EXP) {
    if (EXP == "K") {
        DMG * 1000
    } else if (EXP == "M") {
        DMG * 1e+06
    } else if (EXP == "B") {
        DMG * 1e+09
    } else if (EXP == "") {
        DMG
    } else {
        stop("NOEXP")
    }
}
# mapply can be used to cycle parameters through a function
exponentData$PROP_DAMAGE_AMOUNT <- mapply(applyExponent, exponentData$PROPDMG, exponentData$PROPDMGEXP)
exponentData$CROP_DAMAGE_AMOUNT <- mapply(applyExponent, exponentData$CROPDMG, exponentData$CROPDMGEXP)

Now that we have variables to represent economic consequences with the appropriate magnitude we can apply a similar approach as we did with population health to aggregate the cost data by EVTYPE. We create new variables PROPERTY.DAMAGE and CROP.DAMAGE.

econConseq <- aggregate(PROP_DAMAGE_AMOUNT + CROP_DAMAGE_AMOUNT ~ EVTYPE, data = exponentData, sum)
names(econConseq) <- c("EVENT.TYPE", "PROPERTY.CROP.DAMAGE")
econConseq <- econConseq[order(econConseq$PROPERTY.CROP.DAMAGE, decreasing = TRUE),]

Results

Harm to Population Health

# Reorder factor levels to get highest # of fatalities/injuries on top
library(lattice)
barchart(reorder(EVENT.TYPE, FATALITIES.INJURIES) ~ FATALITIES.INJURIES, 
         data = head(popHealth,10),
         main = "U.S. Casualties by Top 10 Weather Event Types",
         xlab = "Casualties (Fatalities & Injuries)",
         ylab = "Weather Event Type",
         col = "dark grey")

plot of chunk PopHealth Plot

From 1950-2011 in the United States, tornadoes by far cause the most harm to population health with respect to fatalities and injuries caused at 96,979 people.

Economic Consequences

library(lattice)
barchart(reorder(EVENT.TYPE, PROPERTY.CROP.DAMAGE) ~ PROPERTY.CROP.DAMAGE/10^6,
         data = head(econConseq,10),
         main = "Crop & Property Damage by Top 10 Weather Event Types",
         xlab = "Property & Crop Damage (in Millions US Dollars)",
         ylab = "Weather Event Type",
         col = "dark grey")

plot of chunk EconPlot

From 1950-2011 in the United States, floods cause the most economic damage with respect to damage to property and crops at $150,319,678,257 billion.