Peer Assessment 2 - NOAA Storm Database
Title
Identification of storm events in the United States that are the most harmful to the population’s health and have the greatest economic consequences.
Synopsis
This analysis considers data from the National Weather Service - NOAA Storm Data Archive for the years 1950 - 2011 for the purpose of answering 2 questions:
1. Across the United States, which types of events (as indicated in the EVTYPE variable) are most harmful with respect to population health? and
2. Across the United States, which types of events have the greatest economic consequences?
Based on the analysis documented below:
TORNADOS AND EXCESSIVE HEAT are the most harmful to the US populations.
FLOODING results in the most property damage, and FLOODING and DROUGHT result in the most crop damage.
These conclusions are based in the information in the database as downloaded in March 2017.
Data Processing
Load necessary libraries
## Loading required package: 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## Loading required package: lubridate##
## Attaching package: 'lubridate'## The following object is masked from 'package:base':
##
## date## Loading required package: knitrDownload and read the storm data into a data.frame
if(!file.exists("StormData.csv.bz2")){
download.file("https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2", "StormData.csv.bz2")
}
dfNOAA <- read.csv("StormData.csv.bz2", header=TRUE)Summarize Fatality and Injury totals by Year and Event Type
dfEvent <- dfNOAA %>% select(BGN_DATE, EVTYPE, FATALITIES, INJURIES) %>%
mutate(YEAR = year(as.Date(BGN_DATE, format="%m/%d/%Y %H:%M:%S"))) %>%
group_by(YEAR, EVTYPE) %>% summarize(sum(FATALITIES), sum(INJURIES))
names(dfEvent)<- c("YEAR", "EVTYPE", "Total_FATALITIES", 'Total_INJURIES')Summarize the number of Event Type classifications by Year
dfEventTypeCount <- data.frame(YEAR=unique(dfEvent$YEAR), NUM_EVTYPE = group_size(dfEvent))Select the Event Type with the most fatalities for a specific year
dfMaxFatalities <- dfEvent %>% group_by(YEAR) %>% filter(Total_FATALITIES == max(Total_FATALITIES)) %>% arrange(desc(YEAR),EVTYPE,Total_FATALITIES)Select the Event Type with the most injuries for a specific year
dfMaxInjuries <- dfEvent %>% group_by(YEAR) %>% filter(Total_INJURIES == max(Total_INJURIES)) %>% arrange(desc(YEAR),EVTYPE,Total_INJURIES)Calculating Total Property Damage for Each Event
The value of property damage are represented in the data is maintained in two fields PROPDMG and PROPDMGEXP. PROPDMGEXP is the exponent to be applied to the PROPDMG value. The expected values for the exponent are:
1. b, B = Billions
2. m, M = Millions
3. k, K = Thousands
4. h, H = Hundreds
However, there are additional values in the data:
unique(dfNOAA$PROPDMGEXP)## [1] K M B m + 0 5 6 ? 4 2 3 h 7 H - 1 8
## Levels: - ? + 0 1 2 3 4 5 6 7 8 B h H K m M?, -, +, {space} = 0 was used as the exponent. In most cases where these special characters appear, the PROPDMG value is 0, only in the 5 cases where the exponent is + are there no zero data in PROPDMG.
0 - 8: the numeric value was used as the exponent, e.g. 10^PROPDMGEXP
dfEventDamage <- dfNOAA %>% select(BGN_DATE, EVTYPE, PROPDMG, PROPDMGEXP, CROPDMG, CROPDMGEXP) %>%
mutate(YEAR = year(as.Date(BGN_DATE, format="%m/%d/%Y %H:%M:%S")), Total_PropertyDamage = 0, Total_CropDamage = 0)
dfEventDamage[dfEventDamage$PROPDMGEXP %in% c('+', '?', '-', '', '0'), 8] <- dfEventDamage$PROPDMG[dfEventDamage$PROPDMGEXP %in% c('+', '?', '-', '', '0')]
dfEventDamage[dfEventDamage$PROPDMGEXP %in% as.character(1:9), 8] <- dfEventDamage$PROPDMG[dfEventDamage$PROPDMGEXP %in% as.character(1:9)] * 10 ^ as.integer(as.character(dfEventDamage$PROPDMGEXP[dfEventDamage$PROPDMGEXP %in% as.character(1:9)]))
dfEventDamage[dfEventDamage$PROPDMGEXP %in% c("B", "b"), 8] <- dfEventDamage$PROPDMG[dfEventDamage$PROPDMGEXP %in% c('B', 'b')] * 10 ^ 9
dfEventDamage[dfEventDamage$PROPDMGEXP %in% c("M", "m"), 8] <- dfEventDamage$PROPDMG[dfEventDamage$PROPDMGEXP %in% c('M', 'm')] * 10 ^ 6
dfEventDamage[dfEventDamage$PROPDMGEXP %in% c("K", "k"), 8] <- dfEventDamage$PROPDMG[dfEventDamage$PROPDMGEXP %in% c('K', 'k')] * 10 ^ 3
dfEventDamage[dfEventDamage$PROPDMGEXP %in% c("H", "h"), 8] <- dfEventDamage$PROPDMG[dfEventDamage$PROPDMGEXP %in% c('H', 'h')] * 10 ^ 2Calcuating Total Crop Damage For Each Event
The value of crop damage are represented in the data is maintained in two fields CROPDMG and CROPDMGEXP. CROPDMGEXP is the exponent to be applied to the CROPDMG value. The expected values for the exponent are:
1. b, B = Billions
2. m, M = Millions
3. k, K = Thousands
4. h, H = Hundreds
However, there are additional values in the data:
unique(dfNOAA$CROPDMGEXP)## [1] M K m B ? 0 k 2
## Levels: ? 0 2 B k K m M?, -, +, {space} = 0 was used as the exponent.
0 - 8: the numeric value was used as the exponent, e.g. 10^PROPDMGEXP
dfEventDamage[dfEventDamage$CROPDMGEXP %in% c('+', '?', '-', '', '0'), 9] <- dfEventDamage$CROPDMG[dfEventDamage$CROPDMGEXP %in% c('+', '?', '-', '', '0')]
dfEventDamage[dfEventDamage$CROPDMGEXP %in% as.character(1:9), 9] <- dfEventDamage$CROPDMG[dfEventDamage$CROPDMGEXP %in% as.character(1:9)] * 10 ^ as.integer(as.character(dfEventDamage$CROPDMGEXP[dfEventDamage$CROPDMGEXP %in% as.character(1:9)]))
dfEventDamage[dfEventDamage$CROPDMGEXP %in% c("B", "b"), 9] <- dfEventDamage$CROPDMG[dfEventDamage$CROPDMGEXP %in% c('B', 'b')] * 10 ^ 9
dfEventDamage[dfEventDamage$CROPDMGEXP %in% c("M", "m"), 9] <- dfEventDamage$CROPDMG[dfEventDamage$CROPDMGEXP %in% c('M', 'm')] * 10 ^ 6
dfEventDamage[dfEventDamage$CROPDMGEXP %in% c("K", "k"), 9] <- dfEventDamage$CROPDMG[dfEventDamage$CROPDMGEXP %in% c('K', 'k')] * 10 ^ 3
dfEventDamage[dfEventDamage$CROPDMGEXP %in% c("H", "h"), 9] <- dfEventDamage$CROPDMG[dfEventDamage$CROPDMGEXP %in% c('H', 'h')] * 10 ^ 2Summarize Damage totals by Year and Event Type
dfEventDamage <- dfEventDamage %>% select(YEAR, EVTYPE, Total_PropertyDamage, Total_CropDamage) %>% group_by(YEAR, EVTYPE) %>% summarize(sum(Total_PropertyDamage), sum(Total_CropDamage))
names(dfEventDamage)<- c("YEAR", "EVTYPE", "Total_PropertyDamage", 'Total_CropDamage')Select the event type with the most Property damage for a specific year
dfMaxPropDamage <- dfEventDamage %>% group_by(YEAR) %>% filter( Total_PropertyDamage== max(Total_PropertyDamage) & Total_PropertyDamage > 0) %>% arrange(desc(YEAR),EVTYPE,Total_PropertyDamage)Select the Event Type with the most Crop damage for a specific Year
dfMaxCropDamage <- dfEventDamage %>% group_by(YEAR) %>% filter( Total_CropDamage== max(Total_CropDamage) & Total_CropDamage > 0) %>% arrange(desc(YEAR),EVTYPE,Total_CropDamage)Data Exploration
The Number of Event type classifications have changed significantly over the years
Summary of event type count by year:
- For the years 1950 - 1954, there is only 1 event type - Tornados
- For the years 1955 - 1992, there are 3 event types - Tornados, Hail, Tsunami Wind
- For the years 1993 - 2002, there are between 99 and 387 event types
- For the years 2003 - 2011, there are between 38 and 51 event types
Given the change in event classificaitons, it will be difficult to compare data across these 4 periods. For example, to more accurately compare across the last two periods of year listed above, some additional mapping of event types is necessary because the level of granularity in the classification for years 1993 - 2002 is much higher than for 2003 - 2011. For the purpose of this initial analysis, this mapping is NOT present.
plot(dfEventTypeCount$YEAR, dfEventTypeCount$NUM_EVTYPE, main="Event Classification Has Changed Significantly From 1950 to 2011", xlab="Year", ylab = "Number of Event Types")
# Create Year Ranges representing these periods
YearPeriod1 <- seq.int(1950, 1992, 1)
YearPeriod2 <- seq.int(1993, 2002, 1)
YearPeriod3 <- seq.int(2003, 2011, 1)Fatalities:
For the most recent period 2003 - 2011
TORNADOS are the most fatal, followed by EXCESSIVE HEAT
kable(dfMaxFatalities[dfMaxFatalities$YEAR %in% YearPeriod3, c(1,2,3)])| YEAR | EVTYPE | Total_FATALITIES |
|---|---|---|
| 2011 | TORNADO | 587 |
| 2010 | FLASH FLOOD | 67 |
| 2009 | RIP CURRENT | 39 |
| 2008 | TORNADO | 129 |
| 2007 | TORNADO | 81 |
| 2006 | EXCESSIVE HEAT | 205 |
| 2005 | EXCESSIVE HEAT | 158 |
| 2004 | FLASH FLOOD | 58 |
| 2003 | FLASH FLOOD | 69 |
dfSumFatalities3 <- dfMaxFatalities[dfMaxFatalities$YEAR %in% YearPeriod3, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_FATALITIES))
names(dfSumFatalities3) <- c("EVTYPE", "SumTotal_FATALITIES")
dfSumFatalities3 <- arrange(dfSumFatalities3, desc(SumTotal_FATALITIES))
kable(dfSumFatalities3)| EVTYPE | SumTotal_FATALITIES |
|---|---|
| TORNADO | 797 |
| EXCESSIVE HEAT | 363 |
| FLASH FLOOD | 194 |
| RIP CURRENT | 39 |
For the period 1993 - 2002
EXCESSIVE HEAT / HEAT are the most fatal, followed by TORNADOS It is necessary to aggregate Event Types in order to see all the Tornado events
kable(dfMaxFatalities[dfMaxFatalities$YEAR %in% YearPeriod2, c(1,2,3)])| YEAR | EVTYPE | Total_FATALITIES |
|---|---|---|
| 2002 | EXCESSIVE HEAT | 167 |
| 2001 | EXCESSIVE HEAT | 165 |
| 2000 | EXCESSIVE HEAT | 157 |
| 1999 | EXCESSIVE HEAT | 500 |
| 1998 | EXCESSIVE HEAT | 168 |
| 1997 | EXCESSIVE HEAT | 80 |
| 1996 | FLASH FLOOD | 92 |
| 1995 | HEAT | 687 |
| 1994 | LIGHTNING | 65 |
| 1993 | BLIZZARD | 30 |
dfSumFatalities2 <- dfMaxFatalities[dfMaxFatalities$YEAR %in% YearPeriod2, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_FATALITIES))
names(dfSumFatalities2) <- c("EVTYPE", "SumTotal_FATALITIES")
dfSumFatalities2 <- arrange(dfSumFatalities2, desc(SumTotal_FATALITIES))
kable(dfSumFatalities2)| EVTYPE | SumTotal_FATALITIES |
|---|---|
| EXCESSIVE HEAT | 1237 |
| HEAT | 687 |
| FLASH FLOOD | 92 |
| LIGHTNING | 65 |
| BLIZZARD | 30 |
Tornados did not appear at the top, as a result of using a classificaiton system that has multi-event types for Tornado.
paste0("Total Tornado events for this period: ", as.character(sum(dfEvent$Total_FATALITIES[dfEvent$YEAR %in% YearPeriod2 & grepl("tornado", dfEvent$EVTYPE, ignore.case = TRUE)])))## [1] "Total Tornado events for this period: 592"For the period 1950 - 1992
TORNADOS are the most fatal, followed by TSUNAMI WIND in 1986; however, there are only 3 event types for this period, Tornados, Hail, and Tsunami Wind
kable(dfMaxFatalities[dfMaxFatalities$YEAR %in% YearPeriod1, c(1,2,3)])| YEAR | EVTYPE | Total_FATALITIES |
|---|---|---|
| 1992 | TORNADO | 39 |
| 1991 | TORNADO | 39 |
| 1990 | TORNADO | 53 |
| 1989 | TORNADO | 50 |
| 1988 | TORNADO | 32 |
| 1987 | TORNADO | 59 |
| 1986 | TSTM WIND | 36 |
| 1985 | TORNADO | 94 |
| 1984 | TORNADO | 122 |
| 1983 | TORNADO | 34 |
| 1982 | TORNADO | 64 |
| 1981 | TORNADO | 24 |
| 1980 | TORNADO | 28 |
| 1979 | TORNADO | 84 |
| 1978 | TORNADO | 53 |
| 1977 | TORNADO | 43 |
| 1976 | TORNADO | 44 |
| 1975 | TORNADO | 60 |
| 1974 | TORNADO | 366 |
| 1973 | TORNADO | 89 |
| 1972 | TORNADO | 27 |
| 1971 | TORNADO | 159 |
| 1970 | TORNADO | 73 |
| 1969 | TORNADO | 66 |
| 1968 | TORNADO | 131 |
| 1967 | TORNADO | 114 |
| 1966 | TORNADO | 98 |
| 1965 | TORNADO | 301 |
| 1964 | TORNADO | 73 |
| 1963 | TORNADO | 31 |
| 1962 | TORNADO | 30 |
| 1961 | TORNADO | 52 |
| 1960 | TORNADO | 46 |
| 1959 | TORNADO | 58 |
| 1958 | TORNADO | 67 |
| 1957 | TORNADO | 193 |
| 1956 | TORNADO | 83 |
| 1955 | TORNADO | 129 |
| 1954 | TORNADO | 36 |
| 1953 | TORNADO | 519 |
| 1952 | TORNADO | 230 |
| 1951 | TORNADO | 34 |
| 1950 | TORNADO | 70 |
dfSumFatalities1 <- dfMaxFatalities[dfMaxFatalities$YEAR %in% YearPeriod1, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_FATALITIES))
names(dfSumFatalities1) <- c("EVTYPE", "SumTotal_FATALITIES")
dfSumFatalities1 <- arrange(dfSumFatalities1, desc(SumTotal_FATALITIES))
kable(dfSumFatalities1)| EVTYPE | SumTotal_FATALITIES |
|---|---|
| TORNADO | 3997 |
| TSTM WIND | 36 |
Injuries:
For the most recent period 2003 - 2011
TORNADOS are the most injurous, followed by EXCESSIVE HEAT, then by HURRICANE/TYPHOON
kable(dfMaxInjuries[dfMaxInjuries$YEAR %in% YearPeriod3, c(1,2,4)])| YEAR | EVTYPE | Total_INJURIES |
|---|---|---|
| 2011 | TORNADO | 6163 |
| 2010 | TORNADO | 699 |
| 2009 | TORNADO | 397 |
| 2008 | TORNADO | 1690 |
| 2007 | TORNADO | 659 |
| 2006 | EXCESSIVE HEAT | 993 |
| 2005 | TORNADO | 537 |
| 2004 | HURRICANE/TYPHOON | 839 |
| 2003 | TORNADO | 1087 |
dfSumInjuries3 <- dfMaxInjuries[dfMaxInjuries$YEAR %in% YearPeriod3, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_INJURIES))
names(dfSumInjuries3) <- c("EVTYPE", "SumTotal_INJURIES")
dfSumInjuries3 <- arrange(dfSumInjuries3, desc(SumTotal_INJURIES))
kable(dfSumInjuries3)| EVTYPE | SumTotal_INJURIES |
|---|---|
| TORNADO | 11232 |
| EXCESSIVE HEAT | 993 |
| HURRICANE/TYPHOON | 839 |
For the period 1993 - 2002
TORNADOS are the most injurious, followed by FLOODS and by ICE STORMS
kable(dfMaxInjuries[dfMaxInjuries$YEAR %in% YearPeriod2, c(1,2,4)])| YEAR | EVTYPE | Total_INJURIES |
|---|---|---|
| 2002 | TORNADO | 968 |
| 2001 | TORNADO | 743 |
| 2000 | TORNADO | 882 |
| 1999 | TORNADO | 1842 |
| 1998 | FLOOD | 6130 |
| 1997 | TORNADO | 1033 |
| 1996 | TORNADO | 705 |
| 1995 | TORNADO | 1098 |
| 1994 | ICE STORM | 1614 |
| 1993 | TORNADO | 739 |
dfSumInjuries2 <- dfMaxInjuries[dfMaxInjuries$YEAR %in% YearPeriod2, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_INJURIES))
names(dfSumInjuries2) <- c("EVTYPE", "SumTotal_INJURIES")
dfSumInjuries2 <- arrange(dfSumInjuries2, desc(SumTotal_INJURIES))
kable(dfSumInjuries2)| EVTYPE | SumTotal_INJURIES |
|---|---|
| TORNADO | 8010 |
| FLOOD | 6130 |
| ICE STORM | 1614 |
Tornados did appear at the top, but as a result of using a classificaiton system that has multi-event types for Tornado, not all injuries are represented above.
Total Fatalities for all Event types that contain “tornado” case insensitive
paste0("Total Tornado events for this period: ", as.character(sum(dfEvent$Total_INJURIES[dfEvent$YEAR %in% YearPeriod2 & grepl("tornado", dfEvent$EVTYPE, ignore.case = TRUE)])))## [1] "Total Tornado events for this period: 10751"For the period 1950 - 1992
TORNADOS are the most injurious; however, there are only 3 event types for this period, Tornados, Hail, and Tsunami Wind
kable(dfMaxInjuries[dfMaxInjuries$YEAR %in% YearPeriod1, c(1,2,4)])| YEAR | EVTYPE | Total_INJURIES |
|---|---|---|
| 1992 | TORNADO | 1323 |
| 1991 | TORNADO | 864 |
| 1990 | TORNADO | 1177 |
| 1989 | TORNADO | 1270 |
| 1988 | TORNADO | 688 |
| 1987 | TORNADO | 1018 |
| 1986 | TORNADO | 536 |
| 1985 | TORNADO | 1299 |
| 1984 | TORNADO | 2499 |
| 1983 | TORNADO | 756 |
| 1982 | TORNADO | 1276 |
| 1981 | TORNADO | 798 |
| 1980 | TORNADO | 1157 |
| 1979 | TORNADO | 3014 |
| 1978 | TORNADO | 919 |
| 1977 | TORNADO | 771 |
| 1976 | TORNADO | 1195 |
| 1975 | TORNADO | 1457 |
| 1974 | TORNADO | 6824 |
| 1973 | TORNADO | 2406 |
| 1972 | TORNADO | 976 |
| 1971 | TORNADO | 2723 |
| 1970 | TORNADO | 1355 |
| 1969 | TORNADO | 1311 |
| 1968 | TORNADO | 2522 |
| 1967 | TORNADO | 2144 |
| 1966 | TORNADO | 2030 |
| 1965 | TORNADO | 5197 |
| 1964 | TORNADO | 1148 |
| 1963 | TORNADO | 538 |
| 1962 | TORNADO | 551 |
| 1961 | TORNADO | 1087 |
| 1960 | TORNADO | 737 |
| 1959 | TORNADO | 734 |
| 1958 | TORNADO | 535 |
| 1957 | TORNADO | 1976 |
| 1956 | TORNADO | 1355 |
| 1955 | TORNADO | 926 |
| 1954 | TORNADO | 715 |
| 1953 | TORNADO | 5131 |
| 1952 | TORNADO | 1915 |
| 1951 | TORNADO | 524 |
| 1950 | TORNADO | 659 |
dfSumInjuries1 <- dfMaxInjuries[dfMaxInjuries$YEAR %in% YearPeriod1, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_INJURIES))
names(dfSumInjuries1) <- c("EVTYPE", "SumTotal_INJURIES")
dfSumInjuries1 <- arrange(dfSumInjuries1, desc(SumTotal_INJURIES))
kable(dfSumInjuries1)| EVTYPE | SumTotal_INJURIES |
|---|---|
| TORNADO | 68036 |
Property Damage:
For the most recent period 2003 - 2011
FLOODING is the most costly, followed by HURRICANE/TYPHOON
kable(dfMaxPropDamage[dfMaxPropDamage$YEAR %in% YearPeriod3, c(1,2,3)], format.args = list(big.mark = ','))| YEAR | EVTYPE | Total_PropertyDamage |
|---|---|---|
| 2,011 | TORNADO | 9,819,600,700 |
| 2,010 | HAIL | 3,404,109,950 |
| 2,009 | HAIL | 1,440,000,550 |
| 2,008 | STORM SURGE/TIDE | 4,595,035,000 |
| 2,007 | TORNADO | 1,400,756,650 |
| 2,006 | FLOOD | 116,516,105,440 |
| 2,005 | HURRICANE/TYPHOON | 49,786,635,000 |
| 2,004 | HURRICANE/TYPHOON | 18,356,435,000 |
| 2,003 | WILDFIRE | 2,331,537,330 |
dfSumPropDamage3 <- dfMaxPropDamage[dfMaxPropDamage$YEAR %in% YearPeriod3, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_PropertyDamage))
names(dfSumPropDamage3) <- c("EVTYPE", "SumTotal_PROPERTYDAMAGE")
dfSumPropDamage3 <- arrange(dfSumPropDamage3, desc(SumTotal_PROPERTYDAMAGE))
kable(dfSumPropDamage3, format.args = list(big.mark = ','))| EVTYPE | SumTotal_PROPERTYDAMAGE |
|---|---|
| FLOOD | 116,516,105,440 |
| HURRICANE/TYPHOON | 68,143,070,000 |
| TORNADO | 11,220,357,350 |
| HAIL | 4,844,110,500 |
| STORM SURGE/TIDE | 4,595,035,000 |
| WILDFIRE | 2,331,537,330 |
For the period 1993 - 2002
Flooding is the most costly, followed by Storms then by Hurricanes It is necesary to aggregate similar EVTYPE descriptions in order to reach this conclusion…
kable(dfMaxPropDamage[dfMaxPropDamage$YEAR %in% YearPeriod2, c(1,2,3)], format.args = list(big.mark = ','))| YEAR | EVTYPE | Total_PropertyDamage |
|---|---|---|
| 2,002 | TORNADO | 801,254,200 |
| 2,001 | TROPICAL STORM | 5,185,910,000 |
| 2,000 | WILD/FOREST FIRE | 2,109,242,200 |
| 1,999 | HURRICANE | 4,162,191,000 |
| 1,998 | HURRICANE | 3,159,132,000 |
| 1,997 | FLOOD | 6,124,621,200 |
| 1,996 | HURRICANE | 1,382,190,000 |
| 1,995 | HURRICANE OPAL | 3,172,846,000 |
| 1,994 | TORNADO | 694,648,856 |
| 1,993 | WINTER STORM | 5,128,400,000 |
dfSumPropDamage2 <- dfMaxPropDamage[dfMaxPropDamage$YEAR %in% YearPeriod2, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_PropertyDamage))
names(dfSumPropDamage2) <- c("EVTYPE", "SumTotal_PROPERTYDAMAGE")
dfSumPropDamage2 <- arrange(dfSumPropDamage2, desc(SumTotal_PROPERTYDAMAGE))
kable(dfSumPropDamage2, format.args = list(big.mark = ','))| EVTYPE | SumTotal_PROPERTYDAMAGE |
|---|---|
| HURRICANE | 8,703,513,000 |
| FLOOD | 6,124,621,200 |
| TROPICAL STORM | 5,185,910,000 |
| WINTER STORM | 5,128,400,000 |
| HURRICANE OPAL | 3,172,846,000 |
| WILD/FOREST FIRE | 2,109,242,200 |
| TORNADO | 1,495,903,056 |
Hurricanes and Storms appear on the list twice as a result of using a classification system that has multi-event types for Hurricane. In addition, damage from Flooding and Tornados are also under represented in the above totals.
Total Property Damage for all Event types that contain “hurricane” case insensitive
paste0("Total property damage for Flooding events for this period: ", as.character(format(sum(dfEventDamage$Total_PropertyDamage[dfEventDamage$YEAR %in% YearPeriod2 & grepl("flood", dfEvent$EVTYPE, ignore.case = TRUE)]), big.mark = ',' )))## [1] "Total property damage for Flooding events for this period: 24,543,692,335"paste0("Total property damage for Storm events for this period: ", as.character(format(sum(dfEventDamage$Total_PropertyDamage[dfEventDamage$YEAR %in% YearPeriod2 & grepl("storm", dfEvent$EVTYPE, ignore.case = TRUE)]), big.mark = ',' )))## [1] "Total property damage for Storm events for this period: 17,683,357,176"paste0("Total property damage for Hurricane events for this period: ", as.character(format(sum(dfEventDamage$Total_PropertyDamage[dfEventDamage$YEAR %in% YearPeriod2 & grepl("hurricane", dfEvent$EVTYPE, ignore.case = TRUE)]), big.mark = ',')))## [1] "Total property damage for Hurricane events for this period: 13,157,740,010"paste0("Total property damage for Tornado events for this period: ", as.character(format(sum(dfEventDamage$Total_PropertyDamage[dfEventDamage$YEAR %in% YearPeriod2 & grepl("tornado", dfEvent$EVTYPE, ignore.case = TRUE)]), big.mark = ',' )))## [1] "Total property damage for Tornado events for this period: 10,399,450,897"For the period 1950 - 1992
Tornados are the most costly; however, there are only 3 event types classified for these years.
kable(dfMaxPropDamage[dfMaxPropDamage$YEAR %in% YearPeriod1, c(1,2,3)], format.args = list(big.mark = ","))| YEAR | EVTYPE | Total_PropertyDamage |
|---|---|---|
| 1,992 | TORNADO | 1,406,688,900 |
| 1,991 | TORNADO | 1,152,296,430 |
| 1,990 | TORNADO | 1,560,568,070 |
| 1,989 | TORNADO | 1,760,051,050 |
| 1,988 | TORNADO | 1,124,214,650 |
| 1,987 | TORNADO | 407,837,950 |
| 1,986 | TORNADO | 1,034,912,340 |
| 1,985 | TORNADO | 1,608,097,210 |
| 1,984 | TORNADO | 2,131,171,390 |
| 1,983 | TORNADO | 765,855,590 |
| 1,982 | TORNADO | 1,429,218,000 |
| 1,981 | TORNADO | 765,888,310 |
| 1,980 | TORNADO | 2,152,814,350 |
| 1,979 | TORNADO | 946,241,240 |
| 1,978 | TORNADO | 616,068,160 |
| 1,977 | TORNADO | 342,024,230 |
| 1,976 | TORNADO | 308,497,190 |
| 1,975 | TORNADO | 826,723,790 |
| 1,974 | TORNADO | 1,943,517,780 |
| 1,973 | TORNADO | 2,063,066,560 |
| 1,972 | TORNADO | 151,172,570 |
| 1,971 | TORNADO | 202,120,550 |
| 1,970 | TORNADO | 625,205,980 |
| 1,969 | TORNADO | 113,274,230 |
| 1,968 | TORNADO | 210,079,910 |
| 1,967 | TORNADO | 577,019,220 |
| 1,966 | TORNADO | 476,629,360 |
| 1,965 | TORNADO | 1,762,499,040 |
| 1,964 | TORNADO | 165,990,820 |
| 1,963 | TORNADO | 87,230,580 |
| 1,962 | TORNADO | 67,200,730 |
| 1,961 | TORNADO | 179,388,730 |
| 1,960 | TORNADO | 118,224,240 |
| 1,959 | TORNADO | 87,453,040 |
| 1,958 | TORNADO | 128,994,610 |
| 1,957 | TORNADO | 224,388,890 |
| 1,956 | TORNADO | 116,912,350 |
| 1,955 | TORNADO | 82,660,630 |
| 1,954 | TORNADO | 85,805,320 |
| 1,953 | TORNADO | 596,104,700 |
| 1,952 | TORNADO | 94,102,240 |
| 1,951 | TORNADO | 65,505,990 |
| 1,950 | TORNADO | 34,481,650 |
dfSumPropDamage1 <- dfMaxPropDamage[dfMaxPropDamage$YEAR %in% YearPeriod1, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_PropertyDamage))
names(dfSumPropDamage1) <- c("EVTYPE", "SumTotal_PROPERTYDAMAGE")
dfSumPropDamage1 <- arrange(dfSumPropDamage1, desc(SumTotal_PROPERTYDAMAGE))
kable(dfSumPropDamage1, format.args = list(big.mark = ","))| EVTYPE | SumTotal_PROPERTYDAMAGE |
|---|---|
| TORNADO | 30,598,198,570 |
Crop Damage:
For the most recent period 2003 - 2011
DROUGHT, FLOOD, HURRICANE/TYPHOON are the most costly.
kable(dfMaxCropDamage[dfMaxCropDamage$YEAR %in% YearPeriod3, c(1,2,4)], format.args = list(big.mark = ','))| YEAR | EVTYPE | Total_CropDamage |
|---|---|---|
| 2,011 | FLOOD | 154,872,000 |
| 2,010 | FLOOD | 1,104,709,000 |
| 2,009 | HAIL | 350,994,000 |
| 2,008 | FLOOD | 1,078,292,000 |
| 2,007 | FLOOD | 518,507,000 |
| 2,006 | DROUGHT | 2,375,825,000 |
| 2,005 | HURRICANE/TYPHOON | 2,012,682,000 |
| 2,004 | HURRICANE/TYPHOON | 565,820,800 |
| 2,003 | DROUGHT | 572,480,000 |
dfSumCropDamage3 <- dfMaxCropDamage[dfMaxCropDamage$YEAR %in% YearPeriod3, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_CropDamage))
names(dfSumCropDamage3) <- c("EVTYPE", "SumTotal_CROPDAMAGE")
dfSumCropDamage3 <- arrange(dfSumCropDamage3, desc(SumTotal_CROPDAMAGE))
kable(dfSumCropDamage3, format.args = list(big.mark = ','))| EVTYPE | SumTotal_CROPDAMAGE |
|---|---|
| DROUGHT | 2,948,305,000 |
| FLOOD | 2,856,380,000 |
| HURRICANE/TYPHOON | 2,578,502,800 |
| HAIL | 350,994,000 |
For the period 1993 - 2002
FLOODING and DROUGHT are the most costly, followed by ICE STORMS It is necesary to aggregate similar EVTYPE descriptions in order to reach this conclusion…
kable(dfMaxCropDamage[dfMaxCropDamage$YEAR %in% YearPeriod2, c(1,2,4)], format.args = list(big.mark = ','))| YEAR | EVTYPE | Total_CropDamage |
|---|---|---|
| 2,002 | DROUGHT | 737,634,000 |
| 2,001 | DROUGHT | 1,273,860,000 |
| 2,000 | DROUGHT | 2,438,100,000 |
| 1,999 | HURRICANE | 1,378,875,000 |
| 1,998 | DROUGHT | 2,141,971,000 |
| 1,997 | DROUGHT | 252,990,000 |
| 1,996 | DROUGHT | 504,080,000 |
| 1,995 | DROUGHT | 554,500,000 |
| 1,994 | ICE STORM | 5,006,453,500 |
| 1,993 | RIVER FLOOD | 5,005,050,000 |
dfSumCropDamage2 <- dfMaxCropDamage[dfMaxCropDamage$YEAR %in% YearPeriod2, ] %>% group_by(EVTYPE) %>% summarize(sum(Total_CropDamage))
names(dfSumCropDamage2) <- c("EVTYPE", "SumTotal_CROPDAMAGE")
dfSumCropDamage2 <- arrange(dfSumCropDamage2, desc(SumTotal_CROPDAMAGE))
kable(dfSumCropDamage2, format.args = list(big.mark = ','))| EVTYPE | SumTotal_CROPDAMAGE |
|---|---|
| DROUGHT | 7,903,135,000 |
| ICE STORM | 5,006,453,500 |
| RIVER FLOOD | 5,005,050,000 |
| HURRICANE | 1,378,875,000 |
Damage from flooding and hurricanes are unders represented in the above totals, as a result of using a classification system that has multi-event types for Hurricane and Flood
paste0("Total crop damage for Flooding events for this period: ", as.character(format(sum(dfEventDamage$Total_CropDamage[dfEventDamage$YEAR %in% YearPeriod2 & grepl("flood", dfEvent$EVTYPE, ignore.case = TRUE)]), big.mark = ',' )))## [1] "Total crop damage for Flooding events for this period: 8,058,187,200"paste0("Total crop damage for Hurricane events for this period: ", as.character(format(sum(dfEventDamage$Total_CropDamage[dfEventDamage$YEAR %in% YearPeriod2 & grepl("hurricane", dfEvent$EVTYPE, ignore.case = TRUE)]), big.mark = ',')))## [1] "Total crop damage for Hurricane events for this period: 2,726,910,000"For the period 1950 - 1992 - No Crop Damage data is present in the dataset
Results
Which Events are the Most Harmful with Respect to Population Health?
With respect to Fatalities per year, TORNADOS and EXCESSIVE HEAT are the most harmful. With Tornados being the most harmful in recent years.
With respect to Injuries per year, TORNADOS are consistently the most harmful across all years.
lstYearBeginRange <- c(range(YearPeriod3)[1], range(YearPeriod3)[1], range(YearPeriod2)[1], range(YearPeriod2)[1], range(YearPeriod1)[1])
lstYearEndRange <- c(range(YearPeriod3)[2], range(YearPeriod3)[2], range(YearPeriod2)[2], range(YearPeriod2)[2], range(YearPeriod1)[2])
lstFatalEvents <- c(as.character(dfSumFatalities3$EVTYPE[1]), as.character(dfSumFatalities3$EVTYPE[2]), as.character(dfSumFatalities2$EVTYPE[1]), "TORNADO", as.character(dfSumFatalities1$EVTYPE[1]))
lstFatalities <- c(dfSumFatalities3$SumTotal_FATALITIES[1]/length(YearPeriod3), dfSumFatalities3$SumTotal_FATALITIES[2]/length(YearPeriod3), dfSumFatalities2$SumTotal_FATALITIES[1]/length(YearPeriod2), sum(dfEvent$Total_FATALITIES[dfEvent$YEAR %in% YearPeriod2 & grepl("tornado", dfEvent$EVTYPE, ignore.case = TRUE)]) / length(YearPeriod2), dfSumFatalities1$SumTotal_FATALITIES[1]/length(YearPeriod1))
dfFatalities <- data.frame(lstYearBeginRange, lstYearEndRange, lstFatalEvents, lstFatalities)
names(dfFatalities) <- c('Period Start', 'Period End', 'Event Type', 'Fatalities Per Year')
kable(dfFatalities, format.args = list(digits=0), caption = "Tornados and Excessive Heat Result in the Most Fatalities Per Year")| Period Start | Period End | Event Type | Fatalities Per Year |
|---|---|---|---|
| 2003 | 2011 | TORNADO | 89 |
| 2003 | 2011 | EXCESSIVE HEAT | 40 |
| 1993 | 2002 | EXCESSIVE HEAT | 124 |
| 1993 | 2002 | TORNADO | 59 |
| 1950 | 1992 | TORNADO | 93 |
lstInjuryEvents <- c(as.character(dfSumInjuries3$EVTYPE[1]), as.character(dfSumInjuries3$EVTYPE[2]), as.character(dfSumInjuries2$EVTYPE[1]), as.character(dfSumInjuries2$EVTYPE[2]), as.character(dfSumInjuries1$EVTYPE[1]))
lstInjuries <- c(dfSumInjuries3$SumTotal_INJURIES[1]/length(YearPeriod3), dfSumInjuries3$SumTotal_INJURIES[2]/length(YearPeriod3), sum(dfEvent$Total_INJURIES[dfEvent$YEAR %in% YearPeriod2 & grepl("tornado", dfEvent$EVTYPE, ignore.case = TRUE)]) / length(YearPeriod2), dfSumInjuries2$SumTotal_INJURIES[2]/length(YearPeriod2), dfSumInjuries1$SumTotal_INJURIES[1]/length(YearPeriod1))
dfInjuries <- data.frame(lstYearBeginRange, lstYearEndRange, lstInjuryEvents, lstInjuries)
names(dfInjuries) <- c('Period Start', 'Period End', 'Event Type', 'Injuries Per Year')
kable(dfInjuries, format.args = list(digits=0), caption = "Tornados Result in the Most Injuries Per Year")| Period Start | Period End | Event Type | Injuries Per Year |
|---|---|---|---|
| 2003 | 2011 | TORNADO | 1248 |
| 2003 | 2011 | EXCESSIVE HEAT | 110 |
| 1993 | 2002 | TORNADO | 1075 |
| 1993 | 2002 | FLOOD | 613 |
| 1950 | 1992 | TORNADO | 1582 |
Which Events have the greatest Economic consequences?
With respect to Property Damage per year, FLOODING is the most costly, followed by HURRICANES/TYPHOONS and Storms - across multiple varieties.
With respect to Crop Damage per year, FLOODING and DROUGHT are the most costly.
lstYearBeginRange <- c(range(YearPeriod3)[1], range(YearPeriod3)[1], range(YearPeriod2)[1], range(YearPeriod2)[1], range(YearPeriod2)[1], range(YearPeriod1)[1])
lstYearEndRange <- c(range(YearPeriod3)[2], range(YearPeriod3)[2], range(YearPeriod2)[2], range(YearPeriod2)[2], range(YearPeriod2)[2], range(YearPeriod1)[2])
lstPropDamageEvents <- c(as.character(dfSumPropDamage3$EVTYPE[1]), as.character(dfSumPropDamage3$EVTYPE[2]), "FLOOD", "STORM", "HURRICANE", as.character(dfSumPropDamage1$EVTYPE[1]))
lstPropDamage <- c(dfSumPropDamage3$SumTotal_PROPERTYDAMAGE[1]/length(YearPeriod3), dfSumPropDamage3$SumTotal_PROPERTYDAMAGE[2]/length(YearPeriod3), sum(dfEventDamage$Total_PropertyDamage[dfEvent$YEAR %in% YearPeriod2 & grepl("flood", dfEvent$EVTYPE, ignore.case = TRUE)])/length(YearPeriod2), sum(dfEventDamage$Total_PropertyDamage[dfEvent$YEAR %in% YearPeriod2 & grepl("storm", dfEvent$EVTYPE, ignore.case = TRUE)]) / length(YearPeriod2), sum(dfEventDamage$Total_PropertyDamage[dfEvent$YEAR %in% YearPeriod2 & grepl("hurricane", dfEvent$EVTYPE, ignore.case = TRUE)]) / length(YearPeriod2), dfSumPropDamage1$SumTotal_PROPERTYDAMAGE[1]/length(YearPeriod1))
dfPropDamage <- data.frame(lstYearBeginRange, lstYearEndRange, lstPropDamageEvents, lstPropDamage)
names(dfPropDamage) <- c('Period Start', 'Period End', 'Event Type', 'Property Damage Per Year')
kable(dfPropDamage, format.args = list(digits=0), caption = "Flooding Results in the Most Property Damanage Per Year")| Period Start | Period End | Event Type | Property Damage Per Year |
|---|---|---|---|
| 2003 | 2011 | FLOOD | 1.294623e+10 |
| 2003 | 2011 | HURRICANE/TYPHOON | 7.571452e+09 |
| 1993 | 2002 | FLOOD | 2.454369e+09 |
| 1993 | 2002 | STORM | 1.768336e+09 |
| 1993 | 2002 | HURRICANE | 1.315774e+09 |
| 1950 | 1992 | TORNADO | 7.115860e+08 |
# Crop Damage Table of Results
lstYearBeginRange <- c(range(YearPeriod3)[1], range(YearPeriod3)[1], range(YearPeriod2)[1], range(YearPeriod2)[1])
lstYearEndRange <- c(range(YearPeriod3)[2], range(YearPeriod3)[2], range(YearPeriod2)[2], range(YearPeriod2)[2])
lstCropDamageEvents <- c(as.character(dfSumCropDamage3$EVTYPE[1]), as.character(dfSumCropDamage3$EVTYPE[2]), "FLOOD", "DROUGHT")
lstCropDamage <- c(dfSumCropDamage3$SumTotal_CROPDAMAGE[1]/length(YearPeriod3), dfSumCropDamage3$SumTotal_CROPDAMAGE[2]/length(YearPeriod3), sum(dfEventDamage$Total_CropDamage[dfEvent$YEAR %in% YearPeriod2 & grepl("flood", dfEvent$EVTYPE, ignore.case = TRUE)])/length(YearPeriod2), sum(dfEventDamage$Total_CropDamage[dfEvent$YEAR %in% YearPeriod2 & grepl("drought", dfEvent$EVTYPE, ignore.case = TRUE)]) / length(YearPeriod2))
dfCropDamage <- data.frame(lstYearBeginRange, lstYearEndRange, lstCropDamageEvents, lstCropDamage)
names(dfCropDamage) <- c('Period Start', 'Period End', 'Event Type', 'Crop Damage Per Year')
kable(dfCropDamage, format.args = list(digits=0), caption = "Flooding and Drought Result in the Most Crop Damage Per Year")| Period Start | Period End | Event Type | Crop Damage Per Year |
|---|---|---|---|
| 2003 | 2011 | DROUGHT | 3.275894e+08 |
| 2003 | 2011 | FLOOD | 3.173756e+08 |
| 1993 | 2002 | FLOOD | 8.058187e+08 |
| 1993 | 2002 | DROUGHT | 9.286631e+08 |