Synopsis

The data for this assignment can be downloaded from the course web site: Storm Data.

This analysis report aims to answer the following 2 questions:

  1. Across the United States, which types of events (as indicated in the EVTYPE) are most harmful with respect to population health?
  2. Across the United States, which types of events have the greatest economic consequences?

During the data processing, the following steps are conducted:

  1. Subset relevant columns for analysis from raw data file, extract month information from BGN_DATE, calculate economic damage form property damage (PROPDMG, PROPDMGEXP) and crop damage (CROPDMG, CROPDMGEXP), clean and merge similar event types (EVTYPE) which are recorded in different ways.
  2. Calculate fatalities and injuries by event type in each month and top 5 event types of each month are annotated in detail, which shows differential distribution pattern in seasons but generally tornado is the most harmful to population health.
  3. Calculate economic damages by event type in each month and top 5 event types of each month are annotated in detail, which shows differential distribution pattern in seasons but generally flood causes the greatest economic consequences.

1. Setup

Load packages and set language of system time.

library(dplyr)
library(ggplot2)
Sys.setlocale("LC_TIME","en_US.UTF-8")

2. Download and read raw data file

setwd("D:/GXY's documents/R/5-ReproducibleResearch/week4")
url <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
download.file(url,"repdata_data_StormData.csv",method="curl")
data <- read.csv("repdata_data_StormData.csv",header=TRUE)

3. Data Processing

3-1. Pre-processing

Subset relevant columns for later analysis.
Extract the month from BGN_DATE.

data <- subset(data,select=c(BGN_DATE,EVTYPE,FATALITIES,INJURIES,PROPDMG,PROPDMGEXP,CROPDMG,CROPDMGEXP))
data <- mutate(data,month=format(strptime(data$BGN_DATE,"%m/%d/%Y %H:%M:%S"),"%m"))

Format PROPDMGEXP and CROPDMGEXP in an uniform way and calculate economic damage (PROPDMG * PROPDMGEXP + CROPDMG * CROPDMGEXP).

data <- mutate(data,PROPDMGEXP_number=PROPDMGEXP,CROPDMGEXP_number=CROPDMGEXP)
data$PROPDMGEXP_number <- gsub("1",10,data$PROPDMGEXP_number) %>%
  gsub("2",100,.) %>%
  gsub("3",1000,.) %>% 
  gsub("4",10000,.) %>% 
  gsub("5",100000,.) %>%
  gsub("6",1000000,.) %>%
  gsub("7",10000000,.) %>%
  gsub("8",100000000,.) %>%
  gsub("^0$",1,.) %>%
  gsub("[Hh]",100,.) %>%
  gsub("K",1000,.) %>%
  gsub("[Mm]",1000000,.) %>%
  gsub("B",1000000000,.) %>%
  gsub("(^$)|-|(\\+)|(\\?)",0,.) 
data$CROPDMGEXP_number <- gsub("2",100,data$CROPDMGEXP_number) %>%
  gsub("^0$",1,.) %>%
  gsub("[Kk]",1000,.) %>%
  gsub("[Mm]",1000000,.) %>%
  gsub("B",1000000000,.) %>%
  gsub("(^$)|(\\?)",0,.) 
data <- mutate(data,economic_damage=PROPDMG*as.integer(PROPDMGEXP_number)+CROPDMG*as.integer(CROPDMGEXP_number))

Format and merge EVTYPE (upper/lower cases, misspelling, extra space ” ” or other characters, synonymous phrases, etc.).

data <- mutate(data,event_type=trimws(toupper(EVTYPE)))
data$event_type <- gsub("(\\.|/|;|-)$","",data$event_type) %>%
  gsub("AVALANCE","AVALANCHE",.) %>%
  gsub("COASTALSTORM","COASTAL STORM",.) %>%
  gsub("CHIL$","CHILL",.) %>%
  gsub("DEVEL","DEVIL",.) %>%
  gsub("DUSTSTORM","DUST STORM",.) %>%
  gsub("EROSIN","EROSION",.) %>%
  gsub("(FLOODING|FLOOODING|FLDG|FLD|(FLOODIN$))","FLOOD",.) %>%
  gsub("HAIL STORM","HAILSTORM",.) %>%
  gsub("HVY","HEAVY",.) %>%
  gsub("(LIGHTING|LIGNTNING|LIGHTNING  WAUSEON)","LIGHTNING",.) %>%
  gsub("(MICOBURST|MIRCOBURST)","MICROBURST",.) %>%
  gsub("MUD SLIDE","MUDSLIDE",.) %>%
  gsub("(PRECIPATATION|(PRECIP$))","PRECIPITATION",.) %>%
  gsub("SML","SMALL",.) %>%
  gsub("STRM","STREAM",.) %>%
  gsub("(STROM|TSORM)","STORM",.) %>%
  gsub("(THUNDERTROM|THUNDESTORM|THUNERSTORM|TUNDERSTORM|THUNDERTORM|THUNDEERSTORM|THUNDERESTORM|THUDERSTORM)","THUNDERSTORM",.) %>%
  gsub("(TORNADOES|TORNDAO)","TORNADO",.) %>%
  gsub("WAYTER","WATER",.) %>%
  gsub("WATER SPOUT","WATERSPOUT",.) %>%
  gsub("WETNE","WET",.) %>%
  gsub("(WINDCHILL|WINDCHILL TEMPERATURE)","WIND CHILL",.) %>%
  gsub("(WINTRY|WINTERY)","WINTER",.) %>%
  gsub("(WND|W IND|WINS|WINDS|WINDS G|((WI|WIN)$))","WIND",.) %>%
  gsub("WINDHAIL","WIND/HAIL",.) %>%
  gsub("ACCUMULATED SNOWFALL","SNOW ACCUMULATION",.) %>%
  gsub("ASHFALL","ASH",.) %>%
  gsub("(BEACH|CSTL)","COASTAL",.) %>%
  gsub("COASTA(L|L |L  )FLOO(DING|D)","COASTAL FLOOD",.) %>%
  gsub("DAM FAILURE","DAM BREAK",.) %>%
  gsub("DOWNBURST WIND","DOWNBURST",.) %>%
  gsub("(DRYNESS|DRY (PATTERN|WEATHER))","DRY",.) %>%
  gsub("(EXCESSIVE HEAT/DROUGHT|(HEAT( |/| WAVE )DROUGHT))","DROUGHT/EXCESSIVE HEAT",.) %>%
  gsub("EXCESSIVE PRECIPITATION","HEAVY PRECIPITATION",.) %>%
  gsub("EXCESSIVE RAIN","HEAVY RAIN",.) %>%
  gsub("EXCESSIVE SNOW","HEAVY SNOW",.) %>%
  gsub("(EXTENDED|PROLONGED)","PROLONG",.) %>%
  gsub("EXTREME WIND CHILL/BLOWING SNO","BLOWING SNOW/EXTREME WIND CHILL",.) %>%
  gsub("FLASH FLOOD LANDSLIDE","FLASH FLOOD/LANDSLIDE",.) %>%
  gsub("FLOOD(/| & )HEAVY RAIN","HEAVY RAIN/FLOOD",.) %>%
  gsub("^FUNNEL(|S)$","FUNNEL CLOUD",.) %>%
  gsub("GLAZE ICE","GLAZE",.) %>%
  gsub("HEAVY RAIN EFFECT","HEAVY RAIN",.) %>%
  gsub("HEAVY RAIN/LIGHTNING","LIGHTNING/HEAVY RAIN",.) %>%
  gsub("HEAVY SWELL","HEAVY SURF",.) %>%
  gsub("HEAVY SNOW ANDBLOWING SNOW","HEAVY SNOW/BLOWING SNOW",.) %>%
  gsub("HEAVY SNOW/BLIZZARD","BLIZZARD/HEAVY SNOW",.) %>%
  gsub("HIGH WIND(/|/ )BLIZZARD","BLIZZARD/HIGH WIND",.) %>%
  gsub("HIGH WIND DUST STORM","DUST STORM/HIGH WIND",.) %>%
  gsub("HIGH WIND HEAVY RAIN","HIGH WIND/HEAVY RAIN",.) %>%
  gsub("HIGH WIND(/| AND )HEAVY SNOW","HEAVY SNOW/HIGH WIND",.) %>%
  gsub("^HYPOTHERMIA$","HYPOTHERMIA/EXPOSURE",.) %>%
  gsub("ICE(/| AND )SNOW","SNOW/ICE",.) %>%
  gsub("ICE JAM FLOOD \\(MINOR","ICE JAM FLOOD",.) %>%
  gsub("ICE( ON | )ROAD","ICY ROAD",.) %>%
  gsub("ICE STORM AND SNOW","SNOW/ICE STORM",.) %>%
  gsub("LAKE EFFECT","LAKE-EFFECT",.) %>%
  gsub("LAKE FLOOD","LAKESHORE FLOOD",.) %>%
  gsub("LANDSLUMP","LANDSLIDE",.) %>%
  gsub("((LATE(-| )SEASON)|UNUSUALLY LATE)","LATE",.) %>%
  gsub("(LARGE|ROTATING) WALL CLOUD","WALL CLOUD",.) %>%
  gsub("LIGHTNING( | AND )THUNDERSTORM WIND","THUNDERSTORM WIND/LIGHTNING",.) %>%
  gsub("MICROBURST WIND","MICROBURST",.) %>%
  gsub("MISHAP","ACCIDENT",.) %>%
  gsub("NON-","NON ",.) %>%
  gsub("RAINFALL","RAIN",.) %>%
  gsub("RAIN \\(HEAVY\\)","HEAVY RAIN",.) %>%
  gsub("RECORD/EXCESSIVE HEAT","RECORD HEAT",.) %>%
  gsub("RECORD HIGH TEMPERATURE","HIGH TEMPERATURE RECORD",.) %>%
  gsub("RECORD TEMPERATURE","TEMPERATURE RECORD",.) %>%
  gsub("RECORD WARM TEMP","RECORD WARM",.) %>%
  gsub("RIP CURRENTS( |/)","RIP CURRENT/",.) %>%
  gsub("SEVERE COLD","EXCESSIVE COLD",.) %>%
  gsub("^SMALL STREAM$","SMALL STREAM FLOOD",.) %>%
  gsub("(SNOW/ BITTER COLD|COLD AND SNOW)","SNOW/COLD",.) %>%
  gsub("SNOWFALL RECORD","RECORD SNOW",.) %>%
  gsub("(SNOWFALL|SNOW SHOWER)","SNOW",.) %>%
  gsub("SNOW( |/)FREEZING RAIN","FREEZING RAIN/SNOW",.) %>%
  gsub("SNOW/HIGH WIND","HIGH WIND/SNOW",.) %>%
  gsub("(SNOW/RAIN/SLEET|SLEET/RAIN/SNOW)","SNOW/SLEET/RAIN",.) %>%
  gsub("SNOW/RAIN","RAIN/SNOW",.) %>%
  gsub("SNOW(-|/|)SQUALL","SNOW SQUALL",.) %>%
  gsub("(SNOW SLEET|SLEET/SNOW)","SNOW/SLEET",.) %>%
  gsub("THUNDERSTORM DAMAGE","THUNDERSTORM WIND DAMAGE",.) %>%
  gsub("THUNDERSTORM HAIL","THUNDERSTORM WIND/HAIL",.) %>%
  gsub("THUNDERSTORM WIND( |/)FUNNEL CLOU","THUNDERSTORM WIND/FUNNEL CLOUD",.) %>%
  gsub("TSTM HEAVY RAIN","THUNDERSTORM WIND/HEAVY RAIN",.) %>%
  gsub("(UNUSUAL|ABNORMAL|ABNORMALLY) ","UNUSUALLY ",.) %>%
  gsub("UNSEASONA(BLE|L)","UNSEASONABLY",.) %>%
  gsub("URBAN FLOOD LANDSLIDE","LANDSLIDE/URBAN FLOOD",.) %>%
  gsub("VERY DRY","UNUSUALLY DRY",.) %>%
  gsub("VERY WARM","UNSEASONABLY WARM",.) %>%
  gsub("WARMTH","WARM",.) %>%
  gsub("WIND CHILL/HIGH WIND","HIGH WIND/WIND CHILL",.) %>%
  gsub("WIND GUST","GUSTY WIND",.) %>%
  gsub("^WIND/HAIL$","HAIL/WIND",.) %>%
  gsub("(WINTER MIX|WINTER WEATHER( |/)MIX)","WINTER WEATHER",.) %>%
  gsub("WINTER STORM HIGH WIND","WINTER STORM/HIGH WIND",.) %>%
  gsub(" ADVISOR(Y|IE)","",.) %>%
  gsub("(^(BITTER|COLD) WIND CHILL.*)|(^COLD/WIND$)","COLD/WIND CHILL",.) %>%
  gsub("BLIZZARD (SUMMARY|WEATHER)","BLIZZARD",.) %>%
  gsub("BLOWING SNOW(- EXTREME WIND CHI| & EXTREME WIND CH)","BLOWING SNOW/EXTREME WIND CHILL",.) %>%
  gsub("(BRUSH|WILD/FOREST|WILD|FOREST|GRASS) FIRE","WILDFIRE",.) %>%
  gsub("(COASTAL EROSION/COASTAL FLOOD|EROSION/COASTAL FLOOD)","COASTAL FLOOD/EROSION",.) %>%
  gsub("COLD (TEMPERATURE|WAVE|WEATHER)","COLD",.) %>%
  gsub(" (CONDITION|SPELL)","",.) %>%
  gsub("(DRY HOT WEATHER|WARM DRY)","HOT/DRY",.) %>%
  gsub("DRY MICROBURST.+[0-9]+.*","DRY MICROBURST",.) %>%
  gsub("^(FLASH FLOOD/FLOOD|FLOOD FLOOD/FLASH|FLOOD/FLASH/FLOOD|FLOOD/FLASHFLOOD|FLASH FLOOD/ FLOOD)$","FLOOD/FLASH FLOOD",.) %>%
  gsub("^(FLOOD( |/)FLASH)$","FLASH FLOOD",.) %>%
  gsub("(FREEZING RAIN SLEET AND|(SLEET(/| & )FREEZING RAIN))","FREEZING RAIN/SLEET",.) %>%
  gsub("HAIL.+[0-9]+.*","HAIL",.) %>%
  gsub("(HEAT( WAVE|BURST)|HOT (PATTERN|WEATHER)|(^HOT$)|WARM WEATHER)","HEAT",.) %>%
  gsub("HIGH WIND.+[0-9]+.*","HIGH WIND",.) %>%
  gsub("HIGH (SEA|SWELL| SWELL|TIDE|WAVE)","HIGH SURF",.) %>%
  gsub("(HURRICANE( EDOUARD| EMILY| ERIN| FELIX| GORDON| OPAL|/TYPHOON)|TYPHOON)","HURRICANE",.) %>%
  gsub("PATCHY ","",.) %>%
  gsub("(ROGUE WAVE|ROUGH SEA|ROUGH SURF|HEAVY SEA)","HEAVY SURF",.) %>%
  gsub("(TSTMW|TSTM WIND.+[0-9]+.*|THUNDERSTORM.+[0-9]+.*|THUNDERSTORMWIND|(^THUNDERSTORMW$)|THUNDERSTORMW 50|THUNDERSTORM WIND      LE CEN)","THUNDERSTORM WIND",.) %>%
  gsub("(THUNDERSTORM(S|W)|TSTM)","THUNDERSTORM",.) %>%
  gsub("(THUNDERSTORM|THUNDERSTORMS|TSTM)$","THUNDERSTORM WIND",.) %>%
  gsub("TORNADO F[0-9]","TORNADO",.) %>%
  gsub("(TORNADO/WATERSPOUT|(WATERSPOUT(-| )TORNADO))","WATERSPOUT/TORNADO",.) %>%
  gsub("TROPICAL STORM .+","TROPICAL STORM",.) %>%
  gsub("^((URBAN( |/| AND )SMALL)|URBAN/SMALL FLOOD|(URBAN(/| AND )SMALL STREAM)|URBAN/STREET FLOOD|URBAN SMALL STREAM FLOOD|(SMALL STREAM( |/| AND )URBAN FLOOD)|SMALL STREAM AND)$","URBAN/SMALL STREAM FLOOD",.) %>%
  gsub("((^SUMMARY.+)|(.+YEAR$)|(.*MONTH.*))","-",.) %>%
  gsub("  +"," ",.) %>%
  gsub("( TO|S+| AND)$","",.) %>%
  gsub("(S/|/ +|\\\\| (AND|&|-) )","/",.) %>%
  gsub("FLASH FLOOD/THUNDERSTORM WIND","THUNDERSTORM WIND/FLASH FLOOD",.) %>%
  gsub("THUNDERSTORM WIND ","THUNDERSTORM WIND/",.)

3-2. Across the United States, which types of events (as indicated in the EVTYPE) are most harmful with respect to population health?

Group the data by month and event types.

data_month_event <- group_by(data,month,event_type)

Calculate fatalities by event type by month, then order event types in each month by fatalities in descending manner. Top 5 event types of each month are annotated in detail, other event types are annotated as “OTHERS”. Arrange all top events in desired order for better coloring in “Results” section.

Fatality_by_eventtype_month <- as.data.frame(summarize(data_month_event,sum(FATALITIES)))
Fatality_by_eventtype_month <- Fatality_by_eventtype_month[order(Fatality_by_eventtype_month$month,-Fatality_by_eventtype_month[,3]),]
colnames(Fatality_by_eventtype_month)[3] <- "Fatality"
row.names(Fatality_by_eventtype_month) <- NULL
for (i in 1:12){
  Fatality_by_eventtype_month[Fatality_by_eventtype_month$month==sprintf("%02d",i),][-(1:5),]$event_type <- "OTHERS"
}
Fatality_by_eventtype_month$event_type <- factor(Fatality_by_eventtype_month$event_type,levels=c("AVALANCHE","WINTER STORM","EXTREME COLD","COLD/WIND CHILL","EXTREME COLD/WIND CHILL","HEAT","EXCESSIVE HEAT","WILDFIRE","LIGHTNING","THUNDERSTORM WIND","FLOOD","FLASH FLOOD","RIP CURRENT","HURRICANE","TORNADO","HIGH WIND","OTHERS"))

Calculate injuries by event type by month, then order event types in each month by injuries in descending manner. Top 5 event types of each month are annotated in detail, other event types are annotated as “OTHERS”. Arrange all top events in desired order for better coloring in “Results” section.

Injury_by_eventtype_month <- as.data.frame(summarize(data_month_event,sum(INJURIES)))
Injury_by_eventtype_month <- Injury_by_eventtype_month[order(Injury_by_eventtype_month$month,-Injury_by_eventtype_month[,3]),]
colnames(Injury_by_eventtype_month)[3] <- "Injury"
row.names(Injury_by_eventtype_month) <- NULL
for (i in 1:12){
  Injury_by_eventtype_month[Injury_by_eventtype_month$month==sprintf("%02d",i),][-(1:5),]$event_type <- "OTHERS"
}
Injury_by_eventtype_month$event_type <- factor(Injury_by_eventtype_month$event_type,levels=c("HEAVY SNOW","BLIZZARD","WINTER STORM","ICE STORM","WINTER WEATHER","GLAZE","HEAT","EXCESSIVE HEAT","WILDFIRE","LIGHTNING","THUNDERSTORM WIND","TROPICAL STORM","FLOOD","FLASH FLOOD","HURRICANE","TORNADO","HIGH WIND","FOG","HAIL","OTHERS"))

3-3. Across the United States, which types of events have the greatest economic consequences?

Calculate economic damage values by event type by month, then order event types in each month by economic damage values in descending manner. Top 5 event types of each month are annotated in detail, other event types are annotated as “OTHERS”. Arrange all top events in desired order for better coloring in “Results” section.

EconomicDamage_by_eventtype_month <- as.data.frame(summarize(data_month_event,sum(economic_damage)))
EconomicDamage_by_eventtype_month <- EconomicDamage_by_eventtype_month[order(EconomicDamage_by_eventtype_month$month,-EconomicDamage_by_eventtype_month[,3]),]
colnames(EconomicDamage_by_eventtype_month)[3] <- "Economic_Damage"
row.names(EconomicDamage_by_eventtype_month) <- NULL
for (i in 1:12){
  EconomicDamage_by_eventtype_month[EconomicDamage_by_eventtype_month$month==sprintf("%02d",i),][-(1:5),]$event_type <- "OTHERS"
}
EconomicDamage_by_eventtype_month$event_type <- factor(EconomicDamage_by_eventtype_month$event_type,levels=c("FROST/FREEZE","WINTER STORM","ICE STORM","WILDFIRE","DROUGHT","THUNDERSTORM WIND","TROPICAL STORM","FLOOD","FLASH FLOOD","RIVER FLOOD","STORM SURGE","STORM SURGE/TIDE","HURRICANE","TORNADO","TORNADO, THUNDERSTORM WIND, HAIL","HIGH WIND","HAIL","OTHERS"))

4. Results

4-1. Across the United States, which types of events (as indicated in the EVTYPE) are most harmful with respect to population health?

Plot 1 for fatalities by month. Top 5 event types of each month are annotated in detail.
Top event types show differential distribution pattern across the year, e.g., tornado-caused fatalities are enriched in March to June, heat and lightning-caused fatalities are enriched in July to August, winter/cold-related fatalities are enriched in January. But generally tornado causes the most fatalities.

ggplot(Fatality_by_eventtype_month,aes(x=month,y=Fatality,fill=event_type))+
  geom_bar(stat="identity",position="stack")+
  theme_bw()+
  ggtitle("Plot 1. Fatalities by different event types in each month")+
  xlab("Month")+
  ylab("Fatality numbers")+
  scale_fill_manual(values=c("AVALANCHE"="lightblue","WINTER STORM"="dodgerblue","EXTREME COLD"="steelblue","COLD/WIND CHILL"="cornflowerblue","EXTREME COLD/WIND CHILL"="royalblue","HEAT"="orange","EXCESSIVE HEAT"="darkorange2","WILDFIRE"="red2","LIGHTNING"="yellow","THUNDERSTORM WIND"="gold","FLOOD"="green3","FLASH FLOOD"="palegreen","RIP CURRENT"="olivedrab","HURRICANE"="cyan","TORNADO"="plum3","HIGH WIND"="pink","OTHERS"="gray"),name="Event Types")

Plot 2 for injuries by month. Top 5 event types of each month are annotated in detail.
Top event types show differential distribution pattern across the year, e.g., tornado-caused injuries are enriched in March to June though present in all months, heat and lightning-caused injuries are enriched in July to August, flood-caused injuries are enriched in October. But generally tornado causes the most injuries.

ggplot(Injury_by_eventtype_month,aes(x=month,y=Injury,fill=event_type))+
  geom_bar(stat="identity",position="stack")+
  theme_bw()+
  ggtitle("Plot 2. Injuries by different event types in each month")+
  xlab("Month")+
  ylab("Injury numbers")+
  scale_fill_manual(values=c("HEAVY SNOW"="lightblue","BLIZZARD"="skyblue","WINTER STORM"="dodgerblue","ICE STORM"="cornflowerblue","WINTER WEATHER"="steelblue","GLAZE"="mediumblue","HEAT"="orange","EXCESSIVE HEAT"="darkorange2","WILDFIRE"="red2","LIGHTNING"="yellow","THUNDERSTORM WIND"="gold","TROPICAL STORM"="goldenrod","FLOOD"="green3","FLASH FLOOD"="palegreen","HURRICANE"="cyan","TORNADO"="plum3","HIGH WIND"="pink","FOG"="wheat","HAIL"="beige","OTHERS"="gray"),name="Event Types")

4-2. Across the United States, which types of events have the greatest economic consequences?

Plot 3 for economic damages by month. Top 5 event types of each month are annotated in detail.
Top event types show differential distribution pattern across the year, e.g., flood and storm surge-caused economic damages are enriched in January and August, hurricane-caused economic damages are enriched in August to October, tornado-caused economic damages are enriched in March to June. But generally flood causes the greatest economic consequences.

ggplot(EconomicDamage_by_eventtype_month,aes(x=month,y=Economic_Damage,fill=event_type))+
  geom_bar(stat="identity",position="stack")+
  theme_bw()+
  ggtitle("Plot 3. Economic damage by different event types in each month")+
  xlab("Month")+
  ylab("Economic damage values")+
  scale_fill_manual(values=c("FROST/FREEZE"="lightblue","WINTER STORM"="dodgerblue","ICE STORM"="cornflowerblue","WILDFIRE"="red2","DROUGHT"="orange","THUNDERSTORM WIND"="gold","TROPICAL STORM"="goldenrod","FLOOD"="green3","FLASH FLOOD"="palegreen","RIVER FLOOD"="yellowgreen","STORM SURGE"="seagreen","STORM SURGE/TIDE"="darkgreen","HURRICANE"="cyan","TORNADO"="plum3","TORNADO, THUNDERSTORM WIND, HAIL"="violet","HIGH WIND"="pink","HAIL"="beige","OTHERS"="gray"),name="Event Types")