Life and Property/Infrastructure Loss Values in the U.S. NOAA Storm DataBase
iconodo
23 de abril de 2017
“every step deeper into the analysis takes a soul away… Teilhard de Chardin”
SYNOPSIS. (1)The U.S. National Oceanic and Atmospheric Administration’s (NOAA) storm database holds events from 1950 to November 2011. (2)Respecting personal damages, this paper considers the loss of life the utmost consequence of any event in the database related to health condition. (3))Old age and early childhood, not by themselves “health conditions”, are determinant in relation to hot weather events as the second main cause of fatalities. Tornadoes, whose fatalities migth allegedly be excluded from a “health condition” direct relationship due to their striking nature, are the first cause of deaths in the database. (5)In the other hand, infrastructure and property damages/losses provoked by direct weather events have been confronted to their pure economic impact. (6)The analysis seems to show an increase in data recording from the 1990s on, or an increase in events related with heat, as the Fig.3 changes in the bottom line from tornados and surges to heat and hurricanes. (7) In general, despite some peaks, there seems to be an increase of heat and hurricane and flood events and a decrease of deaths in the main deadly events towards the last dates. (8)The system configuration for this April/2017 paper was: Ubuntu 16.04 LTS, Rstudio Version 1.0.136 – © 2009-2016 RStudio, Inc. under R version 3.3.3 (2017-03-06) – “Another Canoe”.
1. DATA PROCESSING
- Preprocessing.0 Download of the compressed data file
fileUrl <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
download.file(fileUrl, destfile="./repdata_data_StormData.csv.bz2")- Preprocessing.1 Loading under dataframe name
"stormdat"the downloaded decompressed data file “rep_data_StormData.csv” into R’s workspace.
stormdat <- read.csv("repdata_data_StormData.csv.bz2")- Preprocessing.2 Load of libraries used in the processing…
library(dplyr)
library(lubridate)
library(ggplot2)- Preprocessing.3 A preliminar scoping of the structure showing the integrity of the downloaded data base
str(stormdat)## 'data.frame': 902297 obs. of 37 variables:
## $ STATE__ : num 1 1 1 1 1 1 1 1 1 1 ...
## $ BGN_DATE : Factor w/ 16335 levels "10/10/1954 0:00:00",..: 6523 6523 4213 11116 1426 1426 1462 2873 3980 3980 ...
## $ BGN_TIME : Factor w/ 3608 levels "000","0000","00:00:00 AM",..: 212 257 2645 1563 2524 3126 122 1563 3126 3126 ...
## $ TIME_ZONE : Factor w/ 22 levels "ADT","AKS","AST",..: 7 7 7 7 7 7 7 7 7 7 ...
## $ COUNTY : num 97 3 57 89 43 77 9 123 125 57 ...
## $ COUNTYNAME: Factor w/ 29601 levels "","5NM E OF MACKINAC BRIDGE TO PRESQUE ISLE LT MI",..: 13513 1873 4598 10592 4372 10094 1973 23873 24418 4598 ...
## $ STATE : Factor w/ 72 levels "AK","AL","AM",..: 2 2 2 2 2 2 2 2 2 2 ...
## $ EVTYPE : Factor w/ 985 levels "?","ABNORMALLY DRY",..: 830 830 830 830 830 830 830 830 830 830 ...
## $ BGN_RANGE : num 0 0 0 0 0 0 0 0 0 0 ...
## $ BGN_AZI : Factor w/ 35 levels "","E","Eas","EE",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ BGN_LOCATI: Factor w/ 54429 levels "","?","(01R)AFB GNRY RNG AL",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ END_DATE : Factor w/ 6663 levels "","10/10/1993 0:00:00",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ END_TIME : Factor w/ 3647 levels "","?","0000",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ COUNTY_END: num 0 0 0 0 0 0 0 0 0 0 ...
## $ COUNTYENDN: logi NA NA NA NA NA NA ...
## $ END_RANGE : num 0 0 0 0 0 0 0 0 0 0 ...
## $ END_AZI : Factor w/ 24 levels "","E","ENE","ESE",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ END_LOCATI: Factor w/ 34506 levels "","(0E4)PAYSON ARPT",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ LENGTH : num 14 2 0.1 0 0 1.5 1.5 0 3.3 2.3 ...
## $ WIDTH : num 100 150 123 100 150 177 33 33 100 100 ...
## $ F : int 3 2 2 2 2 2 2 1 3 3 ...
## $ MAG : num 0 0 0 0 0 0 0 0 0 0 ...
## $ FATALITIES: num 0 0 0 0 0 0 0 0 1 0 ...
## $ INJURIES : num 15 0 2 2 2 6 1 0 14 0 ...
## $ PROPDMG : num 25 2.5 25 2.5 2.5 2.5 2.5 2.5 25 25 ...
## $ PROPDMGEXP: Factor w/ 19 levels "","-","?","+",..: 17 17 17 17 17 17 17 17 17 17 ...
## $ CROPDMG : num 0 0 0 0 0 0 0 0 0 0 ...
## $ CROPDMGEXP: Factor w/ 9 levels "","?","0","2",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ WFO : Factor w/ 542 levels "","2","43","9V9",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ STATEOFFIC: Factor w/ 250 levels "","ALABAMA, Central",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ ZONENAMES : Factor w/ 25112 levels ""," "| __truncated__,..: 1 1 1 1 1 1 1 1 1 1 ...
## $ LATITUDE : num 3040 3042 3340 3458 3412 ...
## $ LONGITUDE : num 8812 8755 8742 8626 8642 ...
## $ LATITUDE_E: num 3051 0 0 0 0 ...
## $ LONGITUDE_: num 8806 0 0 0 0 ...
## $ REMARKS : Factor w/ 436781 levels ""," "," "," ",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ REFNUM : num 1 2 3 4 5 6 7 8 9 10 ...- Preprocessing.4 Rnaming of columns (EVTYPE, STATE, BGN_DATE, END_DATE and FATALITIES) to be used in the personal and wealth issues analysis; and CROPDMG (damages in crops), PROPDMG (damages in properties), CROPDMGEXP (indicator/exponent for billions, millions, thounsands, etc. in crops), PROPDMGEXP (same indicator in properties), to be used in the properties/infrastructure/crops damages analysis.
stormdat <- rename(stormdat, event = EVTYPE, state = STATE,
bgn_date = BGN_DATE, end_date = END_DATE,
cdmg = CROPDMG, pdmg = PROPDMG,
cropdmgexp = CROPDMGEXP, propdmgexp = PROPDMGEXP,
fatalities = FATALITIES)1.1 Personal Issues processing
- 1.1.1 A subsetting of the original dataset (stormdat) into databasees
stormdat.fandstormdat.f.descwith columnsevent, state, bgn_date, end_dateand criterium columnfatalities > 0filtered in descending order…
stormdat.f <- filter(stormdat, fatalities > 0)
stormdat.f.desc <- select(stormdat.f, event, state, bgn_date,
end_date, fatalities) %>%
arrange(desc(fatalities))… and a look into first and last rows of resulting stormdat.f.desc dataframe…
head(stormdat.f.desc)## event state bgn_date end_date fatalities
## 1 HEAT IL 7/12/1995 0:00:00 7/16/1995 0:00:00 583
## 2 TORNADO MO 5/22/2011 0:00:00 5/22/2011 0:00:00 158
## 3 TORNADO MI 6/8/1953 0:00:00 116
## 4 TORNADO TX 5/11/1953 0:00:00 114
## 5 EXCESSIVE HEAT IL 7/28/1999 0:00:00 7/31/1999 0:00:00 99
## 6 TORNADO MA 6/9/1953 0:00:00 90tail(stormdat.f.desc) ## event state bgn_date end_date fatalities
## 6969 HIGH WIND CA 11/30/2011 0:00:00 11/30/2011 0:00:00 1
## 6970 RIP CURRENT FL 11/9/2011 0:00:00 11/9/2011 0:00:00 1
## 6971 COASTAL FLOOD AK 11/8/2011 0:00:00 11/10/2011 0:00:00 1
## 6972 STRONG WIND FL 11/28/2011 0:00:00 11/28/2011 0:00:00 1
## 6973 HIGH WIND WA 11/22/2011 0:00:00 11/22/2011 0:00:00 1
## 6974 AVALANCHE UT 11/13/2011 0:00:00 11/13/2011 0:00:00 1… put us on the track of the types of events (as indicated in the event variable) that are most harmful with respect to population health across the United States; narrowing the scoping to 6974 observations (as shown in the last row of the above section).
- 1.1.2 Grouping and Summarizing the filtered database, results in Dplyr’s tibble object
summarizedby.deaths…
groupedby.evt <- group_by(stormdat.f.desc, event)
summarizedby.deaths <- summarize(groupedby.evt, fatalities = sum(fatalities))
summarizedby.deaths## # A tibble: 168 × 2
## event fatalities
## <fctr> <dbl>
## 1 AVALANCE 1
## 2 AVALANCHE 224
## 3 BLACK ICE 1
## 4 BLIZZARD 101
## 5 blowing snow 1
## 6 BLOWING SNOW 1
## 7 COASTAL FLOOD 3
## 8 Coastal Flooding 2
## 9 COASTAL FLOODING 1
## 10 COASTALSTORM 1
## # ... with 158 more rows- 1.1.3 …that once taken back into data.frame formatted oject
by.deathsand poured in descending order in the setby.deaths.desc…
by.deaths <- as.data.frame(summarizedby.deaths)
by.deaths.desc <- arrange(by.deaths, desc(fatalities))… can be observed in its first and last six rows…
head(by.deaths.desc)## event fatalities
## 1 TORNADO 5633
## 2 EXCESSIVE HEAT 1903
## 3 FLASH FLOOD 978
## 4 HEAT 937
## 5 LIGHTNING 816
## 6 TSTM WIND 504tail(by.deaths.desc)## event fatalities
## 163 URBAN AND SMALL STREAM FLOODIN 1
## 164 Whirlwind 1
## 165 WINDS 1
## 166 WIND STORM 1
## 167 WINTER STORM HIGH WINDS 1
## 168 WINTRY MIX 1… narrowing the observations to 168. Entering command summarizedby.deaths %>% tbl_df %>% print(n=168) you can see this tibble object in its entirety and adding respectively by hand the fatalities in rows containing the words “TORNADO” and “HEAT” in the event column, should account to fatalities (due to tornados = 5661, due to heat events = 3138). Notes: 1) Read ahead in the RESULTS section for an automatized procedure for this totals. 2) We maintained the different descriptions for the same type of event (e.g HEAT VS. EXTREME HEAT; TORNADO VS WATERSPROUT/TORNADO ). We found those diversities useful to illustrate the ahead plotS captioned as Fig. 1 and Fig. 2.
- 1.1.4 Preprocessing for total fatalities by Heat Events and Tornados; and respective plots Fig. 1 and Fig. 2.
Objects fatsby.heat and fatsby.tndo filter database stormdat.f.dec, respectively by heat events and by tornado events. Objects fatsby.heat.arr and fatsby.tndo.arr first mutate their initdate column from factor attribute to character attribute, then transform their column initdate into Lubridate’s date column initdate2. Objects fastby.heat2 and fastby.tndo2 trim undesired columns. Objects toplot.heat and toplot.tndo (to be used in the ahead plots) group the data by date and type of event.
fatsby.heat <- stormdat.f.desc[grepl("HEAT|heat",stormdat.f.desc$event),]
fatsby.tndo <- stormdat.f.desc[grepl("TORNADO|tornado",stormdat.f.desc$event),]
fatsby.heat.arr <- mutate(fatsby.heat, initdate = as.character.Date(bgn_date))
fatsby.heat.arr <- mutate(fatsby.heat.arr, initdate2 = mdy_hms(initdate))
fatsby.heat2 <- select(fatsby.heat.arr, event, initdate2, fatalities)
toplot.heat <- group_by(fatsby.heat2, initdate2, event)
fatsby.tndo.arr <- mutate(fatsby.tndo, initdate = as.character.Date(bgn_date))
fatsby.tndo.arr <- mutate(fatsby.tndo.arr, initdate2 = mdy_hms(initdate))
fatsby.tndo2 <- select(fatsby.tndo.arr, event, initdate2, fatalities)
toplot.tndo <- group_by(fatsby.tndo2, initdate2, event)1.2 Property/Infrastructure Issues processing
- 1.2.1 A first subsettings of the original dataset (under
stormdatname) into dmages in properties and crops databasestormdat.dinPCin order to filter indicators for Billions, Millions and Thousands(“Kilos”)…
stormdat.dinPC <- filter(stormdat, pdmg > 0 & pdmg < 10000 |
cdmg > 0 & cdmg < 10000)
stormdat.dinPC <- filter(stormdat.dinPC, propdmgexp == "B" |
propdmgexp == "b" | propdmgexp == "M" |
propdmgexp == "m" | propdmgexp == "K" |
propdmgexp == "k" |
cropdmgexp == "B" |
cropdmgexp == "b" | cropdmgexp == "M" |
cropdmgexp == "m" | cropdmgexp == "K" |
cropdmgexp == "k") … and formatting of beginning dates of events. As well and renaming the damages and indicators (pplyr for properties indicator, cplyr for crops indicator). Also, columns dlspr (dollars in property damagges), dlscr (dollars in crops damages), and dlstot for summarizing both, were created and initialized with value = 0. Finally, the columns were arranged in a more readable fashion.
stormdat.dinPC <- mutate(stormdat.dinPC, initdate = as.character.Date(bgn_date))
stormdat.dinPC <- mutate(stormdat.dinPC, bgndate = mdy_hms(initdate))
stormdat.dinPC <- rename(stormdat.dinPC, pplyr = propdmgexp, cplyr = cropdmgexp )
stormdat.dinPC <- mutate(stormdat.dinPC, dlspr = 0, dlscr = 0, dlstot = 0 )
stormdat.dinPC <- select(stormdat.dinPC, event, state, bgndate,
pdmg, pplyr, dlspr, cdmg, cplyr, dlscr, dlstot)- 1.2.2 Further processing, to separate in BILLIONS, was made as follows. A first filtering of inidcators “B” or “b” followed by the conversion of column
pdmgtimes 1 billion was poured into respectivedlspranddlscrcolumns of databasesstormdat.dinPC.Bpropandstormdat.dinPC.BcropAfter that, in dataframeBbindthose two sets were binded and put in descending oder. Finally, first and last three rows are displayed…
stormdat.dinPC.Bprop <- filter(stormdat.dinPC, (pplyr == "B" | pplyr == "b" ))
stormdat.dinPC.Bprop <- mutate(stormdat.dinPC.Bprop, dlspr = 1000000000 * pdmg)
stormdat.dinPC.Bcrop <- filter(stormdat.dinPC, (cplyr == "B" | cplyr == "b" ))
stormdat.dinPC.Bcrop <- mutate(stormdat.dinPC.Bcrop, dlscr = 1000000000 * cdmg)
Bbind <- rbind(stormdat.dinPC.Bprop, stormdat.dinPC.Bcrop)
Bbind <- mutate(Bbind, dlstot = dlspr + dlscr)
Bbind <- arrange(Bbind, desc(dlstot))
head(Bbind, 3)## event state bgndate pdmg pplyr dlspr cdmg cplyr
## 1 FLOOD CA 2006-01-01 115.00 B 1.150e+11 32.5 M
## 2 STORM SURGE LA 2005-08-29 31.30 B 3.130e+10 0.0
## 3 HURRICANE/TYPHOON LA 2005-08-28 16.93 B 1.693e+10 0.0
## dlscr dlstot
## 1 0 1.150e+11
## 2 0 3.130e+10
## 3 0 1.693e+10tail(Bbind, 3)## event state bgndate pdmg pplyr dlspr cdmg cplyr
## 45 HEAT AL 1995-08-20 0.0 0e+00 0.4 B
## 46 FREEZE IA 1995-09-21 0.0 0e+00 0.2 B
## 47 HURRICANE OPAL/HIGH WINDS AL 1995-10-04 0.1 B 1e+08 10.0 M
## dlscr dlstot
## 45 4e+08 4e+08
## 46 2e+08 2e+08
## 47 0e+00 1e+08… and a similar procedure was made for MILLIONS, changing accordingly the names of the dataframes, indicators and the number 1,000,000 to be multiplied. The same with the displayin of the first and last three rows…
stormdat.dinPC.Mprop <- filter(stormdat.dinPC, (pplyr == "M" | pplyr == "m" ))
stormdat.dinPC.Mprop <- mutate(stormdat.dinPC.Mprop, dlspr = 1000000 * pdmg)
stormdat.dinPC.Mcrop <- filter(stormdat.dinPC, (cplyr == "M" | cplyr == "m" ))
stormdat.dinPC.Mcrop <- mutate(stormdat.dinPC.Mcrop, dlscr = 1000000 * cdmg)
Mbind <- rbind(stormdat.dinPC.Mprop, stormdat.dinPC.Mcrop)
Mbind <- mutate(Mbind, dlstot = dlspr + dlscr)
Mbind <- filter(Mbind, dlstot > 0)
Mbind <- arrange(Mbind, desc(dlstot))
head(Mbind, 3)## event state bgndate pdmg pplyr dlspr cdmg cplyr dlscr
## 1 HIGH WIND FL 2004-08-13 929.00 M 929000000 175 M 0
## 2 HAIL AZ 2010-10-05 900.00 M 900000000 0 K 0
## 3 HURRICANE NC 1996-09-04 792.15 M 792150000 0 0
## dlstot
## 1 929000000
## 2 900000000
## 3 792150000tail(Mbind, 3)## event state bgndate pdmg pplyr dlspr cdmg cplyr dlscr
## 13243 WILD/FOREST FIRE SD 1994-08-09 3.00 K 0 0.01 M 10000
## 13244 HEAVY SNOW SD 1995-03-26 0.01 M 0 0.01 M 10000
## 13245 HEAVY SNOW SD 1995-04-26 0.01 M 0 0.01 M 10000
## dlstot
## 13243 10000
## 13244 10000
## 13245 10000… finally, same procedure was made for the thounsands (“kilos”) resulting amounts of damages, up to the first and last rows displaying.
stormdat.dinPC.Kprop <- filter(stormdat.dinPC, (pplyr == "K" | pplyr == "k" ))
stormdat.dinPC.Kprop <- mutate(stormdat.dinPC.Kprop, dlspr = 1000 * pdmg)
stormdat.dinPC.Kcrop <- filter(stormdat.dinPC, (cplyr == "K" | cplyr == "k" ))
stormdat.dinPC.Kcrop <- mutate(stormdat.dinPC.Kcrop, dlscr = 1000 * cdmg)
Kbind <- rbind(stormdat.dinPC.Kprop, stormdat.dinPC.Kcrop)
Kbind <- mutate(Kbind, dlstot = dlspr + dlscr)
Kbind <- arrange(Kbind, desc(dlstot))
head(Kbind, 3)## event state bgndate pdmg pplyr dlspr cdmg cplyr dlscr
## 1 THUNDERSTORM WIND NC 2009-07-26 5000 K 5e+06 0 K 0
## 2 FLASH FLOOD IL 2010-05-13 5000 K 5e+06 0 K 0
## 3 FLASH FLOOD IL 2010-05-13 5000 K 5e+06 0 K 0
## dlstot
## 1 5e+06
## 2 5e+06
## 3 5e+06 tail(Kbind, 3)## event state bgndate pdmg pplyr dlspr cdmg cplyr dlscr
## 327036 STRONG WIND KY 2011-11-13 1.0 K 0 0 K 0
## 327037 DROUGHT TX 2011-11-01 2.0 K 0 0 K 0
## 327038 HIGH WIND CA 2011-11-30 7.5 K 0 0 K 0
## dlstot
## 327036 0
## 327037 0
## 327038 0- 1.2.3 Binding of the three previous databases into
BMKbinddata set, and further analysis. In order to narrow our results, we filter the binded data frame with values bigger than the average of the median and mean of house prices in the year 1983 ($75,300 and $89,800 respectively), as obtained from the Median and Average Sales Prices of New Homes Sold in United States of the US Government census page.
BMKbind <- rbind(Bbind, Mbind, Kbind)
medianUSAHomePrice1983 <- 75300
meanUSAHomePrice1983 <- 89800
avgPrice <- (medianUSAHomePrice1983 + meanUSAHomePrice1983) / 2
BMKbind <- filter(BMKbind, dlstot > avgPrice + 300000)
head(BMKbind)## event state bgndate pdmg pplyr dlspr cdmg cplyr
## 1 FLOOD CA 2006-01-01 115.00 B 1.150e+11 32.5 M
## 2 STORM SURGE LA 2005-08-29 31.30 B 3.130e+10 0.0
## 3 HURRICANE/TYPHOON LA 2005-08-28 16.93 B 1.693e+10 0.0
## 4 STORM SURGE MS 2005-08-29 11.26 B 1.126e+10 0.0
## 5 HURRICANE/TYPHOON FL 2005-10-24 10.00 B 1.000e+10 0.0
## 6 HURRICANE/TYPHOON MS 2005-08-28 7.35 B 7.350e+09 0.0
## dlscr dlstot
## 1 0 1.150e+11
## 2 0 3.130e+10
## 3 0 1.693e+10
## 4 0 1.126e+10
## 5 0 1.000e+10
## 6 0 7.350e+09tail(BMKbind)## event state bgndate pdmg pplyr dlspr cdmg cplyr dlscr
## 20522 HEAVY RAIN OK 2007-06-01 0 K 0 387 K 387000
## 20523 WINTER STORM VT 1997-03-05 385 K 385000 0 0
## 20524 FLASH FLOOD KY 2001-08-04 385 K 385000 0 K 0
## 20525 HIGH WIND NC 2004-03-07 385 K 385000 0 0
## 20526 FLASH FLOOD PR 2010-05-31 385 K 385000 0 K 0
## 20527 HAIL GA 2011-05-26 385 K 385000 0 K 0
## dlstot
## 20522 387000
## 20523 385000
## 20524 385000
## 20525 385000
## 20526 385000
## 20527 385000In order to group and summarize the filtered set, a table of the events is extracted from the first 1000 rows of the data frame
table(droplevels.factor(head(BMKbind, 1000)$event))##
## BLIZZARD COASTAL FLOOD
## 10 1
## COASTAL FLOODING COLD AND WET CONDITIONS
## 1 1
## Damaging Freeze DAMAGING FREEZE
## 1 1
## DROUGHT Early Frost
## 81 1
## EXCESSIVE HEAT EXCESSIVE WETNESS
## 1 1
## EXTREME COLD FLASH FLOOD
## 7 107
## FLASH FLOOD/FLOOD FLASH FLOODING
## 5 4
## FLOOD FLOOD/FLASH FLOOD
## 152 4
## FLOODING FLOOD/RAIN/WINDS
## 2 2
## FREEZE FROST
## 5 1
## FROST/FREEZE HAIL
## 9 96
## HAILSTORM HEAT
## 1 1
## HEAVY RAIN HEAVY RAINS
## 10 1
## HEAVY RAIN/SEVERE WEATHER HEAVY SNOW
## 1 10
## HIGH WIND HIGH WINDS
## 17 2
## HIGH WINDS/COLD HURRICANE
## 2 52
## HURRICANE EMILY HURRICANE ERIN
## 1 2
## HURRICANE OPAL HURRICANE OPAL/HIGH WINDS
## 3 1
## HURRICANE/TYPHOON ICE STORM
## 36 30
## LANDSLIDE MAJOR FLOOD
## 2 2
## RECORD COLD RIVER FLOOD
## 1 3
## River Flooding SEVERE THUNDERSTORM
## 1 1
## STORM SURGE STORM SURGE/TIDE
## 6 4
## STRONG WIND THUNDERSTORM WIND
## 2 10
## THUNDERSTORM WINDS TORNADO
## 18 193
## TORNADOES, TSTM WIND, HAIL TROPICAL STORM
## 1 25
## TSTM WIND TSUNAMI
## 20 1
## TYPHOON WATERSPOUT/TORNADO
## 2 1
## WILDFIRE WILDFIRES
## 24 2
## WILD FIRES WILD/FOREST FIRE
## 1 9
## WINTER STORM WINTER STORM HIGH WINDS
## 7 1… putting us on the track of the types of events (as indicated by the event variable) that have had the bigger impacts on properties/infrastructure and crops; allowing us to determine the criteria for grouping/summarizing our scoping.
- 1.2.4 After slendering of consolidated
BMKbind, data frame (leaving columns:event,state,bgndateanddlstot)…
BMKbind <- select(BMKbind, event, state, bgndate, dlstot)- 1.2.4 […continued. The below described procedure for events of type “flood”“, is repeated for events of types”hurricane“,”tornado“, etc.] Firstly, the
dmgsby.floodsubset is populated by a “grepl” command that searches the events column for character strings “FLOOD” or “flood” along the column’s width. After obtaining the subset, the event column is filled with the unique string “FLOOD” and the columndyearis created and populated with the event year’s 4 digits from the date columnbgndate. Next step creates the database variabletoplot.floodand populates it with a grouped (by year + name of event, columns) set. Finally, the first ten rows of the data frame set (that later is going to be binded for plot Fig. 3) are displayed; including its dimensions and attributes.
dmgsby.flood <- BMKbind[grepl("FLOOD|flood", BMKbind$event),]
dmgsby.flood <- mutate(dmgsby.flood, event = "FLOOD", dyear = year(bgndate))
toplot.flood <- group_by(dmgsby.flood, dyear, event)
toplot.flood## Source: local data frame [6,154 x 5]
## Groups: dyear, event [19]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 FLOOD CA 2006-01-01 1.15e+11 2006
## 2 FLOOD IL 1993-08-31 5.00e+09 1993
## 3 FLOOD IL 1993-08-31 5.00e+09 1993
## 4 FLOOD ND 1997-04-18 3.00e+09 1997
## 5 FLOOD TN 2011-05-01 2.00e+09 2011
## 6 FLOOD TN 2010-05-01 1.50e+09 2010
## 7 FLOOD AL 2003-05-07 1.00e+09 2003
## 8 FLOOD MS 2011-05-01 1.00e+09 2011
## 9 FLOOD IA 2008-06-01 7.50e+08 2008
## 10 FLOOD NV 1997-01-01 6.40e+08 1997
## # ... with 6,144 more rowsdmgsby.hurricane <- BMKbind[grepl("HURRICANE|TYPHOON", BMKbind$event),]
dmgsby.hurricane <- mutate(dmgsby.hurricane,
event = "HURRICANE|TYPHOON", dyear = year(bgndate))
toplot.hurricane <- group_by(dmgsby.hurricane, dyear, event)
toplot.hurricane## Source: local data frame [249 x 5]
## Groups: dyear, event [16]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 HURRICANE|TYPHOON LA 2005-08-28 1.693e+10 2005
## 2 HURRICANE|TYPHOON FL 2005-10-24 1.000e+10 2005
## 3 HURRICANE|TYPHOON MS 2005-08-28 7.350e+09 2005
## 4 HURRICANE|TYPHOON MS 2005-08-29 5.880e+09 2005
## 5 HURRICANE|TYPHOON FL 2004-08-13 5.420e+09 2004
## 6 HURRICANE|TYPHOON FL 2004-09-04 4.830e+09 2004
## 7 HURRICANE|TYPHOON FL 2004-09-13 4.000e+09 2004
## 8 HURRICANE|TYPHOON LA 2005-09-23 4.000e+09 2005
## 9 HURRICANE|TYPHOON NC 1999-09-15 3.000e+09 1999
## 10 HURRICANE|TYPHOON AL 2004-09-13 2.500e+09 2004
## # ... with 239 more rowsdmgsby.tropical <- BMKbind[grepl("TROPICAL STORM", BMKbind$event),]
dmgsby.tropical <- mutate(dmgsby.tropical,
event = "TROPICAL STORM", dyear = year(bgndate))
toplot.tropical <- group_by(dmgsby.tropical , dyear, event)
toplot.tropical## Source: local data frame [194 x 5]
## Groups: dyear, event [17]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 TROPICAL STORM TX 2001-06-05 5150000000 2001
## 2 TROPICAL STORM MD 2003-09-18 530470000 2003
## 3 TROPICAL STORM TX 1998-09-07 287180000 1998
## 4 TROPICAL STORM AZ 1997-09-25 200000000 1997
## 5 TROPICAL STORM FL 2004-09-05 179400000 2004
## 6 TROPICAL STORM FL 2004-09-25 134800000 2004
## 7 TROPICAL STORM DC 2003-09-18 125000000 2003
## 8 TROPICAL STORM LA 2002-09-25 108630000 2002
## 9 TROPICAL STORM PR 2004-09-14 101500000 2004
## 10 TROPICAL STORM FL 2005-08-27 100000000 2005
## # ... with 184 more rowsdmgsby.hail <- BMKbind[grepl("HAIL", BMKbind$event),]
dmgsby.hail <- mutate(dmgsby.hail,
event = "HAIL", dyear = year(bgndate))
toplot.hail <- group_by(dmgsby.hail , dyear, event)
toplot.hail## Source: local data frame [2,302 x 5]
## Groups: dyear, event [19]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 HAIL AZ 2010-10-05 1.8e+09 2010
## 2 HAIL FL 1993-03-12 1.6e+09 1993
## 3 HAIL AZ 2010-10-05 9.0e+08 2010
## 4 HAIL KY 1998-04-16 5.1e+08 1998
## 5 HAIL WI 2006-06-25 5.0e+08 2006
## 6 HAIL MN 1998-05-15 4.5e+08 1998
## 7 HAIL MO 2001-04-10 4.0e+08 2001
## 8 HAIL CO 2009-07-20 3.5e+08 2009
## 9 HAIL MO 2001-04-10 3.0e+08 2001
## 10 HAIL NE 2001-04-10 3.0e+08 2001
## # ... with 2,292 more rowsdmgsby.winterstorm <- BMKbind[grepl("WINTER STORM", BMKbind$event),]
dmgsby.winterstorm <- mutate(dmgsby.winterstorm,
event = "WINTER STORM", dyear = year(bgndate))
toplot.winterstorm <- group_by(dmgsby.winterstorm , dyear, event)
toplot.winterstorm## Source: local data frame [244 x 5]
## Groups: dyear, event [19]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 WINTER STORM AL 1993-03-12 5.000e+09 1993
## 2 WINTER STORM OH 2008-03-07 7.500e+08 2008
## 3 WINTER STORM CA 1995-12-09 6.000e+07 1995
## 4 WINTER STORM OH 2004-12-22 5.490e+07 2004
## 5 WINTER STORM CA 1993-01-13 5.000e+07 1993
## 6 WINTER STORM NC 1993-03-12 5.000e+07 1993
## 7 WINTER STORM CO 2003-03-17 3.100e+07 2003
## 8 WINTER STORM KS 2005-01-04 3.000e+07 2005
## 9 WINTER STORM NC 1998-02-03 2.218e+07 1998
## 10 WINTER STORM KS 2006-12-29 2.100e+07 2006
## # ... with 234 more rowsdmgsby.tornado <- BMKbind[grepl("TORNADO", BMKbind$event),]
dmgsby.tornado <- mutate(dmgsby.tornado,
event = "TORNADO", dyear = year(bgndate))
toplot.tornado <- group_by(dmgsby.tornado , dyear, event)
toplot.tornado## Source: local data frame [5,857 x 5]
## Groups: dyear, event [62]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 TORNADO MO 2011-05-22 2.8e+09 2011
## 2 TORNADO FL 1993-03-12 1.6e+09 1993
## 3 TORNADO AL 2011-04-27 1.5e+09 2011
## 4 TORNADO AL 2011-04-27 1.0e+09 2011
## 5 TORNADO AL 2011-04-27 7.0e+08 2011
## 6 TORNADO MS 2011-04-04 5.0e+08 2011
## 7 TORNADO OK 1999-05-03 4.5e+08 1999
## 8 TORNADO OK 1999-05-03 4.5e+08 1999
## 9 TORNADO AL 1989-11-15 2.5e+08 1989
## 10 TORNADO AL 1989-11-15 2.5e+08 1989
## # ... with 5,847 more rowsdmgsby.fire <- BMKbind[grepl("FIRE|fire", BMKbind$event),]
dmgsby.fire <- mutate(dmgsby.fire,
event = "WILD FIRE", dyear = year(bgndate))
toplot.fire <- group_by(dmgsby.fire , dyear, event)
toplot.fire## Source: local data frame [372 x 5]
## Groups: dyear, event [19]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 WILD FIRE NM 2000-05-04 1.500e+09 2000
## 2 WILD FIRE CA 2003-10-25 1.040e+09 2003
## 3 WILD FIRE CA 2003-10-25 6.964e+08 2003
## 4 WILD FIRE CA 1993-11-01 6.190e+08 1993
## 5 WILD FIRE CA 2000-09-29 5.470e+08 2000
## 6 WILD FIRE CA 2007-06-24 5.000e+08 2007
## 7 WILD FIRE CA 2003-11-01 2.786e+08 2003
## 8 WILD FIRE TX 2011-09-04 2.500e+08 2011
## 9 WILD FIRE CO 2010-09-06 2.170e+08 2010
## 10 WILD FIRE FL 1998-07-01 2.000e+08 1998
## # ... with 362 more rowsdmgsby.blizzard <- BMKbind[grepl("BLIZZARD|blizzard", BMKbind$event),]
dmgsby.blizzard <- mutate(dmgsby.blizzard,
event = "BLIZZARD", dyear = year(bgndate))
toplot.blizzard <- group_by(dmgsby.blizzard , dyear, event)
toplot.blizzard## Source: local data frame [69 x 5]
## Groups: dyear, event [14]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 BLIZZARD ND 1997-04-05 102000000 1997
## 2 BLIZZARD CO 2003-03-17 62000000 2003
## 3 BLIZZARD ND 1997-01-09 55080000 1997
## 4 BLIZZARD GA 1993-03-13 50000000 1993
## 5 BLIZZARD NY 1993-03-13 50000000 1993
## 6 BLIZZARD SD 1997-04-05 50000000 1997
## 7 BLIZZARD GA 1993-03-13 50000000 1993
## 8 BLIZZARD GA 1993-03-13 50000000 1993
## 9 BLIZZARD ND 1997-04-04 44720000 1997
## 10 BLIZZARD UT 1997-01-11 40000000 1997
## # ... with 59 more rows#
dmgsby.heat <- BMKbind[grepl("HEAT|heat", BMKbind$event),]
dmgsby.heat <- mutate(dmgsby.heat,
event = "HEAT", dyear = year(bgndate))
toplot.heat2 <- group_by(dmgsby.heat , dyear, event)
toplot.heat2## Source: local data frame [16 x 5]
## Groups: dyear, event [6]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 HEAT AL 1995-08-20 400000000 1995
## 2 HEAT CA 2006-07-16 492400000 2006
## 3 HEAT OH 1994-06-13 5000000 1994
## 4 HEAT MD 1995-07-14 4700000 1995
## 5 HEAT IA 1995-07-12 3800000 1995
## 6 HEAT NE 1999-07-19 3300000 1999
## 7 HEAT NE 2005-07-22 3000000 2005
## 8 HEAT IA 1995-07-10 2400000 1995
## 9 HEAT MN 1995-07-10 2000000 1995
## 10 HEAT NE 2009-06-22 1500000 2009
## 11 HEAT IN 1995-07-13 1000000 1995
## 12 HEAT VA 1995-07-14 600000 1995
## 13 HEAT MO 1995-07-17 400000 1995
## 14 HEAT OH 1995-07-12 600000 1995
## 15 HEAT OH 1995-08-08 500000 1995
## 16 HEAT MO 1995-08-01 400000 1995dmgsby.surge <- BMKbind[grepl("SURGE|surge", BMKbind$event),]
dmgsby.surge <- mutate(dmgsby.surge,
event = "SURGE", dyear = year(bgndate))
toplot.surge <- group_by(dmgsby.surge , dyear, event)
toplot.surge## Source: local data frame [75 x 5]
## Groups: dyear, event [13]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 SURGE LA 2005-08-29 3.130e+10 2005
## 2 SURGE MS 2005-08-29 1.126e+10 2005
## 3 SURGE TX 2008-09-12 4.000e+09 2008
## 4 SURGE TX 2008-09-12 5.000e+08 2008
## 5 SURGE LA 2005-09-23 4.320e+08 2005
## 6 SURGE LA 2008-09-12 7.500e+07 2008
## 7 SURGE FL 1993-03-13 5.000e+07 1993
## 8 SURGE FL 1993-03-13 5.000e+07 1993
## 9 SURGE MD 2003-09-19 5.000e+07 2003
## 10 SURGE NC 2011-08-26 4.000e+07 2011
## # ... with 65 more rowsdmgsby.rain <- BMKbind[grepl("RAIN|rain", BMKbind$event),]
dmgsby.rain <- mutate(dmgsby.rain,
event = "RAIN", dyear = year(bgndate))
toplot.rain <- group_by(dmgsby.rain , dyear, event)
toplot.rain## Source: local data frame [182 x 5]
## Groups: dyear, event [19]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 RAIN LA 1995-05-08 2.50e+09 1995
## 2 RAIN CA 1998-05-01 2.00e+08 1998
## 3 RAIN CA 2005-01-07 1.90e+08 2005
## 4 RAIN CA 1998-05-01 7.36e+07 1998
## 5 RAIN CA 2005-01-07 7.00e+07 2005
## 6 RAIN AL 1994-07-03 5.00e+07 1994
## 7 RAIN ID 1995-05-26 5.00e+07 1995
## 8 RAIN ND 1994-07-03 5.00e+07 1994
## 9 RAIN CA 1995-03-01 4.80e+07 1995
## 10 RAIN CA 2005-02-18 4.00e+07 2005
## # ... with 172 more rowsdmgsby.drought <- BMKbind[grepl("DROUGHT|drought", BMKbind$event),]
dmgsby.drought <- mutate(dmgsby.drought,
event = "DROUGHT", dyear = year(bgndate))
toplot.drought <- group_by(dmgsby.drought , dyear, event)
toplot.drought## Source: local data frame [203 x 5]
## Groups: dyear, event [17]
##
## event state bgndate dlstot dyear
## <chr> <fctr> <dttm> <dbl> <dbl>
## 1 DROUGHT TX 2006-01-01 1000000000 2006
## 2 DROUGHT IA 1995-08-01 500000000 1995
## 3 DROUGHT IA 2003-08-01 645150000 2003
## 4 DROUGHT IA 2001-08-01 578850000 2001
## 5 DROUGHT TX 2000-11-01 515000000 2000
## 6 DROUGHT OK 1998-07-06 500000000 1998
## 7 DROUGHT PA 1999-07-01 500000000 1999
## 8 DROUGHT NE 2002-12-01 480000000 2002
## 9 DROUGHT TX 1998-12-01 450000000 1998
## 10 DROUGHT TX 2001-12-01 420000000 2001
## # ... with 193 more rows- 1.2.4.1 Data set
toplot.alltogetherbinds the rows of all the previous obtained sets that are going to be drawn in the Fig. 3 plot of below RESULTS section.
toplot.alltogether <- rbind(toplot.flood, toplot.hurricane, toplot.surge,
toplot.tornado, toplot.tropical, toplot.fire,
toplot.blizzard, toplot.hail, toplot.heat2,
toplot.winterstorm, toplot.rain, toplot.drought)2. RESULTS
2.1 Personal Issues results
- 2.1.1 Total deaths due to heat events:
sum(fatsby.heat$fatalities)## [1] 3138- 2.1.2 Total deaths due to tornadoes:
sum(fatsby.tndo$fatalities)## [1] 5661- 2.1.3 Time series plot showing fatalities by heat events.
Code for Fig. 1 takes previously preprocessed object toplot.heat as argument for ggplot object ggheat. X (years) and Y (fatalities) intersections painted as dots of size 4 and transparency of 50% are coloured by type of event.
ggheat <- ggplot(toplot.heat, aes(initdate2, fatalities))
ggheat + geom_point(aes(color = event), size = 4, alpha = 1/2) +
scale_y_continuous(limits = c(0, 600), breaks = seq(0, 600, by = 50)) +
labs(title = " Deaths by Heat Events") +
labs(x = "Year", y = "Number of fatalities",
caption = "Fig. 1 NOOA Storm Database. Deaths by Heat Events")
Note. The highest point in the plot corresponds to 583 deaths in one day of the widely documented 1995 Chicago heat wave here referenced from the Wikipedia.
- 2.1.4 Time series plot showing fatalities by tornadoes.
Code for Fig. 2 takes previously preprocessed object toplot.tndo as argument for ggplot object ggtndo. X (years) and Y (fatalities) intersections painted as dots of size 4 and transparency of 50% are coloured by type of event.
ggtndo <- ggplot(toplot.tndo, aes(initdate2, fatalities))
ggtndo + geom_point(aes(color = event), size = 4, alpha = 1/2) +
scale_y_continuous(limits = c(0, 600), breaks = seq(0, 600, by = 50)) +
labs(title = " Deaths by Tornado Events") +
labs(x = "Year", y = "Number of fatalities",
caption = "Fig. 2 NOOA Storm Database. Deaths by Tornado Events")
Marginal Note. A single life loss can show its value for this paper, increasing the number of fatalities in section 1.1.1 and looking the effect in the Fig. 1 plot. E.g., when reaching the fatalities > 9 conditional, more than ten years of deaths by heat disappear from the plot; and both plots lose a lot of density at their bottom lines; and in year 2004 there was only one death by heat, as point for year 2004 disappear from the plot bottom line at fatalities > 1
2.2 Propery/Infrastructure and Crops Issues results
- 2.2.1 Time series plot showing damages in Property/Infrastructure and Crops.
ggdmgs <- ggplot(toplot.alltogether, aes(bgndate, dlstot/1000000))
ggdmgs + geom_point(aes(color = event), size = 4, alpha = 1/2) +
labs(title = " Damages in Property/Infrastructure and Crops") +
labs(x = "Year", y = "Millions of US Dollars",
caption = "Fig. 3 NOOA Storm Database. Property/Infrastructure and Crops Issues")
Note. The events of the first days of January of year 2006 reflected in the 115 billion dollars by flood damages reported in this plot, are in depth described in the
Storms and Flooding in California in December 2005 and January 2006 paper from the U.S. Department of the Interior, and the U.S. Geological Survey.
3. FINAL REMARKS
Data gathering of fatalities due to heat events begins around 1993, while fatalities by tornados start in the early 1950s. In general, recordings of fatalities lie beneath the 50 deaths by event line. With no further analysis, the damages in Property/Infrastructure and Crops evolve towards events (reaching peaks in years 2005 and 2006) related to heat and drought events, seemingly showing correspondence with “El niño” and “La niña” latest aggravated phenomenons in the Pacific Ocean warmest shore regions.