Story 5 - The Effect of Earth’s Temperatures on Cyclonic Storms

Packages Used

Show Packages Used

library(tidyverse)
library(data.table)
library(janitor)
library(scales)
library(patchwork)

Load/Tidy Data

Show code for loading data

Earth’s Average Temperature

The original data represents temperature measured as difference from the average temperature for the baseline period of 1951 -1980; this baseline temperature, according to NASA, is 14 degrees Celsius. I chose to show the yearly average temperature as their actual values in Fahrenheit, so I added 14 to all values and applied the Fahrenheit conversion.

Source: NASA’s Goddard Institute for Space Studies

temp_data <- read_table("earth_temp.txt") |> 
  select(year = Year, avg_temp = No_Smoothing) |> 
  mutate(year = as.numeric(year),
         avg_temp = ((avg_temp + 14) * 9/5) + 32) |> 
  filter(year %in% 1980:2024)

Storm Intensity

The storm data contains wind and pressure telemetry measures for cyclonic storms from several agencies across the globe. I selected only the highest wind and pressure measures for each storm. To do this, I found the max reading from each agency per storm, eliminating the time-stamp granularity. Then, for each storm I kept only the agency measurement that was highest.

Source: National Oceanic and Atmospheric Administration (NOAA)

data <- read_csv("storm_data.csv", na = c("", "NaN", "null", "NULL")) |> 
  clean_names() |>  
  select(sid, season, basin, name, contains("wind"), contains("pres")) |> 
  mutate(across(where(~ is.character(.x) && any(grepl("^[+-]?[0-9.]+$", .x))), ~ as.numeric(.x))) |>  
  filter(season %in% 1980:2024) |> 
  filter(!if_all(contains("wind"), is.na),
         !if_all(contains("pres"), is.na))

DT <- as.data.table(data)

# Some storms were in >1 basin so we are selecting the basin where they spent the most time
basin_mode <- DT[
  , .N, by = .(sid, basin)
][
  order(sid, -N)
][
  , .SD[1], by = sid
][
  , .(sid, basin)
]

key_cols <- "sid"
value_cols <- grep("wind|pres", names(DT), value = TRUE)

# Selecting the max wind and pressure values from each agency for each storm
df <- DT[, lapply(.SD, max, na.rm = TRUE),
         by = key_cols,
         .SDcols = value_cols]

df2 <- merge(
  basin_mode,
  df,
  by = "sid",
  all.x = TRUE
)

meta <- unique(DT[, .(sid, season, name)])
df3 <- merge(meta, df2, by = "sid", all.x = TRUE)

wind_cols <- grep("wind", names(DT), value = TRUE)
pres_cols <- grep("pres", names(DT), value = TRUE)


# Selecting the max wind and pressure among all agency measures
storm_extremes <- DT[
  ,
  {
    list(
      max_wind = max(as.matrix(.SD[, ..wind_cols]), na.rm = TRUE),
      max_pres = max(as.matrix(.SD[, ..pres_cols]), na.rm = TRUE)
    )
  },
  by = sid
]

meta <- df3[, .(sid, season, name, basin)]
storm_data <- merge(meta, storm_extremes, by = "sid", all.x = TRUE)

basin_lookup <- tibble(
  basin_abbv = c("NA", "EP", "WP", "NI", "SI", "SP", "SA"),
  basin_full = c(
    "North Atlantic",
    "Eastern North Pacific",
    "Western North Pacific",
    "North Indian",
    "South Indian",
    "Southern Pacific",
    "South Atlantic"
  )
)

storm_data <- as.tibble(storm_data) |> 
  rename(year = season) |> 
  left_join(basin_lookup, by = join_by(basin == basin_abbv)) |> 
  mutate(basin = basin_full,
         max_pres = as.numeric(max_pres)) |> 
  select(-basin_full)

Combine Data

storms <- storm_data |> 
  left_join(temp_data, by = "year")

print(storms, n=5)

## # A tibble: 4,260 × 7
##   sid            year name            basin           max_wind max_pres avg_temp
##   <chr>         <dbl> <chr>           <chr>              <dbl>    <dbl>    <dbl>
## 1 1979238S08085  1980 TONY            South Indian          45     1005     57.7
## 2 1979343S08093  1980 CLAUDETTE:VIOLA South Indian         115     1006     57.7
## 3 1979344S12170  1980 OFA             Southern Pacif…       55      997     57.7
## 4 1979357S10095  1980 DANITZA:WILF    South Indian          70     1005     57.7
## 5 1980001S13173  1980 PENI            Southern Pacif…       65      997     57.7
## # ℹ 4,255 more rows