Replication materials for Jing, R., Heft-Neal, S., Chavas, D.R. et al. Global population profile of tropical cyclone exposure from 2002 to 2019. Nature (2023). https://doi.org/10.1038/s41586-023-06963-z

The materials in this repository reproduce the figures, tables, and calculations appearing in the main text and extended data of the paper.
If you find meaningful errors in the code or have questions or suggestions, please contact Renzhi Jing at jingrenzhi.go@gmail.com

Organization of repository

• scripts: scripts for replication of figures, tables, and calculations.
• figures/published: published versions of the figures.
• figures/raw: scripts plot_Figure*.R will generate pdf figures in this directory.
• data/tc: gridded high-resolution tropical cyclone exposure data
• data/worldpop: high-resolution gridded population data
• data/misc: uncategorized data that is needed for the paper
• results: folder to save intermediate results

Data

• *data/tc *(All files under this path can be generated by unzipping the provided {year}.zip files.)

  • duration  tropical cyclone exposure assuming up to 6 hours, 12 hours and no limit on duration of overland sustained wind. The temporal resolution is 3 hour. The files follow a structured naming convention: duration_{year}{wind_level}{landfall}.tif For instance, a file named duration_2002_cat1_12h.tif contains gridded data representing the duration that each grid endured the impact of Category 1 tropical cyclones, assuming up to 12 hours of sustained winds over land. With a temporal resolution of 3 hours, a duration value of 2 indicates that in 2002 the grid was exposed for a total of 6 hours.

• *data/worldpop *(We note that all files under this path can be downloaded using the links provided below. We provide this path to align with subsequent analysis and visualization)

  • worldpop_total  high-resolution gridded population data from WorldPop, downloaded at https://hub.worldpop.org/geodata/listing?id=29. The datasets are available for each year from 2000 to 2019 with a resolution of 30 arc seconds.
  • worldpop_age_gender  high-resolution gridded age and sex distributions data from Worldpop, downloaded at https://data.worldpop.org/GIS/AgeSex_structures/. Similarly, the datasets are available for each year from 2000 to 2019 with a resolution of 30 arc seconds.

• *data/misc *(We note that all files under this path can be downloaded using the links provided below. We provide this path to align with subsequent analysis and visualization)

  • gadm_410.gpkg  GADM data version 4.1, providing maps and spatial data for all countries and their sub-divisions. The data is downloaded at: https://gadm.org/data.html.
  • world_countries_2020  country boundaries from IPUMS, downloaded at: https://international.ipums.org/international/gis.shtml.
  • world_continent  continent boundaries downloaded from ArcGIS Hub, downloaded at: https://hub.arcgis.com/datasets/esri::world-continents/about.
  • global_mask.tif   2D array with the same resolution as the worldpop data, which is only used to generate a dictionary, saving the indices of the map for each country and continent.
  • povmap-grdi-v1_high_res_global.tif   high-resolution gridded relative deprivation index. The raw data is downloaded at SEDAC https://sedac.ciesin.columbia.edu/data/set/povmap-grdi-v1 with a resolution of ~1 km. We adjust the map's resolution to align with that of the Worldpop data.

  • supplementary_table1.csv   data used to generate supplementary table 1. The columns are:

    • country: Country name
    • avg_person_day: Averaged person-days exposures during 2002-2019 for each country
    • share: Share of global person-days exposure during the study period for each country
    • avg_rdi: Population-averaged relative deprivation index for each country

    • rdi_rate: The ratio of population-averaged relative deprivation index between exposed and unexposed population for each country

      'NA' means not applicable.

  • ibtracs_data_1989_2019.csv   historical tropical cyclone tracks, derived from IBTrACS dataset. The tracks are used in Figure 1. Missing values are filled with 'NA'. The columns are:

    • sid: Storm identifier
    • basin: Basins include: NA - North Atlantic EP - Eastern North Pacific WP - Western North Pacific NI - North Indian SI - South Indian SP - Southern Pacific SA - South Atlantic
    • name: Storm name provided by the agency
    • date_time, year, month, day, hour: Time in Universal Time Coordinates (UTC).
    • tclat, tclon: Storm latitude and longitude
    • max_wind: Maximum sustained wind speed for the current location, in knots
    • rmax: Radius of max winds for the current location, in nmile
    • tspd: Storm translation speed, in m/s

Scripts

Python files calculate intermediate data that is used to generate the figures. The scripts are maintained at: https://github.com/jingrenzhi/tropical_cyclone_exposure/tree/main/scripts

• Script helper_functions.py includes global variables and self-defined functions.
• Script script_Figure*.py replicate the calculations reported in the paper.

Figures

R files (plot_Figure*.R) generate the figures in the paper and write them to figures/raw. The figures produced by these scripts will be slightly visually different than the published figures because post-processing was done in Adobe Illustrator. Published versions of the figures are available in figures/clean. The scripts are maintained at: https://github.com/jingrenzhi/tropical_cyclone_exposure/tree/main/scripts

• Script plot_Figure1.R generates Figure 1.
• Script plot_Figure2.R generates Figure 2, Figure ED2, Figure ED3 and Figure ED8.
• Script plot_Figure3.R generates Figure 3, Figure ED4.
• Script plot_Figure4.R generates Figure 4 and Figure ED6.
• Script plot_Figure5.R generates Figure 5 and Figure ED7.
• Script plot_FigureED1.R generates Figure ED1.
• Script plot_FigureED5.R generates Figure ED5.

Code/Software

Scripts were written in Python 3.6.1 and R 4.2.3.
[Link to GitHub repo with all the replication data and code] (https://github.com/jingrenzhi/tropical\\_cyclone\\_exposure)

Python packages required

• numpy
• math
• pandas
• os
• time
• pickle
• geopandas
• rasterio
• itertools
• scipy
• osgeo
• shapely
• global_land_mask
• multiprocessing for parallel computing

R packages required

• ggplot2
• pracma
• scales
• dplyr
• tidyr
• tidyverse
• reshape2
• sf
• sp
• MetBrewer