Dataset overview
GPC/m is a daily precipitation dataset produced using three machine-learning methods with outgoing longwave radiation (OLR) and atmospheric circulation data from reanalysis. It is designed for climatological analyses, including composite, correlation, variability, and climate-change studies.
wget or curl. No Dropbox page, login, or
browser-only download procedure is required.
Scientific precipitation values are unchanged from the original v1-2024
release. The reorganized files use the variable name pr and
coordinate names lat and lon.
Example animations
The following animations show examples of the daily GPC/m precipitation fields for 2015. They are embedded as ordinary MP4 files in the same directory as this HTML page. Playback starts only when selected, avoiding an unnecessary automatic download of both videos.
Open or download the MP4 file
Open or download the MP4 file
These animations were created using GrADS CAMO.
Yearly files: 1979–2020
Each year is a separate NetCDF file. The corresponding CTL file is a small descriptor for GrADS; the precipitation data themselves are in the NetCDF file.
Terminal download examples
Download one year
wget -c https://camo.fpark.tmu.ac.jp/data/gpcm1/pr-day.y2020.GPCm-v1-2024.ncDownload a range of years
for year in $(seq 2001 2010); do
wget -c "https://camo.fpark.tmu.ac.jp/data/gpcm1/pr-day.y${year}.GPCm-v1-2024.nc"
doneDownload all yearly NetCDF files
for year in $(seq 1979 2020); do
wget -c "https://camo.fpark.tmu.ac.jp/data/gpcm1/pr-day.y${year}.GPCm-v1-2024.nc"
done
The -c option allows an interrupted download to resume.
The direct file URLs used above are identical to the links in the table.
Complete archive and citation
The independent yearly files above are intended for convenient direct download from this web server. The complete archived release, DOI, license, and contact information remain available through Zenodo.
Creator: Hiroshi G. Takahashi, Tokyo Metropolitan University
License: CC BY-NC 4.0
Related paper
Takahashi, H. G. (2024), GPC/m: Global Precipitation Climatology by Machine Learning. Preprint
Calendar note: the files use calendar = "365_day".
February 29 is not included, including in leap years. See the README
before interpreting time coordinates.