Metadata-Version: 2.4
Name: satrain
Version: 1.0
Summary: A machine-learning benchmark dataset for satellite-based precipitation retrievals
Author-email: Simon Pfreundschuh <simon.pfreundschuh@colostate.edu>
Maintainer-email: Simon Pfreundschuh <simon.pfreundschuh@colostate.edu>
License-Expression: MIT
Project-URL: Source, https://github.com/simonpf/satrain/
Project-URL: Docs, https://satrain.readthedocs.io
Keywords: machine-learning,benchmark,dataset,meteorology,geophyisics
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE.txt
Requires-Dist: appdirs
Requires-Dist: click
Requires-Dist: ipywidgets
Requires-Dist: numpy<2
Requires-Dist: netCDF4<1.7
Requires-Dist: h5netcdf
Requires-Dist: hdf5plugin
Requires-Dist: jupyter-book
Requires-Dist: matplotlib
Requires-Dist: requests-cache
Requires-Dist: rich
Requires-Dist: scipy
Requires-Dist: seaborn
Requires-Dist: toml
Requires-Dist: xarray
Provides-Extra: complete
Requires-Dist: pytest; extra == "complete"
Requires-Dist: torch; extra == "complete"
Requires-Dist: torchvision; extra == "complete"
Requires-Dist: lightning; extra == "complete"
Requires-Dist: cartopy; extra == "complete"
Dynamic: license-file

# SatRain: A machine-learning-ready benchmark dataset for satellite precipitation estimation

The Satellite Precipitation Retrieval (SatRain) benchmark dataset developed by the
International Precipitation Working Group (IPWG) is a benchmark dataset for
machine-learning-based satellite precipitation retrievals, i.e., algorithms for
detecting and quantifying precipitation from satellite imagery.

## Features

 - Training, validation, and test splits derived from four years of overpasses of passive-microwave sensors
   over the conterminous united states (CONUS)
 - Collocated satellite observations from visible, infrared, and microwave sensors
 - Comprehensive ancillary data covering atmospheric parameters and surface properties
 - Independent *testing* datasets derived from different regions and measurement techniques
 - Flexible evaluation framework
 

![Precipitation estimates from three SatRain-based retrievals  applied to observations from Typhoon Khanun during landfall on
the Korean peninsula](docs/figures/retrieval_example.png)]

Retrieved precipitation from three SatRain-trained retrievals applied to
observations of Typhoon Khanun during landfall on the Korean peninsula. Each
retrieval relies on a different type of input from the SatRain dataset: a single
IR window channel (panel d), Himawari-9 observations (panel e), and GMI
observations (panel f). The results are compared with reference ground-based
radar measurements (panel a) and baseline estimates from ERA5 (panel b) and
GPROF V7 (panel c).

## Documentation

For instructions on how to get started using the dataset refer to the documentation available [here](https://satrain.readthedocs.io/en/latest/intro.html).
