Skip to content

readme.md

Latest commit

 

History

History
71 lines (51 loc) · 1.64 KB

readme.md

File metadata and controls

71 lines (51 loc) · 1.64 KB

MLGEFS: Machine Learning-based Global Ensemble Forecast System

This package contains scripts to run and an ensemble-based cascaded version of the GraphCast weather model for the Global Ensemble Forecast System (GEFS). It also provides the pre-trained model (weights) to run MLGEFS:

Table of Contents

Overview

The National Centers for Environmental Prediction (NCEP) provides GEFS data that can be used for ensemble weather prediction and analysis.

Prerequisites and Installation

To install the package, run the following commands:

conda create --name mlgefs python=3.10
conda activate mlgefs
pip install dm-tree boto3 xarray netcdf4
conda install --channel conda-forge cartopy
pip install --upgrade https://github.com/deepmind/graphcast/archive/master.zip
pip install ecmwflibs
pip install iris
pip install iris-grib

This will install the packages and most of their dependencies.

Usage

Input data

python gdas_utility.py YYYYMMDDHH -l 13 -m wgrib2(or pygrib) -s s3 -l /path/to/output -d /path/to/download -k no

Run the model

python run_gencast.py -i /path/to/inputfile -w /path/to/model/ -l lead_time(steps) -m num_of_ensemble_members -o /path/to/output -p num_of_pls -u yes(no) -k yes(no)

Output

Contact

For questions or issues, please contact: Linlin.Cui@noaa.gov Jun.Wang@noaa.gov