Description
The Global Environmental eMuLator (GEML) is a component of the experimental Global Deterministic Prediction System (GDPS) and is an artificial intelligence (AI)-based weather emulator trained on past atmospheric states. More specifically, this GEML model is based on data compatible with the ¼°, 13-level version of the GraphCast model (Lam et al. 2023) from DeepMind. It was trained and refined by ECCC, using ECMWF's ERA5 data (1979-2016) and operational analyses (2016-2021). The weights have been recalculated and are also available to the public. Forecasts are carried out twice daily, each with a 10-day lead time. It generates the reference large-scale temperature and horizontal wind fields, toward which GDPS's GEM forecasts are spectrally nudged. The geographical coverage is global with a horizontal resolution of 28 km. Data is available on 13 pressure levels, and employs a uniform latitude-longitude grid with 0.25-degree grid resolution. Six atmospheric variables defined on the 13 pressure levels, along with 4 surface variables are available every 6 hours.
Resources
| Name | Format | Description | Link |
|---|---|---|---|
| Datamart du SMC AMQP | GRIB2 | ||
| MSC Datamart AMQP | GRIB2 | ||
| Datamart du SMC | GRIB2 | Open resource | |
| MSC Datamart | GRIB2 | Open resource | |
| MSC Open Data documentation | HTML | Open resource | |
| Documentation des données ouvertes du SMC | HTML | Open resource | |
| SGPD-GEML - Température de l'air à 2m au dessus de la surface [°C] [expérimental] | WMS | Open resource | |
| GDPS-GEML - Air temperature at 2m above ground [°C] [experimental] | WMS | Open resource |