Open source dataset loading and creation from Planetary Computer, GCP, and AWS. To support reproducible training of weather, energy, and mapping models.
A weather forecasting dataset would be ideal to have, to more easily train weather forecasting models. Most use ERA5, as it has a long time period and is consistent, although there are issues with ERA5, especially around precipitation, clouds, and radiation.
An interest would be to also allow for mixed model usage (i.e. global data from ERA5/GFS/etc, but have a high resolution model like HRRR for over part of the globe)
A weather forecasting dataset would be ideal to have, to more easily train weather forecasting models. Most use ERA5, as it has a long time period and is consistent, although there are issues with ERA5, especially around precipitation, clouds, and radiation.
An interest would be to also allow for mixed model usage (i.e. global data from ERA5/GFS/etc, but have a high resolution model like HRRR for over part of the globe)
But for an initial one, this could easily use