NWP up sampling predictions paper

[jacobbieker]

https://arxiv.org/abs/2203.12297

Detailed Description

The paper shows making a Gan that can up sample global forecasts for precipitation to high resolution local forecasts, which is similar to something we want to try here, of training on various resolution graph inputs so that we can make high res forecasts over arbitrary areas.

Context

Possible Implementation

[jacobbieker]

Also, this paper uses ecmwrf ensemble global model which we could try with our gfs or gefs forecast data, hrrr and mrms data, which I'm already collecting and transforming for training dgmr, metnet, and this model on. Our dataset seems like it will be significantly larger, covering 2016 to now.

[jacobbieker]

Another similar kinda one using ground observations to correct ECMWF forecasts for 2m temperature and precipitation, could do similar, with the gauge-corrected rainfall rates over the whole US that are now on HuggingFace and the GFS operational forecasts, or ERA5: https://www.nature.com/articles/s41598-022-11936-9