search

openclimatefix / graph_weather

PyTorch implementation of Ryan Keisler's 2022 "Forecasting Global Weather with Graph Neural Networks" paper (https://arxiv.org/abs/2202.07575)
MIT License
179 stars 44 forks source link

Pull in recent observations to model #64

Open jacobbieker opened 1 year ago

jacobbieker commented 1 year ago

We could try pulling in new observations when running the model, somewhat faster data assimilation essentially. Could use something like https://github.com/blaylockbk/SynopticPy

Detailed Description

Context

If we can have a model that can create fast forecasts, including observations could mean better forecasts, similar to how we can do that for PV forecasting by using live readings from PV data.

Possible Implementation

jacobbieker commented 10 months ago

One method could be to use https://github.com/tom-andersson/deepsensor for creating grids from the observations to take as inputs.