Re:NOAA upgrades the U.S. global weather forecast model
by naavis on Friday June 14, 2019 @05:41AM
(#20181647)
One fundamental problem we have with weather forecasts is that our input data for the starting point is fairly sparse. GFS calculates the forecast on a grid with 13 km horizontal resolution and 64 vertical layers. We don't have accurate weather information at that resolution from all over the globe, so the starting point for the forecasts is a combination of previous simulations and interpolated observational data.
So even if we had a forecast engine that would perfectly simulate everything given some start state, we wouldn't have enough input data to have an accurate start state.
Re:NOAA upgrades the U.S. global weather forecast model
by nestorD on Friday June 14, 2019 @02:49AM
(#20181001)
You can also use ML to learn a metamodel : a model trained on an accurate (but too costly to be run in real time) simulation.
Re:NOAA upgrades the U.S. global weather forecast model
by SubiculumCode on Friday June 14, 2019 @04:45AM
(#20181415)
Almost like super resolution techniques?
Re:NOAA upgrades the U.S. global weather forecast model
by namibj on Friday June 14, 2019 @05:36AM
(#20181624)
Pretty much.
Maybe OpenAI will try to throw transformers at it.
Re:NOAA upgrades the U.S. global weather forecast model
by lambdamore on Friday June 14, 2019 @01:05PM
(#20184571)
We can compute the radiative transfer processes to infer the physical processes from the satellite images.
Machine learning is all about finding an unknown function that underlies known data. This is sort of the opposite issue: we know the underlying function but can’t compute it.
That said, ML models are being applied to situations where the whether data doesn’t translate directly into known physical quantities, like satellite images (see https://developmentseed.org/projects/hurricane-intensity/)