We know all weather forecasts are wrong in some way, but can we take advantage of the agreement or disagreement between multiple weather sources to further boost our model’s accuracy? If three different weather sources are saying there will be a significant storm, we’d want our model to trust that more than if only one source was forecasting rain.
This paper by our frequent collaborators describes the approach we use to include multiple weather input sources into our model and quantifies the resulting accuracy improvements.
