Currently, our error bounds are based on a simple heuristic that looks at the range in the input data. Ideally, we want to replace this with a more sophisticated mechanism which is based either on expert soliciation or analysis of the ERA5 ensembles. @juntyr is working on developing the bounds.
Currently, our error bounds are based on a simple heuristic that looks at the range in the input data. Ideally, we want to replace this with a more sophisticated mechanism which is based either on expert soliciation or analysis of the ERA5 ensembles. @juntyr is working on developing the bounds.