search

judithberner / climpred_CESM1_S2S

MIT License
3 stars 1 forks source link

Add climatology as 'reference' #8

Closed judithberner closed 3 years ago

judithberner commented 3 years ago

@aaronspring You were waiting for this, correct? The docu says, open an issue :-) We really want to have the RPSS wrt climatology. The good thing is that - thanks to @bradyrx !!!!- we compute the climatology in https://github.com/judithberner/climpred_CESM1_S2S/blob/main/0.02_generate_climatology_S2S.ipynb

Can this be added as reference, please?

Related to issue https://github.com/judithberner/climpred_CESM1_S2S/issues/7 it should be considered if the reference should be computed outside and passed into hindcast2_djf.verify(metric="rps",...)

judithberner commented 3 years ago

@abjaye

aaronspring commented 3 years ago

Thanks @judithberner. Could you please open this issue in climpred and explain how a persistence or climatological forecast for probabilistic metrics looks like? For deterministic we just prescribe a constant or seasonally vary forecast but I don’t know what the assumption would be for probabilistic? What’s the member dim? Or is just clim probability needed?

aaronspring commented 3 years ago

Related issue about more reference forecasts: https://github.com/pangeo-data/climpred/issues/202

aaronspring commented 3 years ago

I went back into my SubX freeze example. There I used the observations.mean(‘time’) (maybe better time.day of year or time.weekofyear) as the climatology no resolution line in the reliability plot