search

ISI-MIP / attrici

Produce counterfactual climates for ISIMIP.
GNU General Public License v3.0
9 stars 2 forks source link

Compare GSWP3 to newer ERA5-W5E5 #69

Closed matthiasmengel closed 4 years ago

matthiasmengel commented 4 years ago

We have not decided, which dataset to put out first for attributions runs. Should be decided based on our analysis. So we should run W5E5 as well and see which one we trust more.

matthiasmengel commented 4 years ago

As discussed with Katja. One important test would to compare the trends in the non-counterfactuals datasets. Over the same time period, like 1900-2010, are the trends similar or very different. If very different, that would mean trends are induced by the concatenation of the two datasets, which would then produce spurious trends in the counterfactuals.

matthiasmengel commented 4 years ago

@SimonTreu

SimonTreu commented 4 years ago

so if trends are similar, take ERA5-W5E5 if trends differ, take GSWP3 ?

matthiasmengel commented 4 years ago

Pseudocode, ha! Yes, I would think as you suggest. There may be other measures as well that influence our decision.

matthiasmengel commented 4 years ago

This is largely done. Close here.