Issues with Climdex Indices

[Pouriya-J]

@ssobie

• The dtr (Daily temperature range) for CanESM5 model (all realizations) does not look right. It seems as if the calculation was paused and resumed.

This can be seen in time-series plot of the annual averages for dtr (i.e., CanESM5_ssp585_r9i1p2f1).

• As previously noted (Missing Climdex Indices #27 ), the EC-Earth3-Veg_ssp126_r1i1p1f1 was missing 13 indices. In addition to that, the calculation for "su30ETCCDI_ann_BCCAQ2v2+ANUSPLIN300_EC-Earth3-Veg_historical+ssp126_r1i1p1f1_1950-2100.nc" is not done correctly.

[cpomer10]

Hey @Pouriya-J just adding Charles comments here:

1) The replacement files we sent for EC-Earth3-Veg should have addressed the glitch in SU: could you please regenerate that t-series plot to confirm? That should allow you to proceed to generate the multi-model means for this index.

2) Stephen traced the problem seen in DTR in CanESM5 (and subsequently detected in the few other GCMs that project a large change in this index) to the QDM part of the BCCAQv2 calculation that uses a 30-yr period. Some experimenting with this period has led to improvements in this behaviour, which will be used to improve the MBCn downscaled products. At present, however, we do not plan to regenerate the BCCAQv2 outputs for these models.

Since most models project very small changes in DTR over the century, we suspect this issue will not be visible in the ensemble mean. Therefore, we recommend computing the ensemble mean DTR as a check, and post the results for both indices here to see whether any further follow-up is needed.

[Pouriya-J]

su30ETCCDI_ann_BCCAQ2v2+ANUSPLIN300_EC-Earth3-Veg_historical+ssp126_r1i1p1f1_1950-2100.nc Looks good!

[Pouriya-J]

Here is the ensemble mean DTR for all models (including the CanESM5_r10i1p2f1). The drop is visible, mostly in ssp585, but the changes in DTR are very small. @ssobie