FIXED: bad tasmin and tasmax values in ISIMIP2a/3a/3b climate input data

Posted by Martin Park on June 3, 2020

Dear ISIMIP modelers,

The bug in the ISIMIP3BASD code that caused the badly bias-adjusted tasmin and tasmax values has been fixed and we have re-done the bias adjustment of these variables for all five primary GCMs. All files that were affected by the bug have been replaced at PIK, DKRZ and ESGF. Detailed information on which files and grid cells were affected can be found in the caveat reports for GFDL-ESM4, IPSL-CM6A-LR, MRI-ESM2-0, and UKESM1-0-LL. No files were affected for MPI-ESM1-2-HR. Data for the secondary GCMs will be replaced this week.

Below you find an overview of which files were affected per scenario and GCM together with information about which simulations should be re-done to be included in the ISIMIP repository, based on the replaced input data (see the legend below this table).

The bug also affected the tasmax and tasmin in the datasets GSWP3-EWEMBI and GSWP3-W5E5 provided in ISIMIP2a and ISIMIP3a because ISIMIP3BASD was used to harmonize GSWP3 with EWEMBI and W5E5 in these datasets. Also these harmonizations were re-done and tasmax and tasmin data replaced for 1901-1978. Since almost only ocean grid cells were affected (see the detailed caveat reports for GSWP3-EWEMBI and GSWP3-W5E5), model re-runs over land are deemed unnecessary in these two cases. The ISIMIP3a counterclim datasets were not changed since they only contain data over land.

Best wishes,
the ISIMIP team at PIK