cdo eca_hwfi wrong result for cdo-1.9.10 (correct result for cdo-1.9.9)
Added by Laura Hövel over 3 years ago
If I calculate the HWFI as illustrated below, the result forthe "warm spell days index wrt 90th percentile of reference period" is different depending on the cdo version I use.
Using cdo 1.9.10 gives a unrealisticly high count of heatwave days (more than 1000 per year for some gridcells) and using the older version cdo 1.9.9 gives a realistic result (well below 365 per year). The result for the warm spell periods per time period are the same in both cdo versions.
cdo eca_hwfi,6 HIPPO_n22_tsurf_1961.nc temp_n22_runpctl_tsurf_1961.nc HW_HIPPO_n22_tsurf_1961.nc
Thanks for your help!
Cheers,
Laura
temp_n22_runpctl_tsurf_1961.nc (25.7 MB) temp_n22_runpctl_tsurf_1961.nc | 90th percentile threshold | ||
HIPPO_n22_tsurf_1961.nc (25.7 MB) HIPPO_n22_tsurf_1961.nc | Daily mean Temperature |
Replies (1)
RE: cdo eca_hwfi wrong result for cdo-1.9.10 (correct result for cdo-1.9.9) - Added by Fabian Wachsmann over 3 years ago
Dear Laura,
thanks for raising this issue.
I introduced a keyvalue argument with 1.9.10 named "freq" which can be "year" or "month". Unfortunately, I added two bugs to the code while trying to make it compatible with older versions.
I debugged this and it will be made available in a new release.
Best,
Fabi