Closed crjones-amath closed 6 years ago
Symptom: ne4 run with m2005 SAM microphysics and 15 minute GCM timestep crashed on step 2 with following error:
ERROR: *** parampollu_check_adjust_inputs - bad acen_tfin 1 0.0000000E+00
This is the same symptom noted by @whannah1 in PR #37
Cause: When the GCM timestep is larger than 10 minutes, it must evenly be divided by 10 minutes. Otherwise, mod(itavg2, ntavg2) > 0 at the end of the CRM evaluation, so area_cen_final gets zeroed out. https://github.com/E3SM-Project/ACME-ECP/blob/800dc0826da55c09a2f258e50aaaf11490a6c4ca/components/cam/src/physics/crm/module_ecpp_crm_driver.F90#L650-L654 This eventually gets passed to acen_tfin, which fails the check in parampollu_check requiring that the total area for each diagnosed ECPP class (quiescent/updraft/downdraft) sums to 1 at each vertical level.
mod(itavg2, ntavg2) > 0
area_cen_final
parampollu_check
Symptom: ne4 run with m2005 SAM microphysics and 15 minute GCM timestep crashed on step 2 with following error:
This is the same symptom noted by @whannah1 in PR #37
Cause: When the GCM timestep is larger than 10 minutes, it must evenly be divided by 10 minutes. Otherwise,
mod(itavg2, ntavg2) > 0at the end of the CRM evaluation, soarea_cen_finalgets zeroed out. https://github.com/E3SM-Project/ACME-ECP/blob/800dc0826da55c09a2f258e50aaaf11490a6c4ca/components/cam/src/physics/crm/module_ecpp_crm_driver.F90#L650-L654 This eventually gets passed to acen_tfin, which fails the check inparampollu_checkrequiring that the total area for each diagnosed ECPP class (quiescent/updraft/downdraft) sums to 1 at each vertical level.