#2581 closed Defect (fixed)
icb speed limit and speeding ticket
| Reported by: | mathiot | Owned by: | systeam |
|---|---|---|---|
| Priority: | low | Milestone: | Unscheduled |
| Component: | ICB | Version: | |
| Severity: | minor | Keywords: | ICB, |
| Cc: |
Description
Context
In the icb module, we can enforce the icb speed limit via rn_speed_limit and issue a speeding ticket if needed. Scaling down the icb speed at the end of icb_accel does not have any effect and the collection of speeding ticket is hard to interpret (all domains report the same number of speeding tickets, even if no icb on the domain).
Analysis and recommandation
- Icb speed is not an output of icb_accel, so scaling the speeding down according to the cfl will not have any effect on the simulation. Probably we should scale down the velocity and recompute the acceleration accordingly.
- About the icb report, I suggest to remove the mpp_sum on nspeeding_ticket in icb_dia. Like this, every process will show the number of speeding ticket on its domain.
- probably additional information on what cause the speeding ticket could be useful (is it the atmo, the oce, the ice ...)
It affects the trunk and v4.
Commit History (8)
| Changeset | Author | Time | ChangeLog |
|---|---|---|---|
| 14400 | mathiot | 2021-02-04T19:49:00+01:00 | ticket #2581: merge ticket branch into trunk |
| 14399 | mathiot | 2021-02-04T19:43:16+01:00 | ticket #2581: update branch to head of the trunk |
| 14382 | mathiot | 2021-02-03T14:12:29+01:00 | ticket #2581: implement change made in v4 to trunk version |
| 14373 | mathiot | 2021-02-02T18:46:10+01:00 | ticket #2581: build branch for correction in trunk |
| 14372 | mathiot | 2021-02-02T18:42:36+01:00 | ticket #2581: merge ticket branch into r4.0-HEAD after Dave review |
| 14361 | mathiot | 2021-01-29T14:52:00+01:00 | ticket #2581: correction based on Dave review (missing _wp, improve information wrote in iceberg.stat file) |
| 14356 | mathiot | 2021-01-28T10:30:26+01:00 | ticket #2581 (NEMO_4.0-HEAD version): create branch |
| 14268 | davestorkey | 2021-01-06T09:46:41+01:00 | UKMO branch to implement Pierre's version of the iceberg speed limiter (ticket #2581). |
Change History (13)
comment:1 Changed 3 years ago by davestorkey
comment:2 Changed 3 years ago by mathiot
In 14356:
comment:3 Changed 3 years ago by mathiot
As discuss with Dave, a safe value for iceberg cfl is 0.4 as icb ssh is evaluate at icb%xi+/-0.1 and extra haloes for variable at T points are not fill with data. Once icb and nn_hls will be compatible this restriction will goes away for case with nn_hls > 1.
comment:4 Changed 3 years ago by mathiot
In 14361:
comment:5 Changed 3 years ago by mathiot
In 14372:
comment:6 Changed 3 years ago by mathiot
In 14373:
comment:7 Changed 3 years ago by mathiot
In 14382:
comment:8 Changed 3 years ago by mathiot
In 14399:
comment:9 Changed 3 years ago by mathiot
No change requested by Dave S. on trunk version, ready to be merge in trunk.
comment:10 Changed 3 years ago by mathiot
In 14400:
comment:11 Changed 3 years ago by mathiot
- Resolution set to fixed
- Status changed from new to closed
comment:12 Changed 3 years ago by smasson
comment:13 Changed 3 years ago by mathiot
Change in results is normal with Dave we also fix a small bug in the call of icb_ground in the process:
The icb ground use as input non defined yet variable. It should be zu2 and zv2 for step 3 (instead of zu3 and zv3) and zu3 and zv3 for step 4 (instead of zu4 and zv4). Tested made without this change in r4.0 do not change the results.
In 14268: