Opened 11 years ago

Closed 6 years ago

#273 closed help (completed)

updating snow ancil in vn4.5

Reported by: agt Owned by: um_support
Component: UM Model Keywords: snow updating in slab model
Cc: Platform:
UM Version: 4.5

Description

Hi,

ages ago Jeff made me a userprestashmod which allowed hourly/daily etc updating of snow depth anomalies from ancil. This works great in hadam3. Now I want to repeat a couple of the experiments in the slab version of the model. I have been through the calibration run and now am doing the actual run. This seems to fail somewhere on the first day. An example is xdzta in directory /home/n02/n02/agt/work/um/xdzt
and with leave file: /home/n02/n02/agt/um/umui_out/xdzta000.xdzta.d09121.t191752.leave

The model fails with the classic negative theta detected. Originally I thought this related to the extreme sensitivity test I was using not being liked by the slab, but in the version indicated I am testing with updating the supplied qrclim.smow.newA version. If instead "not used" is selected, the slab run will begin, which leads me to think it is the updating itself that has become a problem. Is it possible the ~jeff/umui_jobs/userprestash/snowdepth_update somehow is in conflict with the slab model, or also needs adding to the userprestash section of the slab component?

Thanks, Andy

Change History (3)

comment:1 Changed 11 years ago by agt

(original pre-userstash mod was in ticket 138)

comment:2 Changed 11 years ago by agt

Ok I have more thoughts. Looking at http://ncas-cms.nerc.ac.uk/html_umdocs/UM55_User_Guide/node31.html and scroll down to item 4d (although those instructions are for 5.5 and this info is not included in the 4.5 user guide). It seems to suggest that there can be a problem with "inappropriate" snow from the ancil being converted to sea ice in the first time step, eg at places like Greenland. It suggests "not used" for the snow depth ancil!

What is the best solution to this, or is the below feasible?

One possibility is simply not to use the ice part of the slab model (I study the tropics, not the poles in any case)? I have a test-run xeafa which is updating to my snow ancil hourly, and has almost integrated for 1 month at time of writing. If this is an ok method then what is the cleanest way to turn off the ice part of the slab? Currently in umui for xeafa I have the settings in slab/science/slab ice as thermodynamics excluded and dynamics set to "thermodynamics only". Then I update the qrclim.ice ancil as for an atmosphere run, but use the heat convergence file for the sst. Does it matter that the heat convergence ancil made in the calibration was designed for the ice and sst, but I am now effectively overwriting the ice part?

comment:3 Changed 6 years ago by ros

  • Resolution set to completed
  • Status changed from new to closed

Conversation migrated to email/offline.

Note: See TracTickets for help on using tickets.