Opened 7 months ago

Last modified 6 months ago

#2053 new help

Simplest way to run UKCA nudged to 2016 meteorology?

Reported by: earhg Owned by: um_support
Priority: normal Component: UM Model
Keywords: Cc:
Platform: MONSooN UM Version: 8.4

Description

hi,
I would like to run UKCA in nudged mode for a few months of 2016, ideally June to August. I am happy if the ancillaries from a previous year are used.

I set up run xmmlb, and was able to reconfigure and change the date on the start dump I had for a previous run from year 2000 but then the model run failed (leave file is at /home/hamgo/output/xmmlb003.xmmlb.d17012.t200023.leave on MONSOON) because the ancillaries don't contain data for 2016. The first failure is for sea ice, I expect it would fail on others too.

Is there a straightforward way to force the model to use ancillaries for a different year? I don't mind using sea ice (or other) ancillaries from 2000, I just need the forcing meteorology to be from 2016, and I checked that the era-interim data is present on MONSOON for this (until end September). Or does someone already have ancillary files for 2016 i could use?

Many thanks Hamish

Change History (6)

comment:1 Changed 7 months ago by grenville

Hamish

It may be simplest to change the dates is the ancillary files (I haven't counted how many there are which will fail for a 2016 run) - there are utilities in MULE which help you to do that, see https://code.metoffice.gov.uk/doc/um/mule/vn1.3/user_guide_basics.html#header-components for example. You'll need UMDP F03 handy while trying this.

Grenville

comment:2 Changed 7 months ago by earhg

hi Grenville,
Many thanks for the help. I don't know how to use MULE but I was able to change the sea ice and SST ancillaries using xconv and xancil, and the model seemed then to get past the step where the ancillaries are updated, so I assume (or naively hope) the rest are ok. However, the run xmmlb then crashes on the first or second timestep with a seg fault. I tried starting in February rather than June, and tried halving the timestep, but neither of these seemed to work. A sample .leave file is /home/hamgo/output/xmmlb003.xmmlb.d17017.t114430.leave on MONSOON.
Please can you suggest other things I could try?
Many thanks, best
Hamish

comment:3 Changed 7 months ago by earhg

…for example should I request a start dump or restart file to be generated for some month in 2016, so that I can run without a discontinuity in the meteorological forcing between the dump file and the model run (which sounds like the most likely cause for the crash)?

comment:4 Changed 7 months ago by earhg

hi,
Does anyone have any recommendations for this? I should add that I don't have to use UKCA 8.4, any version will do - I just happened to have 8.4 runs to hand as a starting point.

many thanks Hamish

comment:5 Changed 7 months ago by willie

Hi Hamish,

You could try modifying the ancillary reference time in the UMUI. It's at Ancillary and Inpur > In file related options > Ancillary Reference Time. I'm not sure whether you should set it to zero or to the date you want.

Regards
Willie

comment:6 Changed 6 months ago by grenville

Hamish

Is this still a problem?

Grenville

Note: See TracTickets for help on using tickets.