[MITgcm-support] using low resolution spin up as input to a high resolution model

[Dimitris Menemenlis]

... or, given that you are changing resolution and that the tendency terms are likely to be either irrelevant
(and even potentially problematic), you could just initialize model with:

input file "data"

# Input datasets
 &PARM05
 bathyFile      = ... ,
 hydrogThetaFile= ...,'
 hydrogSaltFile = ...,
 uVelInitFile   = ...,
 vVelInitFile   = ...,
 pSurfInitFile  = ...,
 &

input file "data.seaice"

# SEAICE parameters
 &SEAICE_PARM01
...
      AreaFile           = '...,
      HsnowFile          = '...,
      HsaltFile          = ...,
      HeffFile           = ...,
      UiceFile           = ...,
      ViceFile           = ...,
 &

On Jun 13, 2014, at 6:23 AM, Martin Losch <Martin.Losch at awi.de> wrote:

> Hi Jonny,
> 
> if you can live with pickup files that are NOT in netcdf format (mnc_read_pickup = .false.) then you can simply provide one global pickup file in MDS-format (the *.data *.meta pairs). A file pickup.??????????.data will be read by any model decomposition. Personally I prefer this method over the netcdf pickups, even when I use netcdf for other output. With “globalFiles = .TRUE.”, or useSingleCPUio = .TRUE., (and mnc_write_pickup=.false.) you’ll even get “global” pickup files directly from the model.
> 
> Martin
> 
> On Jun 13, 2014, at 1:13 PM, Jonny Williams <Jonny.Williams at bristol.ac.uk> wrote:
> 
>> Hi everyone
>> 
>> I am currently trying to work out how I can use a 1000 year long spinup of a 1 degree resolution model as the input to a 0.1 degree equivalent. 
>> 
>> Both versions have 48 processors (although clearly I could vary this).
>> 
>> I know how to smooth, regrid and impose the new (high resolution) land/sea mask on to my pickup files already.
>> 
>> The pickup files are of the form...
>> 
>> pickup.0000324000.t040.nc
>> 
>> The problem is that processor number 40 (as indicated at the end of the above file name) of the 1 degree model does not spatially correspond to processor number 40 in the 0.1 degree model due to the different processor decomposition.
>> 
>> One way round this would be to 'glue' the pickup files together into a global field (using the gluemncbig script) BUT then I will surely have to 'unglue' the resulting global field together in order to use it as a restart for the 0.1 model!
>> 
>> Does anyone have experience of this kind of issue? Presumably I am not the first to want to do this kind of thing?...
>> 
>> Many thanks for your time!!
>> 
>> Jonny