[MITgcm-devel] woes with exf interpolating

[Matthew Mazloff]

Hi JMC,

I copied it below -- I think its pretty standard.  Is there something  
I'm missing in prescribing the forcing dimensions -- do I need to  
manually provide an overlap, or provide overlap information?

Thanks
Matt



login4% more data.exf
# $Header: /u/gcmpack/MITgcm/verification/lab_sea/input/data.exf,v 1.7  
2004/07/11 06:21:44 dimitri Exp
  $
#
# *********************
# External Forcing Data
# *********************
  &EXF_NML_01
#
  useExfYearlyFields = .TRUE.,
  useExfCheckRange = .FALSE.,
  exf_iprec         = 32,
  exf_yftype        = 'RL',
  useRelativeWind   = .TRUE.,
  /
#
  &EXF_NML_02
#
  atempstartdate1   = 20050101,
  atempstartdate2   = 00000,
  atempperiod       = 21600.0,
#
  aqhstartdate1     = 20050101,
  aqhstartdate2     = 00000,
  aqhperiod         = 21600.0,
#
  precipstartdate1  = 20050101,
  precipstartdate2  = 00000,
  precipperiod      = 21600.0,
#
  uwindstartdate1   = 20050101,
  uwindstartdate2   = 00000,
  uwindperiod       = 21600.0,
#
  vwindstartdate1   = 20050101,
  vwindstartdate2   = 00000,
  vwindperiod       = 21600.0,
#
  runoffstartdate1  = 20050101,
  runoffstartdate2  = 00000,
  runoffperiod      = 0.0,
#
  swdownstartdate1  = 20050101,
  swdownstartdate2  = 00000,
  swdownperiod      = 21600.0,
#
  lwdownstartdate1  = 20050101,
  lwdownstartdate2  = 00000,
  lwdownperiod      = 21600.0,
#
  atempfile         = 'SO6input/NCEP_R6_tmp2m_degC_SO',
  aqhfile           = 'SO6input/NCEP_R6_spfh2m_SO',
  uwindfile         = 'SO6input/NCEP_R6_u10m_SO',
  vwindfile         = 'SO6input/NCEP_R6_v10m_SO',
  precipfile        = 'SO6input/NCEP_R6_rain_SO_DT',
  runoffFile        = 'SO6input/run-off_SO.6',
  lwdownfile        = 'SO6input/NCEP_R6_dlw_SO',
  swdownfile        = 'SO6input/NCEP_R6_dsw_SO',
#
  /
#
  &EXF_NML_03
#
  exf_inscal_uwind = 1.,
  exf_inscal_vwind = 1.,
  exf_inscal_runoff = 3.170979198E-8,
  exf_offset_atemp = 273.15,
#
  /
#
  &EXF_NML_04
#
  atemp_lon0       = 0.5D0,
  atemp_lon_inc     = 1.D0,
  atemp_lat0        = -79.5D0,
  atemp_lat_inc     = 60*1.D0,
  atemp_nlon        = 360,
  atemp_nlat        = 60,
#
  aqh_lon0          = 0.5D0,
  aqh_lon_inc       = 1.D0,
  aqh_lat0          = -79.5D0,
  aqh_lat_inc       = 60*1.D0,
  aqh_nlon          = 360,
  aqh_nlat          = 60,
#
  precip_lon0       = 0.5D0,
  precip_lon_inc    = 1.D0,
  precip_lat0       = -79.5D0,
  precip_lat_inc    = 60*1.D0,
  precip_nlon       = 360,
  precip_nlat       = 60,
#
  uwind_lon0         = 0.5D0,
  uwind_lon_inc      = 1.D0,
  uwind_lat0         = -79.5D0,
  uwind_lat_inc      = 60*1.D0,
  uwind_nlon         = 360,
  uwind_nlat         = 60,
#
  vwind_lon0         = 0.5D0,
  vwind_lon_inc      = 1.D0,
  vwind_lat0         = -79.5D0,
  vwind_lat_inc      = 60*1.D0,
  vwind_nlon         = 360,
  vwind_nlat         = 60,
#
  swdown_lon0          = 0.5D0,
  swdown_lon_inc       = 1.D0,
  swdown_lat0          = -79.5D0,
  swdown_lat_inc       = 60*1.D0,
  swdown_nlon          = 360,
  swdown_nlat          = 60,
#
  lwdown_lon0          = 0.5D0,
  lwdown_lon_inc       = 1.D0,
  lwdown_lat0          = -79.5D0,
  lwdown_lat_inc       = 60*1.D0,
  lwdown_nlon          = 360,
  lwdown_nlat          = 60,
#
  /
#
  &EXF_NML_OBCS
#
  obcsNstartdate1     = 20050101,
  obcsNstartdate2     = 00000,
  obcsNperiod         = 2635200.0,
#
  /



On Apr 3, 2010, at 7:07 AM, Jean-Michel Campin wrote:

> Hi Matt,
>
> Could you send also the "data.exf" you are using ? or could I find it
> somewhere ?
> Thanks,
> Jean-Michel
>
> On Fri, Apr 02, 2010 at 03:17:51PM -0700, Matthew Mazloff wrote:
>> Hello,
>>
>> A problem has been found (thanks Fabien!) regarding a discontinuity  
>> in
>> the wind fields around longitude 0/360.
>>
>> My set-up is 1/6 of a degree and I use NCEP on a 1 degree grid and  
>> use
>> exf interpolation.  Within ~1 degree east and west of longitude 0  
>> there
>> is a discontinuity in the wind field -- it is small but quite
>> significant in the x-derivative.  My guess is that the interpolation
>> routine is not properly accounting for periodicity -- and instead
>> perhaps incorporating zeros.
>>
>> Some important stuff -- I am using NCEP winds -- not stress -- and I
>> have switched to interp_method=2 instead of the hard-coded
>> interp_method=1.  This is because with interp_method 1 (linear) there
>> are discontinuities in the wind-stress curl everywhere.  With bicubic
>> interp the curl is smooth -- with this exception around longitude 0
>>
>> I was hoping someone with more expertise could shed light on this  
>> before
>> I go digging in the code
>>
>> Thanks!
>> -Matt
>>
>>
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel