Running cam5 with prescribed SST

[xianbiaokang@gmail_com]

I want to run cam5 with varying prescribed SST from Jan 1997 - Dec 1998 .

Following is my settings:

(1) Based on F_1850_CAM5 case, f19-19 grid

(2) download and interpolate the Renolds SST into the atm grid: avhrr.regrid.1981-2009.nc
the metadata of the regrided SST is:

netcdf avhrr.regrid.1981-2009 {
dimensions:
time = UNLIMITED ; // (10349 currently)
lat = 96 ;
lon = 144 ;
variables:
float ice_cov(time, lat, lon) ;
ice_cov:_FillValue = -999.f ;
ice_cov:long_name = "BCS Pseudo Sea-ice concentration" ;
ice_cov:units = "fraction" ;
ice_cov:missing_value = -999.f ;
float time(time) ;
time:units = "days since 1978-01-01 00:00:00" ;
time:long_name = "Center time of the day" ;
time:calendar = "365_day" ;
float zlev ;
zlev:actual_range = "0, 0" ;
zlev:units = "meters" ;
zlev:long_name = "Sea surface height" ;
double lat(lat) ;
lat:units = "degrees_north" ;
lat:long_name = "latitude" ;
double lon(lon) ;
lon:units = "degrees_east" ;
lon:long_name = "longitude" ;
float anom(time, lat, lon) ;
anom:_FillValue = -999.f ;
float err(time, lat, lon) ;
err:_FillValue = -999.f ;
float SST_cpl(time, lat, lon) ;
SST_cpl:_FillValue = -999.f ;
SST_cpl:units = "deg_C" ;
SST_cpl:missing_value = -999.f ;
SST_cpl:long_name = "BCS Pseudo SST" ;

(3) after creating the case, set RUN_STARTDATE = 1997-01-01, DOCN_SSTDATA_FILENAME = avhrr.regrid.1981-2009.nc, DOCN_SSTDATA_YEAR_START=1997, DOCN_SSTDATA_YEAR_END = 1998 in env_conf.xml :


(4) then configure and build the case successfully

(5) while running, the model crashed immediately with the following messages. I re-run the case with case provided SST: sst_HadOIBl_bc_1.9x2.5_clim_c061031.nc, the model can process smoothly, it seems that there is something wrong with the SST data I processed based on Renolds SST. I then compare the two SST and found that the main difference is that the renolds SST has missing value over land while sst_HadOIBl_bc_1.9x2.5_clim_c061031.nc not. I am not sure whether this is the cause of the problem. another, I am puzzling why we have SST over land, and how can we get that reasonably based on only the data over ocean.


error message:

QNEG3 from vertical diffusion/ncl_a1:m= 16 lat/lchnk=797 Min. mixing ratio violated at 1 points. Reset to 1.0E-36 Worst =-1.2E-10 at i,k= 12 30
QNEG3 from vertical diffusion/num_a3:m= 25 lat/lchnk=535 Min. mixing ratio violated at 1 points. Reset to 1.0E-36 Worst =-1.5E+06 at i,k= 8 30
_pmii_daemon(SIGCHLD): [NID 00416] [c2-1c0s0n2] [Wed Jun 22 13:50:41 2011] PE 178 exit signal Floating point exception
QNEG3 from vertical diffusion/num_a3:m= 25 lat/lchnk=797 Min. mixing ratio violated at 1 points. Reset to 1.0E-36 Worst =-2.2E+05 at i,k= 12 30
_pmii_daemon(SIGCHLD): [NID 01118] [c4-1c0s0n2] [Wed Jun 22 13:50:41 2011] PE 140 exit signal Floating point exception



Sincere thanks for any suggestions.

 

[eaton]

The error message is indicating that you are getting some bad surface values in the model. This could be due to the issue that you've noticed with missing data. The reason the SST dataset has values over land is to deal with problems where model grids don't match the real coastlines and models use ocean fraction and land fraction within a coastal gridcell. So your dataset should contain a smooth SST field. The datasets supplied with CESM have interpolated values over land. There is an algorithm for doing this. This paper gives details of how the datasets are prepared.

2008 Hurrell, J. J., J. J. Hack, D. Shea, J. M. Caron, and J.
Rosinski, A new sea surface temperature and sea ice boundary dataset
for the NCAR Community Atmospheric Model, Journal of Climate, 21, pp.
5145-5153.

 

[xianbiaokang@gmail_com]

that's really helpful, Thank you very much, eaton.