Port validation of 1.9x2.5_gx1v6 resolution, B_1850_CN run: b40.1850.track1.2deg.003

[cnrwil004@myuct_ac_za]

HiI am a South African postgraduate student new to running operational models. We have ported cesm1.2.0 onto a local linux cluster and can produce output. However, a number of problems have arisen in trying to validate the porting. Using a start-up run, two apparently equivalent cases, run on a different number of cores, appear to produce different results. Additionally running a 20-yr 1.9x2.5_gx1v6 resolution, B_1850_CN Pre-Industrial Control simulation (as suggested in the cesm1.2 userguide and using the b40.1850.track1.2deg.003 restart files as is done automatically when choosing the appropriate values for the input parameters when running ./create_newcase), produces a climatology which, as assessed using Ferret and cdo plotting and analysis tools, gives appears to be very different to that of produced on the NCAR bluefire machine. The mean clm surface temperature (TS) is 281.9K, rather than the suggested 280.659K. Interannual variability also does not appear to match up. Any suggestion on whether this an acceptable level of discrepancy and if not how I might go about addressing it, will be greatly appreciated.I attach a file showing output annual mean temperature for clm (TSA_b40_chpc.gif) and the file from bluefire to which I compared it (b40_TSA.gif), as well as a comparison of the mean tempearture climatology (this is a comparison of 20 and 30 yr files respectively. NCAR bluefire images are from http://www.cesm.ucar.edu/experiments/cesm1.0/diagnostics/b40.1850.track1.2deg.003/atm_501-530-obs/ and links from this page).

Thanks
Stefaan