Question about ECMWF data

[ymamoutos]

Greetings,

I have been running simulations for the north Aegean region over a month now,
and I have used two different atmospheric datasets. The configuration for all
simulations was the same and the only difference was the atmospheric forcing files.

I used the BULK_FLUXES formula, clamped obc for my 3D fields (u,v,T,S), Chapman explicit
for free surface and Shchepetkin for ubar and vbar.Only the south boundary is open.
For the first run I used data from the NCEP reanalysis dataset (NCEP Climate Forecast
System Reanalysis (CFSR) http://rda.ucar.edu/datasets/ds093.1/) and for second
was from ECMWF-Interim (http://apps.ecmwf.int/datasets/data/interim_full_daily/).

The data from NCEP had a 6 hour timestep and ~0.3x0.3 spatial resolution and for the data
from ECMWF I used the matlab script that is available through SVN repository (3 hour timestep
and 0.25x0.25 spatial resolution). Comparing the results there is difference in temperature
during the whole year and especially during the summer months. Also the results from the first
simulation (using NCEP data) it seems to be more accurate according to the available bibliography.

I also compared the values for shflux from the ROMS average file for each run because the values are
at the same time (every day at 12:00) and there is a difference between two sets. The average value for
the whole region from the first run (NCEP) is approximately -98.89 W/m^2 and from the second (ECMWF)
is -12.81 W/m^2. I am trying to understand if I had done something wrong with forcing but to be honest
I am a bit confused. The data for boundary conditions files are from MFS system reanalysis with a daily
time step (http://gnoo.bo.ingv.it/mfs/dailysys_des.htm/) and the climatology from MEDATLAS dataset.

I am attaching two figures from the same day, and my *.h file. Any help or suggestion will
be highly appreciated.

Giannis