reconfigured ECMWF global dump

[hanneke]

Hello,

For one of my runs I am using a reconfigured ECMWF global start dump that Chang created for me. I noticed some differences compared to the UKMO start dump. For example, the boundary layer depth is constant globally (500m). After running reconfiguration and getting a start dump for my LAM, this is still the case. Isn't this a problem for the model run?

The reason I am looking into this is that my model run completes, but gives out mostly NaN's and shows this in the leave file:

==============================================

initial Absolute Norm : 281436461.073034465

GCR( 2 ) failed to converge in 100 iterations.

Final Absolute Norm : 725.022702810517217

==============================================

Atm_Step: Timestep 2

RHS zero so GCR( 2 ) not needed

============================================== (etc)

So my guess was that something is wrong with the start dump. I was wondering if things like the constant boundary layer depth could be causing this problem?

The start dump can be found here: /hpcx/home/n02/n02/hanneke/work/globaldumps/feb02/02022100.astart
The run I am working on is xcncc.

Cheers,
Hanneke