[Met_help] grid_stat different means of obs field for exact same obs field, but different model output comparisons

[Shaw, Michael J CTR USAF AFWA 16 WS/WXE]

Hi.

I'm grid_stat'ing precip, with a fcst_thresh of ge0 and am masking against stage IV (over most of conus) file that's projected to the model grid and for all stage IV values ge10 (it's a cumulative file over about a week).  When I compare model output A to stage IV, it uses ~2500 points with one OBS mean, and when comparing model output B to stage IV, it uses ~2600 points and finds another OBS mean.  I would have thought that this mask and fcst_thresh would use the same number of points and would compute the same OBS mean.  I.e., if there's a model value ge0 to compare with obs ge10 mask, the pair would be counted, and there'd be same # pairs for both case A and case B.  I would think this would be the same for either case as I believe the model and obs files are all ge0 for this mask (ge10 obs).  I'm using the Monte Carlo error bounds with random selection of samples which I thought might have been the issue (random sample selection resulting in slightly different means and error bars), but it actually results in the exact same values of mean of A etc for all reruns of model comparison A, and the exact same values for all values of mean of B etc for all reruns of model  comparison B.

Any thoughts?

Thanks,
Mike

______________________________________________

Michael Shaw, Contractor
SAIC/NASA Support Scientist
michael.shaw.ctr at offutt.af.mil
16WS Environmental Characterization
HQ AFWA-SAIC
101 Nelson Drive
Offutt AFB, NE 68113-1023
402-232-7690 Comm * 402-272-7690 DSN * 402-294-8230 Fax

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/met_help/attachments/20100429/3458bf0c/attachment.html