Date: Wed, 25 Oct 2000 17:30:51 -0500 From: "Mike Baldwin" X-Accept-Language: en MIME-Version: 1.0 To: "Lance F. Bosart" CC: David Schultz , map@atmos.albany.edu, map_disco@nssl.noaa.gov, ddowell@ou.edu, hblue@ou.edu, wicker@nssl.noaa.gov Subject: Re: [Map_disco] The Kicker and the Kickee Content-Transfer-Encoding: 7bit All- Should we be surprised to find out that output from a "mesoscale" NWP data assimilation/forecast system is causing problems when computing QG diagnostics? The Eta Model is continuously violating QG assumptions as it "dreams" up the forecast! Running at 22km, it contains plenty of friction and diabatic heating and ageostrophic advection, and should have sub-synoptic scale waves running all over the place. I'm reminded of the great pains that people have taken to filter out the "noise" while trying to keep the Rossby modes in NWP output, for example, Barnes et al 1996 BAMS, and an interesting FSL Forum paper by Caracena ( http://www.fsl.noaa.gov/~vondaust/f398/f398b.html ). Should one be using NWP output at all when doing QG diagnostics? I don't think so! Operational data assimilation procedures are intended to initialize a NWP forecast, not to produce the optimal fields for computing forcing terms in the QG omega equation. So, do your own analysis of the observed data, and take care to set up the objective analysis scheme to keep the synoptic-scale features and filter out the smaller scales. That way, you'll know exactly what went into those analyses, rather than relying on NCEP to do that work for you and wondering how the NWP model "dreamed" up those wacky initial conditions. Please listen to Dr. Bosart, and don't use the Eta output on high resolution grids when doing QG diagnostics. If you must use NWP output to do your diagnostics, NCEP produces Eta output on the old LFM grid, filtering the fields before interpolation to avoid aliasing as much as possible. Those grids are available via anonymous ftp at ftp.ncep.noaa.gov in /pub/data1/eta/erl.YYMMDD/, filenames containing *GRB5FM* and *GRB_FM* contain LFM grids. Or, if you really want to return to the good old days, get out your MM5 or download the workstation version of the Eta, configure it to run at 190.5 km, smooth the heck out of the topography, and you should be all set. Mike Baldwin