NOTES from Global Climate and Weather Modeling Branch Map Briefing
----------------
  3/17/05:

        Glenn White - New GFS, a.k.a. parallel-Y- 
	          see: http://wwwt.emc.ncep.noaa.gov/gmb/noor/oct98op/textwea.html 
                  click on "PRY experiment with T382 system Dec4-Jan21"

		1. Mark Iredell's recent email listed fixes to PRY problems:
			a. PROBLEM 1 - "analysis not converging", FIXES:
			     -use current operational background error,
			     -tune low-humidity criterion in Convergence algorithm,
			     -increase iterations from 100 to 150.
			b. PROBLEM 2 - "T190 instability", FIXES:
			     -increase horizontal/vertical diffusion above 10mb,
			     -invoke wind speed dependent diffusion at 10% lower speed. 
		2. Compared day1/day5; 500mb height AC+RMS+wavenumber 'die-off' curves 
		      and found:
			-NH significantly better in para-Y-
			-SH para-Y- is somewhat worse, especially in waves 1-3
		3. N.America precipitation EQTS, BIAS (12-36),(60-84) much better in para-Y-
		4. Tropical winds: para-Y- bit better at 850mb, perhaps a bit worse at 200mb.
		5. Suru's 'Fit to OBS' similar, plus improved stratosphere in the para-Y-, plus 
				better (1000-700)mb wind fits to ACARS data (N America).
				better (300-150)mb wind fits to aircraft (Global).
		6. Conclusions: NH days1-6: paraY is better, but worse after day 7!
		                SH/Tropics: neutral results!

        Yuejian Zhu - Moisture verif with UK (1,2 below) and Analysis 
			comparisons (3-5 below) ... SEE 
			http://www.emc.ncep.noaa.gov/gmb/yzhu/html/qvrfy.html:

		1. Reviewed GFS presentation from 3 Feb 05 - correlation, bias, 
			 and RMS error, using GDAS as 'truth'.
		2. Compares UK and GDAS (only up to 500mb), correlation between analyses 
			is lowest in lower atmosphere and higher at 500mb; UK drier 
			below 850mb and more moist above (especially the tropics).
		3. Compares 500mb height and 850/200mb winds (AC/RMSE) from 
			UK, ECMWF, Canada, and NAVY for the past 45 days, WHERE 
			all are referenced to NCEP Final analysis.
		4. Suggestion that GFS analysis be used, as that is the one 
			driving the forecast model.
		5. NCEP analysis appears to be closest to the ECMWF analysis.
 
	Mike Baker - STATS

                1. Reviewed Pete Caplan's web page for operational model 'stats':
                www.emc.ncep.noaa.gov/gmb/STATS/STATS.html:
                        a. GFS has improved over the past month, but in terms of
                           500 mb AC, ECMWF is a day better in both hemispheres 
			   at day 7.
                        b. Comparing NH 500mb AC 'die off' curves for the past
                           month, all 4 GFS cycles are similar in SH, but in NH at 
			   days 8-9, 18Z seems best.  However yearly data does not 
			   show any cycle being the best.