Real-time Verification of CPC Long-Lead Forecasts Using Heidke Skill 
				Scores from Station Data

                	    All Stations     Non-EC Only      % Coverage 

	Season      T		16.7		 38.2		 43.6 
	JJA 05      P		-4.7		-13.4		 35.0

	Month (LL)  T		 7.3		 38.6		 18.8 
	AUG 05      P		 0.0		-----		  0.0

	Updated Mo. T		 6.4		 35.7		 17.9
	AUG 05      P		-0.8		 -2.1		 40.2 

	Change      T		-0.9		 -2.9		 -0.9 
	(U-LL Mo.)  P		-0.8		 -2.1		+40.2


		Brief Discussion of Forecasts and Verification
		----------------------------------------------

	The temperature forecast for JJA 2005 had solidly positive Heidke skill
	scores of 16.7 for all stations and 38.2 for non-EC only, with coverage
	of 43.6 %.  The scores were both above the long-term averages of 8.7 for
	all stations and 18.5 for non-EC only, but the coverage was a shade less
	than the long-term average of 46.9% for the entire period of record.

	The JJA 05 seasonal forecast was based on a variety of statistical and
	dynamical models, and multi-model ensembles from IRI and CDC, as well as
	CPC's CCA and OCN statistical tools.  The NCEP CFS model was minimally
	considered, as it seems to have a cold and wet bias during the warm
	seasons.  Indications over the equatorial Pacific suggested that ENSO
	would be in a neutral phase during summer 2005.  The temperature forecast
	for the CONUS was primarily based on the traditional CPC statistical
	tools, all of which favored warmth in the interior Southwest, much of the
	Southeast, and along the Pacific Northwest Coast.  Most of the models at
	least hinted at some relative warmth in these areas as well.  Below
	normal temperatures were forecast over the upper Mississippi Valley on
	the basis of a consensus of most of the models with support from CCA. 
	Only the CDC statistical ensemble of six models gave a good forecast
	indication of the extensive area of warmth that was observed over much of
	the East and Midwest, although all the statistical tools correctly
	indicated substantial portions of the warmth that was observed over most
	of the Southeast, interior Southwest, and Pacific Northwest coast.  The
	unanimous indications for warmth in Florida from the statistical tools
	were unfortunately not heeded, and the state was left as EC due to an
	incorrect assumption that below normal SSTs would persist in the area and
	cancel out the indications for warmth.  The forecast of above normal
	temperatures in Alaska was very successful, scoring 87.0 for both all
	stations and non-EC only, with 100% coverage.  Contributing to its
	success was the fact that all the tools, both statistical and dynamical,
	correctly had warmth. 

	The precipitation forecast for the CONUS had a moderately negative skill
	overall.  The area of above median centered over the northern Great Basin
	and northern Great Plains had marginally positive skill,  with most of
	the skill coming from the eastern part due to frequent MCS activity.  The
	area of dryness predicted in the Southeast, based mainly on almost
	unanimous indications from the models, was all wrong for the second
	consecutive season, most of it with two-class errors.  The area of
	relative dryness predicted over Arizona and New Mexico had modest skill,
	correctly anticipating that an expected late monsoon would lead to a
	deficit in the seasonal total precipitation.  The all-station score was
	-4.7 and the non-EC only score was -13.4, with a moderate coverage of
	35.0%.  These forecasts were based on almost unanimous indications from
	the dynamical models that it would be wetter than normal over much of the
	north-central part of the country and parts of the northern Great Basin
	and northern Rockies, with some weak support in parts of the area from
	CCA and CAS.   The CFS model did correctly forecast much of the wetness
	in the West and the northern Great Plains, but would not have scored very
	well overall because it indicated wet conditions over nearly all the
	CONUS.   A rather strong signal for dryness over the Southeast from the
	CDC model ensemble was unfortunately followed, as there were no contrary
	indications and even the CAS statistical tool and the NCEP CFS model had
	indications for dry over most of Florida.   Alaska was left with all EC,
	as there were no consistently strong or reliable forecast indications
	from any of the tools for that area.
	
	The 0.5-month lead temperature forecast for August was fairly successful,
	with an all-station Heidke Skill score of 7.3 and a non-EC score of 38.6,
	with a modest coverage of 18.8%.  The areas predicted to have above
	normal temperatures in the far West, the interior Southwest, and the
	Florida Peninsula were nearly all correct, but the portion of the above
	normal predicted in the Great Basin was largely wrong.  Alaska again had
	a very good forecast of above normal for much of the state with an
	all-station score of 60.9 and a non-EC score of 100.0, with coverage of
	60.9%.  The areas forecast to be warm were based primarily on OCN with
	general agreement from CCA and CAS in some areas.   Pre-existing above
	normal SSTs around Alaska also contributed to the largely successful
	forecast for relative warmth there, along with generally strong agreement
	among the statistical tools.  

	Lack of any statistically significant tools led to a 0.5-month lead
	precipitation forecast for August that  showed EC everywhere, so the
	all-station skill was exactly zero (0.0) by definition, and of course the
	non-EC only score was indeterminate and the coverage was 0.0%.    Alaska
	was also left as all EC due to lack of any statistically significant
	signals from any of the tools.

	The revised updated temperature forecast for August issued at the end of
	July, which also incorporated input from several dynamical models used to
	make medium-range forecasts for the first 10 days of August, showed
	relatively small changes from the 0.5-month lead version, due to the good
	strength and consistency of the signals from the statistical tools
	compared with the somewhat lower anticipated skill of the dynamical
	forecasts.  (The model forecasts for 8-14 days were discounted due to
	their rather low skill and the fact that they showed a relatively large
	change in the forecast pattern).  The net result was a very small and
	probably insignificant decline in the scores to 6.4 for all stations and
	35.7 for non-EC only, along with a slight decrease in the coverage to
	17.9%.  The area of above normal temperatures forecast for Alaska was
	increased, resulting in an increase in the all-station score to 84.8 and
	the percent coverage to 91.3.  The non-EC score decreased slightly to
	92.9 (from a perfect 100) because one of the stations added verified as
	near normal.

	Consideration of the precipitation forecasts from the GFS models for the
	first ten days of August enabled the forecaster to add extensive forecast
	areas totaling up to 40.2% coverage, a rather large increase from the
	long-lead version that was left with all EC due to the lack of any
	significant statistical signals.   Unfortunately, Heidke skill scores
	were slightly negative, -0.8 for all stations and -2.1 fir non-EC only. 
	The area of above median predicted for most of the South and East from
	the central Gulf Coast and the Appalachians to the Atlantic Coast by
	itself actually had modest skill, with 13 or the 32 stations in the
	forecast area verifying correctly.  The problem was in the area of below
	median added over much of the West from the central and southern
	Intermountain Region to the Pacific Coast, which did very poorly, with
	only one correct station out of the 15 forecast.  Many of the stations
	that were wrong showed totals of only zero or one hundredth, but the
	colors indicated that they were in the near median or above median
	category.  The summer climate, especially in California, is so dry that
	most of the time no rain at all falls, and even a trace or a hundredth of
	an inch is considered to be near or above the median.  This can occur
	with drizzle in the marine layer at coastal stations, or a stray shower
	in the interior locations. Neither of these types of precipitation is
	well-determined from the large-scale circulation or handled well by the
	dynamical models.  The addition of an area of above median to much of
	Alaska resulted in an all-station score of 2.2 and a non-EC only score of
	3.6 for the state, with coverage of 60.9%.

	NOTE:  The total precipitation shown by the available data for New
	Orleans was in the below median category, but due to the devastation from
	Hurricane Katrina, no observations were recorded in the local station
	data for the last three days of the month.  It was judged to be almost
	certain that the actual total rainfall for the month was above the median
	due to Katrina.  The total rainfall for Birmingham, Alabama, was also
	suspiciously low, in the below median category.  However, inspection of
	the daily station data showed a very low amount from Katrina, and there
	were no missing observations.  Whether Birmingham just slipped between
	the rain bands, or whether there was an actual malfunction of the
	rain-measuring equipment, could not be definitively determined, so the
	reported monthly total was assumed to be correct despite it low value.
	
	--- EOF