NOCN02 CWAO 061830
GENOT FAX NO. 007

GENOT NO. 007 EST LA VERSION ANGLAISE DE GENOT NO. 008

MODIFICATION TO THE REGIONAL FORECAST SYSTEM AT CMC.

A MODIFICATION TO THE REGIONAL FORECAST SYSTEM WILL BE IMPLEMENTED 
AT THE CANADIAN METEOROLOGICAL CENTRE TUESDAY, JUNE 10 1997, AT 
1200 UTC. THE MODIFICATION INVOLVES A CHANGE IN THE NATURE OF THE 
VERTICAL COORDINATE OF THE ANALYSIS FEEDING THE GEM REGIONAL MODEL.  
THIS NEW ANALYSIS WILL RESULT FROM A DIRECT INTERPOLATION FROM THE 
21 SIGMA LEVELS OF THE GLOBAL ANALYSIS ON THE ETA LEVELS OF THE GEM 
MODEL, WHEREAS THE CURRENT OPERATIONAL ANALYSIS FEEDING THE MODEL 
IS DONE ON THE 16 STANDARD PRESSURE LEVELS.

THE ANALYSIS IN THE REGIONAL SYSTEM WILL CONTINUE TO BE PRODUCED 
FROM THE SAME GLOBAL ASSIMILATION CYCLE (USING THE GLOBAL SEF MODEL 
TO GENERATE THE TRIAL FIELDS). THE MODIFICATION WILL AFFECT THE 
INTERFACE BETWEEN THE GLOBAL ASSIMILATION CYCLE AND THE REGIONAL 
FORECAST MODEL.

THE REGIONAL MODEL WILL THEREFORE REMAIN UNCHANGED, EXCEPT FOR THE 
MODEL ENTRY. THIS NEW VERSION OF THE ENTRY ALLOWS THE INTERPOLATION 
FROM THE 21 SIGMA LEVELS OF THE GLOBAL ANALYSIS TO THE 28 ETA 
LEVELS OF THE GEM MODEL, WHILE TAKING INTO ACCOUNT DIFFERENCES 
BETWEEN THE TOPOGRAPHY FIELDS REPRESENTED IN BOTH. IT IS THEREFORE 
A MODIFICATION THAT AFFECTS ONLY THE PREPARATION OF THE INITIAL 
FIELDS PROVIDED TO THE MODEL.

THE REGIONAL MODEL THEREFORE BENEFITS FROM AN ANALYSIS HAVING AN 
INCREASED VERTICAL RESOLUTION IN THE LOWER TROPOSPHERE, WITH 8 
LEVELS BELOW 3 KM AGAINST 3 LEVELS AS IT IS CURRENTLY THE CASE FOR 
THE ANALYSIS IN ISOBARIC COORDINATES.

THIS IMPLEMENTATION REPRESENTS ONE STEP TOWARDS IMPROVING THE 
INITIAL CONDITIONS PROVIDED TO THE REGIONAL MODEL. THIS IS NO 
SUBSTITUTE TO THE FORESEEN IMPLEMENTATION OF A REGIONAL 
ASSIMILATION SYSTEM WHERE THE REGIONAL MODEL WILL BE USED TO 
PRODUCE THE TRIAL FIELDS.

THIS NEW SYSTEM HAS BEEN RUN IN PARALLEL AT CMC OVER THE LAST 
WEEKS, AND HAS BEEN ASSESSED BOTH OBJECTIVELY AND SUBJECTIVELY BY 
CMC METEOROLOGISTS.

PERFORMANCE EVALUATION:

1. INITIAL CONDITIONS:  0 HOUR OUTPUTS

SIGNIFICANT IMPROVEMENTS TO THE INITIAL TEMPERATURE AND MOISTURE 
FIELDS HAVE BEEN NOTED WHEN THEY WERE COMPARED TO UA SOUNDINGS. IT 
IS OBVIOUS THAT THE PARALLEL MODEL CARRIES A BETTER REPRESENTATION 
OF THE INITIAL STATE OF THE BOUNDARY LAYER. VERTICAL GRADIENTS OF 
MOISTURE AND TEMPERATURE ARE MORE REALISTIC WHEN COMPARED TO UA 
OBSERVATIONS, MOST ESPECIALLY IN THE LOW LEVELS AND IN VICINITY OF 
FRONTAL ZONES.

WITH THE NEW VERSION, THE AMPLITUDE OF THE VERTICAL MOTION FIELD 
NEAR THE SURFACE (AT 0 HOUR) IS ALSO SIGNIFICANTLY REDUCED. THIS 
AMPLITUDE IS TOO LARGE IN THE CURRENT OPERATIONAL SYSTEM, 
ESPECIALLY OVER OCEANS.  THIS IMPROVEMENT IS CHARACTERISTIC OF A 
BETTER INITIAL DYNAMICAL BALANCE FOR THE MODEL.

GIVEN THE ABSENCE OF A REGIONAL ASSIMILATION SYSTEM AND THE 
CURRENTLY USED INITIALIZATION METHOD, MOST STATEMENTS PUT FORWARD 
IN THE GENOT NOCN02 CWAO ISSUED ON FEBRUARY 14 1997 AT 1930 RELATED 
TO THE IMPLEMENTATION OF THE GEM MODEL IN REGIONAL MODE ARE STILL 
VALID.

2. EVALUATION OF FORECASTS:

THE NEW SYSTEM HAS BEEN ASSESSED DAILY BOTH OBJECTIVELY AND 
SUBJECTIVELY FOR THE MONTH OF MAY. DURING THIS PERIOD, THE 
IMPROVEMENT OF THE INITIAL CONDITIONS PROVIDED TO THE MODEL DID NOT 
GENERALLY PRODUCE MARKED CHANGES ON THE MODEL FORECASTS.

HOWEVER, THE OBJECTIVE EVALUATION OF FORECASTS PRODUCED BY THE NEW 
SYSTEM AGAINST OBSERVATIONS SHOWS A SLIGHT IMPROVEMENT OF THE LOW 
LEVEL MOISTURE FORECASTS OUT TO 48 HOURS, AS WELL AS A DECREASE OF 
THE LOW BIAS OF THE MODEL AT THE SURFACE. QUITE OFTEN DIFFERENCES 
OF 1 OR 2 HPA IN THE DEPTH OF LOWS AND IN FAVOR OF THE NEW SYSTEM 
HAVE BEEN OBSERVED.  IN SOME CASES, ESPECIALLY OVER THE EAST COAST, 
THE DIFFERENCES BETWEEN THE NEW SYSTEM AND THE CURRENT OPERATIONAL 
ONE REACHED 4 HPA FAVORING THE NEW SYSTEM.

THE FORECAST PRECIPITATION FIELD HAS SHOWN MODERATE SENSITIVITY TO 
THIS CHANGE. MOST NOTICEABLE DIFFERENCES BETWEEN THE OPERATIONAL 
AND PARALLEL MODELS FORECASTS HAVE BEEN OBSERVED WHEN THE 
CONVECTIVE COMPONENT WAS SIGNIFICANT. OVERALL, ONLY SLIGHT 
IMPROVEMENTS TO THIS FIELD ARE TO BE EXPECTED. 

THE NEW SCHEME HAS BEEN RUN ON A FEW HISTORICAL WEATHER EVENTS. ON 
SOME OF THESE CASES, IT PERFORMED SIGNIFICANTLY BETTER.  IT IS 
LIKELY THAT THESE IMPROVEMENTS TO THE REGIONAL MODEL STARTING STATE 
LEAD TO SIGNIFICANT IMPROVEMENTS TO THE FORECASTS IN CASES WHERE 
MODELS ARE MORE SENSITIVE TO INITIAL CONDITIONS.

MCBEAN/ADMA/TORONTO