NOCN03 CWAO  121930
GENOT NO. 004

GENOT NO. 003 EST LA VERSION FRANCAISE DU GENOT NO. 004

IMPROVEMENT TO GEM REGIONAL SURFACE TEMPERATURE FORECASTS 

THE CANADIAN METEOROLOGICAL CENTRE WILL IMPLEMENT A MINOR CHANGE TO 
THE REGIONAL MODEL FORECAST SYSTEM ON JANUARY 13TH 2005 FOR THE
12Z RUN, WITH A GOAL OF IMPROVING NOCTURNAL SURFACE 
TEMPERATURE FORECASTS.  

MOTIVATION
IT HAS BEEN NOTED THAT THE OPERATIONAL GEM REGIONAL MODEL FORECASTS 
EXCESSIVELY COLD NOCTURNAL SURFACE TEMPERATURES, ESPECIALLY DURING 
WINTER, UNDER CLEAR SKIES AND IN WEAK LOW-LEVEL CIRCULATION REGIMES 
(TYPICALLY IN ANTICYCLONES).   THIS PROBLEM HAS BEEN NOTED MORE 
SYSTEMATICALLY IN MOUNTAINOUS REGIONS AS WELL AS IN THE ARCTIC.  THE 
CHANGE AIMS TO CORRECT THIS PROBLEM.  

THE SURFACE TEMPERATURE CALCULATION IS THE RESULT OF A DELICATE 
BALANCE BETWEEN SEVERAL COMPLEX PHYSICAL PARAMETERIZATIONS.  THE 
POSSIBLE CAUSES CONTRIBUTING TO THE PROBLEM INCLUDE A SLIGHT DEFICIT 
IN THE SURFACE INFRARED RADIATIVE BALANCE, UNDERESTIMATION OF WINDS 
NEAR THE SURFACE (PARTICULARLY IN MOUNTAINOUS TERRAIN), AS WELL AS THE 
TREATMENT OF ENERGY STORED IN THE SOIL, PARTICULARLY IN THE PRESENCE 
OF SNOW COVER.  RESEARCH AND DEVELOPMENT PROJECTS ON THIS SUBJECT 
ARE CONTINUING IN THE METEOROLOGICAL RESEARCH BRANCH, IN COLLABORATION 
WITH THE DEVELOPMENT BRANCH OF THE CANADIAN METEOROLOGICAL CENTRE.

THE APPEARANCE OF THESE EXCESSIVELY COLD TEMPERATURES IS ASSOCIATED 
WITH THE FORMATION OF VERY PRONOUNCED BUT VERY SHALLOW TEMPERATURE 
INVERSIONS NEAR THE MODEL SURFACE.  AS SUCH, IN THE MAJORITY OF CASES 
WHERE THESE EXAGGERATEDLY COLD TEMPERATURES ARE FORECAST, THERE 
EXISTS AN UNREALISTIC TEMPERATURE DIFFERENCE BETWEEN THE SURFACE AND 
THE MODEL LEVEL JUST ABOVE IT AT 40 M (ATTAINING AS MUCH AS 25 DEGREES 
AT TIMES).

DESCRIPTION OF THE CHANGE
THE CHANGE CONSISTS OF MODIFYING THE DIAGNOSTIC TEMPERATURE CALCULATION 
AT THE STEVENSON SCREEN HEIGHT, LIMITING THE VERTICAL GRADIENT TO 8 
DEGREES BETWEEN THIS LEVEL AND THE NEXT MODEL LEVEL, SITUATED AT A 
HEIGHT OF ABOUT 40 M.  HENCE THIS CHANGE ONLY COMES INTO EFFECT IN THE
CALCULATION OF THIS TEMPERATURE UNDER THE PRECISE SCENARIO WHEREIN THE
COOLING IS JUDGED TO BE EXCESSIVE.

EVALUATION
THIS MODIFICATION HAS BEEN TESTED BY THE RESEARCH AND DEVELOPMENT  
TEAMS ON SEVERAL CASES, AND A REGIONAL FORECAST SYSTEM EQUIPPED WITH  
THIS CHANGE WAS THEN INSTALLED IN PARALLEL ON NOVEMBER 11, 2004.  

THE OBJECTIVE EVALUATION OF SURFACE TEMPERATURES SHOWS ONLY A SLIGHT 
IMPROVEMENT WHEN SCORED AGAINST THE FULL NORTH AMERICAN SURFACE DATA 
SET.  THIS WAS EXPECTED, GIVEN THE LOCALIZED AND OCCASIONAL NATURE OF 
THE CHANGE.  THE CHANGE HAS LITTLE SIGNIFICANT IMPACT ON OTHER MODEL 
FIELDS, SUCH AS MASS FIELDS, CLOUD AND PRECIPITATION.

THE EVALUATION OF THE OPERATIONAL TEAM INDICATES THAT FORECAST 
TEMPERATURES ARE CLOSER TO THE OBSERVATIONS IN THE PARALLEL SYSTEM IN 
80 PERCENT OF CASES WHERE THE DIFFERENCE BETWEEN THE OPERATIONAL AND 
PARALLEL FORECASTS IS GREATER THAN 3 DEGREES.  THE NUMBER OF SURFACE 
DATA REJECTS WAS DIMINISHED IN THE PARALLEL 3DVAR ASSIMILATION CYCLE.  

A WEAK DRY BIAS IN SURFACE RELATIVE HUMIDITY WAS DETECTED IN AREAS 
WHERE THE CORRECTION WAS ACTIVATED.  THIS IS JUDGED TO BE A SECOND 
ORDER PROBLEM IN COMPARISON WITH THE BENEFITS OF BETTER TEMPERATURE
FORECASTS, IN LIEU OF A MORE COMPREHENSIVE SOLUTION TO COME.

THE UMOS STATISTICAL FORECAST SYSTEM WAS SET UP IN PARALLEL, SO THAT 
THIS SYSTEM COULD TAKE THE CORRECTION INTO ACCOUNT.  A COMPARISON OF  
UMOS OPERATIONAL AND PARALLEL TEMPERATURE FORECASTS REVEALED THAT BOTH 
SYSTEMS PERFORMED TO AN EQUIVALENT LEVEL FOR THE MONTH OF DECEMBER  
2004.  IT IS EXPECTED THAT THE UMOS SYSTEM WILL IMPROVE GRADUALLY OVER 
TIME.  WE ALSO NOTE A BETTER DIAGNOSTIC OF THE MEAN SEA LEVEL
PRESSURE, PARTICULARLY OVER MOUNTAINOUS TERRAIN.

EVERELL / ADMA / SMA - SMC TORONTO