NOCN02 CWAO 161930 GENOT FAX NO. 001 NO. 001 EST LA VERSION FRANCAISE DE GENOT FAX NO.002 MODIFICATIONS TO THE SCHEME "FORCE-RESTORE" WILL BE IMPLEMENTED IN THE CODE OF THE REGIONAL GEM MODEL ON WEDNESDAY, JANUARY 21 1998 AT 1200 UTC. THIS SCHEME ALLOWS TO CALCULATE IN THE MODEL SURFACE TEMPERATURE AND HUMIDITY AND IS USED AS BOUNDARY CONDITIONS IN THE SOLVING OF THE VERTICAL DIFFUSION EQUATION. THE MODIFICATIONS WILL INFLUENCE THE DIRECT MODEL OUTPUT OF SURFACE TEMPERATURES AND DEW POINTS DURING THE WINTER SEASON, THUS ALLOWING TO SIGNIFICANTLY IMPROVE FORECASTS OF THESE VARIABLES. THREE MODIFICATIONS ARE INVOLVED: 1. THE THERMAL CHARACTERISTICS OF SNOW HAVE BEEN MODIFIED TO RENDER IT MORE INSULATING. 2. THE TREATMENT APPLIED TO ICE COVERED WATER SURFACES HAS BEEN MODIFIED; FROM NOW ON, IF AN ICE SURFACE IS COVERED BY AT LEAST 50 CM OF SNOW, IT WILL BE TREATED AS A SNOW SURFACE AND NOT AN ICE SURFACE; THE THRESHOLD PREVIOUSLY USED WAS 10 CM. 3. THE CALCULATION OF THE INFRARED FLUXES ORIGINATING FROM THE EARTH'S SURFACE HAS BEEN IMPROVED. THE FIRST MODIFICATION IS THE MOST IMPORTANT; RENDERING SNOW MORE INSULATING ALLOWS LARGER VARIATIONS OF SOIL TEMPERATURE IN RESPONSE TO A GIVEN ENERGY BUDGET. THE GOAL BEHIND THE SECOND CHANGE IS TO OBTAIN WARMER TEMPERATURES THAN PREVIOUSLY FOR WATER SURFACES THAT ARE ICE COVERED. THE IMPACT OF THE THIRD CHANGE IS TO ENHANCE RADIATION LOSSES AT THE SURFACE THE PRESENCE OF VERY DRY ARCTIC AIRMASSES. PRACTICALLY, UNDER ARCTIC RIDGES, TEMPERATURE FALLS AT NIGHT AS FORECAST BY THE MODIFIED MODEL MAY EXCEED BY 10 DEGREES THE FALLS FORECAST BY THE CURRENT MODEL. THIS WILL GREATLY REDUCE THE WARM BIAS THAT WAS ASSOCIATED TO THE CURRENT OPERATIONAL MODEL UNDER SUCH CONDITIONS. INVERSELY, IN REGIONS UNDERGOING WARM AIR ADVECTION, THE TEMPERATURE FORECAST WILL BE HIGHER THAN THE ONE FORECAST BY THE CURRENT OPERATIONAL MODEL. HOWEVER, THE NEW MODEL RESPONDING MORE STRONGLY TO A GIVEN ENERGY BUDGET, THE ERRORS CAN BE LARGER IF THE CLOUD COVER IS NOT CORRECTLY FORECAST. THE OBJECTIVE VERIFICATION OF TEMPERATURE FORECASTS FOR 264 CANADIAN STATIONS GENERATED BY THAT MODEL FOR THE DURATION OF THE PARALLEL RUN HAS SHOWN A SIGNIFICANT IMPROVEMENT OF THE SURFACE TEMPERATURE FORECASTS. THE WARM BIAS HAS BEEN GREATLY REDUCED AND THE ROOT MEAN SQUARE ERROR HAS BEEN SUBSTANTIALLY REDUCED, ESPECIALLY FOR MINIMUM TEMPERATURES (WITH A GAIN OF OVER 0.5 DEG. C). THE FORECASTS HAVE IMPROVED OVER ALL CANADIAN REGIONS, EXCEPT IN THE ARCTIC FOR FORECASTS VALID AT 00Z AND IN BRITISH COLUMBIA FOR THOSE VALID AT BOTH 00Z AND 06Z. HOWEVER, IN THOSE CASES, THE SLIGHT REDUCTION IN QUALITY WAS SMALL COMPARED TO THE GAINS OBTAINED FOR THE OTHER VALID TIMES. SUBJECTIVE EVALUATION CARRIED OUT BY OPERATIONAL METEOROLOGISTS AT CMC GENERALLY HAS SHOWN THAT FORECAST SURFACE TEMPERATURES UNDER ARCTIC HIGHS CLOSE TO THE HOURS OF MINIMUM TEMPERATURE WERE MUCH COLDER AND VERIFIED BETTER AGAINST OBSERVATIONS FOR FORECAST ISSUED FROM THE MODIFIED MODEL IF THE CLOUD FORECAST BY THE MODEL FITTED REALITY. IN THOSE CASES WHERE THE MODEL DID NOT FORECAST CLOUDS AND IN FACT LOW CLOUDS WERE PRESENT, SURFACE TEMPERATURES FORECAST BY THE MODEL WERE SOMETIMES TOO COLD; HOWEVER, THOSE SITUATIONS ARE MUCH COMPENSATED BY THE SUBSTANTIAL IMPROVEMENT NOTED IN THE MAJORITY OF CASES. THE FORECAST MASS FIELD AS WELL AS PRECIPITATION HAVE SHOWN VERY LITTLE DIFFERENCE WITH THOSE OF THE OPERATIONAL MODEL. STATISTICAL TEMPERATURE FORECASTS BASED ON THE PERFECT PROG SYSTEM DRIVEN BY THE REGIONAL MODEL WILL BE AFFECTED INDIRECTLY BY THIS MODIFICATION SINCE THE SOME OF THE 1000 HPA FIELDS USED BY THE SCHEME WILL BE MODIFIED BY CHANGES IN SURFACE TEMPERATURES. EXPECTED IMPROVEMENTS IN STATISTICAL TEMPERATURE GUIDANCE WILL BE HOWEVER NOT AS LARGE AS THE ONES ASSOCIATED TO THE DIRECT MODEL OUTPUTS. NOTE THAT THIS MODIFICATION DOES NOT AFFECT THE GLOBAL SEF MODEL. MCBEAN/ADMA/TORONTO