NOCN02 CWAO 071930 GENOT FAX NO. 021 LE GENOT NO. 022 EST LA VERSION FRANCAISE DU GENOT NO 021 IMPLEMENTATION OF A NEW GLOBAL MODEL AT CMC... ON WEDNESDAY OCTOBER 14 1998, AT 1200 UTC, THE CANADIAN METEOROLOGICAL CENTRE WILL IMPLEMENT ITS NEW CANADIAN GLOBAL MODEL. THE GLOBAL ENVIRONMENTAL MULTI-SCALE (GEM) MODEL WILL REPLACE THE T199 GLOBAL SPECTRAL MODEL CURRENTLY USED TO PRODUCE THE MEDIUM AND LONG-RANGE FORECASTS. THE NEW MODEL WILL ALSO PROVIDE THE TRIAL FIELDS REQUIRED BY THE GLOBAL DATA ASSIMILATION SYSTEM TO PRODUCE THE GLOBAL ANALYSIS. THE CHANGE TO A NEW GLOBAL MODEL WILL ALSO IMPACT ON THE REGIONAL FORECAST SYSTEM SINCE THE GLOBAL TRIAL FIELDS ARE USED TO INITIALIZE THE 12-HOUR SPIN-UP OF THE REGIONAL FORECAST SYSTEM. THIS NEW MODEL IS THE SAME AS THE ONE CURRENTLY RUN BY CMC TO PRODUCE THE SHORT-TERM (48 H) FORECASTS. THE MODEL USES A VARIABLE RESOLUTION MESH FOR THE LATTER APPLICATION, WHILE FOR ITS NEW APPLICATION, I.E. MEDIUM-RANGE FORECASTS AND GLOBAL DATA ASSIMILATION, IT WILL BE RUN ON A UNIFORM LAT-LONG GRID WITH UNROTATED POLES. ALL PRODUCTS CURRENTLY USING THE GLOBAL T199 MODEL OUTPUT WILL FROM NOW ON BE PRODUCED USING THE OUTPUT OF THE UNIFORM GLOBAL GEM MODEL. A. SOME CHARACTERISTICS OF THE TWO MODELS. OPERATIONAL GLOBAL T199 NEW GLOBAL GEM MODEL ======================== ======================= HORIZONTAL SPECTRAL FORMULATION FINITE ELEMENTS DISCRETIZATION HORIZONTAL GAUSSIAN GRID 400 X 200 GRILLE LAT-LONG 0.9 DEG. RESOLUTION VERTICAL 21 SIGMA LEVELS 28 HYBRID LEVELS. THE RESOLUTION ADDITIONAL LEVELS ARE IN THE STRATOSPHERE AND AT THE TROPOPAUSE. TIME 3 TIME-LEVEL SEMI- 2 TIME-LEVEL SEMI- DISCRETIZATION LAGRANGIAN SCHEME LAGRANGIAN SCHEME TIMESTEP 1800 SEC 2700 SEC HORIZONTAL INCREASES LINEARLY CONSTANT THROUGHOUT DIFFUSION FROM INITIAL TIME TO 96 THE INTEGRATION HOURS, REMAINS CONSTANT AFTERWARDS BOTH MODELS USE A SIMILAR SET OF PHYSICAL PARAMETERIZATIONS, THE ONE EXCEPTION BEING THE STRATIFORM CONDENSATION SCHEME: THE NEW GLOBAL GEM MODEL USES A SUNDQVIST SCHEME WHILE THE GLOBAL T199 MODEL USES A SIMPLE STRATIFORM CONDENSATION SCHEME. THE MEAN-SEA LEVEL REDUCTION ALGORITHM AND THE ONE USED TO OBTAIN GEOPOTENTIAL AND TEMPERATURE VALUES ON PRESSURE LEVELS WHERE THOSE LEVELS ARE BELOW THE TOPOGRAPHY ARE DIFFERENT IN THE TWO MODELS. THE T199 SPECTRAL MODEL USES A VERTICAL EXTRAPOLATION TO OBTAIN THE DESIRED VALUES, WHILE THE NEW GLOBAL GEM MODEL USES AN ALGORITHM EQUIVALENT TO A LINEAR HORIZONTAL INTERPOLATION USING THE TEMPERATURES SURROUNDING THE TOPOGRAPHY FROM WHICH THE GEOPOTENTIAL AND TEMPERATURE FOR THE REQUIRED PRESSURE LEVELS AS WELL AS A MSLP ARE EXTRACTED. THE TWO METHODS CAN GIVE SIGNIFICANTLY DIFFERENT RESULTS, ESPECIALLY IN WINTER AND ABOVE HIGH ELEVATIONS SUCH AS GREENLAND WHERE ANTICYCLONES FORECAST BY THE NEW GEM MODEL COULD BE UP TO 50 HPA LESS INTENSE THAN IN THE T199 MODEL. B. PERFORMANCE EVALUATION BEFORE THE PARALLEL EVALUATION OF THE GEM MODEL BEGAN, A THOROUGH STUDY OF THE MODEL FOR ALL FOUR SEASONS WAS PERFORMED. AN ENSEMBLE OF CONTINUOUS DATA ASSIMILATION CYCLES WERE DONE, AS WELL AS A LARGE NUMBER OF LONG-TERM FORECASTS. THIS STUDY WAS FOLLOWED BY SEVERAL MONTHS OF A COMPLETE (ANALYSIS AND FORECASTING) PARALLEL RUN AT CMC. THE RESULTS OBTAINED WHEN COMPARED WITH THOSE OF THE OPERATIONAL T199 MODEL RUN IN SIMILAR CONDITIONS HAVE SHOWN THAT: FOR THE ASSIMILATION SYSTEM, THE VERIFICATIONS AGAINST RADIOSONDE DATA OF THE ANALYSIS AND TRIAL FIELDS OBTAINED USING THE NEW GEM MODEL HAVE SMALLER RMS ERRORS FOR ALL VARIABLES, AND MORE NOTABLY FOR THE TEMPERATURE AND WINDS. THE RESULTS FOR THE BIAS VARY WITH THE SEASON, AND THE VERIFICATION REGION. HOWEVER WE OBSERVE IN GENERAL SLIGHTLY LARGER GEOPOTENTIAL BIASES IN THE STRATOSPHERE FOR THE NEW GEM MODEL, AND TEMPERATURE BIASES THAT ARE EQUIVALENT. THE WIND BIAS IS IN GENERAL SLIGHTLY REDUCED, WHILE THE STRONG HUMIDITY BIAS WHICH WAS CHARACTERISTIC OF THE T199 MODEL IS GREATLY REDUCE IN THE ASSIMILATION SYSTEM USING THE GEM MODEL. FOR THE FORECAST SYSTEM, THE BIAS AND RMS ERROR OF THE WINDS AND THE DEW POINT DEPRESSION ARE REDUCED OR UNCHANGED WITH THE NEW MODEL FOR ALL FORECAST TIMES (FROM 0 TO 240 HOURS). THE IMPROVEMENTS IN WINDS FORECASTS IN THE STRATOSPHERE ARE PROBABLY DUE TO THE INCREASED VERTICAL RESOLUTION OF THE NEW MODEL. IN THE TROPOSPHERE, AND FOR UP TO 144H, THE BIAS AND RMS ERROR OF THE GEOPOTENTIAL AND TEMPERATURE ARE GENERALLY VERY SIMILAR. BEYOND THAT FORECAST RANGE AND IN THE STRATOSPHERE, THE BIAS AND THE RMS ERRORS FOR GEOPOTENTIAL AND TEMPERATURE ARE SLIGHTLY LARGER THAN THOSE OBTAINED WITH THE T199 OPERATIONAL MODEL. DURING THE PARALLEL RUN, THE FORECASTS WERE EVALUATED BOTH OBJECTIVELY AND BY THE OPERATIONAL METEOROLOGISTS AT CMC. THESE EVALUATIONS HAVE SHOWN THAT: 1. THE ANALYSIS DONE USING THE TRIAL FIELD OF THE NEW GLOBAL GEM MODEL ARE OFTEN JUDGED TO BE SLIGHTLY SUPERIOR TO THE OPERATIONAL ANALYSIS. THE AMPLITUDE OF THE SYSTEMS IS OFTEN BETTER PRESERVED BY THE NEW MODEL, AND THIS IS PARTICULARLY OBVIOUS WITH THE JET STREAM ANALYSIS WHICH HAS BETTER DEFINED CENTRES AND SLIGHTLY LARGER MAXIMUM IN THE NEW SYSTEM. 2. THE FORECAST FIELDS OF THE NEW MODEL ARE GENERALLY BETTER THAN THOSE OF THE T199 OPERATIONAL MODEL. THE NEW MODEL IS DEFINITELY MORE ACTIVE, AND FREQUENTLY FORECASTS LOWS THAT ARE SLIGHTLY DEEPER AND WHOSE POSITION VERIFIES BETTER THAN THOSE OF THE OPERATIONAL T199 MODEL. OCCASIONALLY THE NEW MODEL CAN OVERDEVELOP SYSTEMS. 3. THE NEW MODEL'S FORECAST HUMIDITY IS SIGNIFICANTLY LOWER THAN THAT OF THE OPERATIONAL T199 MODEL, ESPECIALLY IN THE LOWER LEVELS. THIS BEHAVIOUR CLEARLY VERIFIES BETTER WITH OBSERVATIONS. 4. THE DIFFERENCES IN FORECAST HUMIDITY HAVE A SIGNIFICANT IMPACT ON THE PRECIPITATION FIELDS. THESE FIELDS ARE SOMETIMES SMOOTHER WITH THE NEW MODEL AND SHOW LOWER MAXIMUM VALUES, ESPECIALLY IN THE CASE OF CONVECTIVE PRECIPITATION AND DURING THE FIRST FEW DAYS OF THE INTEGRATION. IN GENERAL HOWEVER, THE FORECAST PRECIPITATION IS JUDGED BETTER IN THE NEW SYSTEM, THOUGH IT WAS NOTED IN SEVERAL INSTANCES THAT THE FORECAST MAXIMUM COULD HAVE BEEN HIGHER. 5. THE VERTICAL MOTION FIELD IS SLIGHTLY SMOOTHER IN THE EARLY PART OF THE FORECASTS OF THE NEW MODEL. THIS FEATURE ATTENUATES ITSELF AS THE INTEGRATION PROGRESSES, PROBABLY BECAUSE THE DIFFUSION IN THE OPERATIONAL T199 MODEL INCREASES WITH TIME, WHILE IT REMAINS CONSTANT IN THE NEW MODEL. 6. THE IMPACT ON THE REGIONAL SYSTEM OF USING AT THE BEGINNING OF ITS 12-HOUR SPIN-UP A TRIAL FIELD FROM THE NEW GLOBAL GEM MODEL INSTEAD OF ONE FROM THE OPERATIONAL GLOBAL T199 MODEL WAS CONSIDERED NEUTRAL. MCBEAN/ADMA/TORONTO