A Glimpse into the Techniques and Interpretation of the CPC
Transcription
A Glimpse into the Techniques and Interpretation of the CPC
A Glimpse into the techniques and interpretation of CPC‟s Extended-Range Forecasts (ERF) TEMPERATURE PRECIPITATION Anthony Artusa NOAA Climate Prediction Center (301) 763-8000 ext. 7530 anthony.artusa@noaa.gov WeatherBug Roosevelt Hotel NYC October 7, 2010 1 COMPONENTS FLOW PATTERN TEMPERATURE PRECIPITATION WRITE-UP Valid: Sep 01 - 05, 2010 (6-10 Day Outlook) Valid: Sep 03 - 09, 2010 (8-14 Day Outlook) Issued: Aug 26, 2010 PROGNOSTIC DISCUSSION FOR 6 TO 10 AND 8 TO 14 DAY OUTLOOKS NWS CLIMATE PREDICTION CENTER CAMP SPRINGS, MD 300 PM EDT THU AUGUST 26 2010 6-10 DAY OUTLOOK FOR SEP 01 - 05 2010 THE MODEL SOLUTIONS ARE IN FAIR AGREEMENT TODAY, LESS AGREEMENT THAN EARLIER IN THE WEEK. THE MODEL SOLUTIONS FEATURE A WEAK RIDGE NEAR THE ALEUTIANS, WEAK TROUGHING ALONG THE WEST COAST OF NORTH AMERICA, RIDGING ACROSS THE VERIFICATION 2 Difference Between Forecasts Long-waves Short-waves 5-day Average Height Field (D+8) Components of D+8 Height Field (132 -240 Hour Forecast) 3 What is the main difference between the 6-10 and Week 2 forecasts? 6-10 500-hPa Height/Anomalies May 23-27, 2010 8-14 500-hPa Height/Anomalies May 25-31, 2010 Averaging period is the only difference between these two forecasts. 4 Comparison of Models 5 Background notes: 00z (8:00 PM (EDT)) 12z (8:00 AM (EDT)) 6-10 day forecast: 5-day mean centered on day 8 (D+8) 8-14 day forecast: 7-day mean centered on day 11 (D+11) 6 ENSEMBLES t+14 t=0 {IC} ANALYSIS (data assimilation provides best estimate of initial state of atmosphere) M2 M2 M1 CTRL M1 low-res version of analysis CTRL M3 M3 7 Comparison of Models – D+8 Heights Ensemble Means (top) & Operationals (bottom) 0z GFS ENS MEAN 0z CAN ENS MEAN y-12z EC ENS MEAN 0z GFS 6z GFS y-12z ECMWF 8 Comparison of Models – D+8 Anomalies Ensemble Means (top) & Operationals (bottom) 0z GFS ENS MEAN 0z CAN ENS MEAN y-12z EC ENS MEAN 0z GFS 6z GFS y-12z ECMWF 9 D+8 Spaghetti Contours (00z GFS ENM) Valid May 23-27, 2010 5520 meters 5760 meters Ensembles provide a measure of forecast uncertainty Ridge over eastern US (right panel) is more amplified than normal (compare with black climatology line) 10 Model 6-10 day 8-14 day 00z GFS Blend Weights: assigned using a subjective/objective mix; consider the Anomaly Correlation (AC) scores for the models over past 60 days. 06z GFS 10% y12z GFS y18z GFS 00z GFS ENM 20% 06z GFS ENM 25% y12z GFS ENM y18z GFS ENM GFS SUPER ENM 40% 00z ECMWF y12z ECMWF 00z ECMWF ENM 35% 50% 10% 10% y12z ECMWF ENM 00z CAN ENM y12z CAN ENM 11 500-hPa Blend Forecast made: May 17, 2010 This product is available to the public via the web. AWIPS version has heights and anomalies as two separate figures. 12 30-year historical temperature observations (PDF) Middle tercile Coldest tercile Warmest tercile N B 33% A 67% Credit: (Copyright) Robert Niles, http://www.robertniles.com/stats/stdev.shtml 13 Partitioning of Probabilities 100% 80% 13 30 33 33 53 60% 33 39 40% 20% 53 33 33 Above Near Below 30 13 0% CLIM NOR B>N>A B<N<A TERCILE BOUNDARIES 14 Interpretation of ERF Surface Outlook N Central MN Below: 31.3% Normal: 37.3% Above: 31.3% N=Near Normal Category Contours are labeled with the total probability for the indicated category. N 33 N 40 Southern NV Below: 45% Normal: 33% Above: 22% 33 40 B A N 50 Northern GA Above: 55% N Normal: 33% Below: 12% 15 Generating temperature and precipitation outlooks for the 6-10/8-14 day period: Temperature Tools Precipitation Tools Klein Equations Calibrated Precipitation NAEFS NAEFS Analogs Analogs Neural Network Neural Network Re-forecasts Re-forecasts Automatic Forecast Automatic Forecast 16 Temperature Tool (Kleins) Klein – 0z GFS ENS Klein – 0z GFS Klein – Manual Blend Klein – 6z GFS Screening Regression that derives multiple linear regression (MLR) equations which express 5-day mean surface temperature as a function of 5-day mean 700-mb heights centered 2 days earlier, and local 5-day mean surface temp for a period 4 days earlier than the forecast period. 17 Temperature Tool (0z NAEFS) B E L O W N E A R A B O V E Joint Ensemble: 20 Canadian members & 20 GEFS members per Cycle (00z & 12z) Equally weighted members and both ensemble systems are also equally weighted Both ensembles are „postprocessed‟ (bias-corrected & calibrated) before combined into one NAEFS. Valid May 25-31, 2010 Bias Correction: www.emc.ncep.noaa.gov/gmb/yzhu/gif/pub/manscript_bocui_bias_correction_20100709.pdf Calibration: Unger, David A., Huug van den Dool, Edward O‟Lenic, Dan Collins, 2009: Ensemble Regression. 18 Mon.Wea.Rev., 137, 2365-2379 (journals.ametsoc.org/doi/abs/10.1175/2008MWR2605.1) Precipitation Tool (NAEFS) Wet Tercile Dry Tercile 19 ANALOGS A1 A10 A2 A3 A9 Height prog Avg Composite Analog A4 A8 A5 A7 A6 20 COMPOSITE HEIGHT ANALOG ANALOG WINDOW: +/- 6 WEEKS 21 Temperature & Precipitation Tools (Analogs) 0z GFS ENM 0z GFS y-12z EC ENM y-12z EC y-12z EC ENM y-12z EC 22 Temperature & Precipitation Tools (Neural Net) 0z CAN ENM 0z EC ENSM Derives statistical relationships between 500-mb heights and surface fields; uses larger domain than Kleins 6z GFS 0z EC 23 Temperature & Precipitation Tools (Reforecast) MRF (T62) model (Jan thru Jun 1998) ; Whitaker Initial conditions use reanalysis (ncep / ncar) 15-member ensemble (Refcst from 11/78 - 2005) Calibrated probabilities TEMP PRCP 24 Temperature & Precipitation Tools (Auto) TEMP PRCP 25 Sample 06Z GFS model (12-hr total precipitation; MSLP; 850 mb temperatures) 26 Nocturnal Thunderstorm Clusters (MCS) often track along the southern edge of the westerlies (582-588 “corridor”) 27 Teleconnections (Temperature) Made May 17, 2010 500hPa 500hPa + Surface Temp - Surface Temp 28 Teleconnections (Precipitation) Made May 17, 2010 500-hPa 500-hPa + Surface Prcp - Surface Prcp 29 Official CPC Surface Forecasts TEMP PRCP 30 PROGNOSTIC MAP DISCUSSION ( 6-10 day & 8-14 day ) SYNOPSIS: Ridges/Troughs Blend Weights & Confidence Temperature TEMPERATURE TOOLS USED Tools Used Precipitation PRECIPITATION TOOLS USED Tools Used 31 500-hPa Heights/Anomalies Verification (May 23-27, 2010) OBSERVED HEIGHTS OBSERVED ANOMALIES 32 Temperature Verification (May 23-27, 2010) FORECAST OBSERVATIONS General Heidke Skill Score for 3 equal classes: (H – E) * 100 ---------------------------------(T – E) H: # of “Hits” = E: expected # of hits (=1/3) +71.4 T: total # of stations 33 Precipitation Verification (May 23-27, 2010) OBSERVATIONS FORECAST General Heidke Skill Score for 3 equal classes: (H – E) * 100 ---------------------------------(T – E) H: # of “Hits” = E: expected # of hits (=1/3) +17.7 T: total # of stations 34 Everglades Heron 35