Objective Interpretation of Numerical Weather Prediction Model Output A Perspective Based on Verific

1
Objective Interpretation of Numerical Weather
Prediction Model Output A Perspective Based on
Verification of Temperature and Precipitation
Guidance

• J. Paul Dallavalle and Valery J. Dagostaro
• Meteorological Development Laboratory
• National Weather Service/NOAA
• Symposium on the 50th Anniversary of Operational
  Numerical Weather Prediction
• June 14-17, 2004
• University of Maryland
• College Park, MD

2
Outline

• History of NWS verification program
• Methodology
• Results
• Conclusions

3
History of NWS Verification Program

• April 1966 begins
• Initially compared NMC subjective guidance and
  NWS local forecasts for max/min temperature and
  PoP
• April 1970 perfect prog max/min guidance
  replaces NMC guidance
• January 1972 MOS PoP guidance replaces NMC
  guidance
• August 1973 MOS max/min guidance replaces
  perfect prog guidance

4
Forecast Elements

• Max/Min Temperature
• Probability of Precipitation

• Local forecast is for
• daytime max/nighttime min
• Local forecast is for 1200-0000 UTC (day) and
  0000-1200 UTC (night)

5
Forecast Projections

• 0000 UTC cycle
• ( Added in October 1975
• for temperature in Summer 2002 for PoP)
• 1200 UTC cycle

• Today
• Tonight
• Tomorrow
• Tomorrow Night
• Tonight
• Tomorrow
• Tomorrow Night
• Day after Tomorrow

6
Methodology

• 80 CONUS sites
• Verification statistics for warm (Apr. - Sept.)
  and cool (Oct. Mar.) seasons
• Official verification observation
• Official guidance
• Fixed climate standard (1961- 90 normals for
  max/min 1972- 85 relative frequencies for PoP)

7
Verification Sites
8
Verification Measures

• Max/Min Temperature
• Mean Absolute Error
• PoP
• Improvement in Brier Score relative to climate

9
Significant Changes to Official Guidance

• January 1972 -- PE-based MOS PoP
• August 1973 -- PE-based MOS max/min
• April 1980 -- LFM-based MOS
• November 1985 -- daytime max/nighttime min
• June 1993 -- NGM-based MOS
• Summer 2002 -- GFS-based MOS

10
Max Temperature Cool Season (0000 UTC Cycle)
11
Max Temperature Warm Season (0000 UTC Cycle)
12
Min Temperature Cool Season (1200 UTC Cycle)
13
Min Temperature Warm Season (1200 UTC Cycle)
14
PoP Cool Season(0000/1200 UTC Cycles Combined)
15
PoP Warm Season(0000/1200 UTC Cycles Combined)
16
Conclusions

• Public weather forecasts and guidance have
  improved during the period
• Quality of day 2 local forecasts now is equal
  to that of the day 1 forecasts 20 years ago
• Quality of day 2 guidance now is equal to that
  of the day 1 guidance 10 years ago
• Skill of PoPs is higher in cool season errors
  in max/min are greater in cool season, but so is
  rate of improvement
• High correlation between quality of guidance and
  local forecasts