Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

Intercomparison of low visibility prediction methods

VIEWS: 7 PAGES: 11

									Intercomparison of low visibility
      prediction methods

        COST-722 (WG-i)
   Frédéric Atger & Thierry Bergot
           (Météo-France)




                                     1
                     Proposal
• Horizontal surface (2m) visibility
• Main goal : learn about the value of the different
  existing methods
• Not a competition




                                                       2
                       Observations
• Participants propose airports where hourly visibility
  observations are available
• WG-i select one airport (validated by MC)
• Selected participant provides observations for 4 « winter »
  seasons (October to March) from Oct. 1999 to March 2003
    2 seasons for adjusting statistical methods and models
    characteristics
    2 seasons for evaluating existing methods




                                                                3
             Observed parameters
• Minimum required : hourly visibility (reference
  for verification)
• Any potentially useful parameters (according to
  local observation capabilities)
   for initializing models
   as predictors for statistical methods




                                                    4
                           Forecasts
• Forecast basis : 00 UTC and 12 UTC
• Lead times : +3h to +24h by 3h step
• It is not a competition :
    Participants indicate the main characteristics of the forecasting
     method (e.g. « 1D model coupled to ECMWF » or « MOS based on
     Aladin »)
    Participants are encouraged to provide alternative sets of forecasts
     obtained by modifying these characteristics




                                                                        5
                   Visibility thresholds
• Depend on WG-ii conclusions (requirements from the
  forecasters and from the customers)
• Proposal :
      200 m (roads)
      600 m (airports)
      1000 m (fog)
      5000 m (mist)
• Participants provide probabilistic or deterministic
  forecasts for as many thresholds as possible




                                                        6
                         Verification
• Comparison for a given validity (e.g. 06 UTC) and a given
  lead time (e.g. +6h)
• 2 verification aspects :
    Contingency tables  hit rate and false alarm rate  ROC or
     « pseudo-ROC » diagram
    Reliability diagram  Brier Score + reliability and resolution
     components
• Deterministic forecasts are considered as a special case of
  probabilistic forecasts




                                                                      7
ROC and pseudo-ROC curves




 2 definitions for the False Alarm Rate !
                                            8
Reliability diagram and Brier score
           decomposition


                  •   BS=(pi-oi)2/N
                  •   BS=REL-RES+UNC
                  •   REL=nk(pk-ok)2/N
                  •   RES=nk(ok-o)2/N
                  •   UNC=o(1-o)




                                          9
                    Evaluation tasks
• More efficient if performed centrally
    Could be performed by a WG-i participant not involved in the
     intercomparison
    Data (observations and forecasts) should be provided in due time
     and in a defined format by the participants
• Alternatively, each participant conducts the evaluation of
  its own forecasts (following common verification rules)
    All data and results should circulate among the participants




                                                                        10
Appendix : list of additional observed
             parameters
   2m temperature
   2m humidity
   10m wind
   1h/3h rainfall
   Total cloud cover
   Soil temperature
   Surface pressure
   Net short wave radiation near the ground
   Soil type (soil + vegetation)
   Radio-sounding observations




                                               11

								
To top