ESRL%20Model%20Probabilistic%20Post%20Processor%20(EMP3)

1
ESRL Model Probabilistic Post Processor (EMP3)

• And
• Personal Weather Advisor
• Holub, McCaslin, McGinley, Sahm
• GSD/FAB

2
Full Ensemble Forecast System with Decision
Support Attributes.Assumes everyone wants a
Yes-No Forecast
Location, Time
Cost/Loss Thresholds
Ensemble
Decision Engine
6-D model grids (variable,x,y,z,t,prob)
Yes or No
Customer
3
Goal Test User Interface for Public Access
Surface data only Threshold Settings Probabilist
ic Thresholding (decision criteria) Start with
SREF Simple assumptions on PDF recovery
4
Defining acceptable risk

• Danger threshold defined by user
• Probability density function is used to find
  probability of threshold
• P(danger) area B/ (area A area B)
• Interrogator determines users risk factor, R
• No go if P(danger) gt R

Probability density of event
A
B
Weather element
Danger threshold
5
Summary
EMP3 allows user to set meteorological criteria
critical to success of some task Probabilistic
criteria is set based on Cost-Lost, acceptable
risk factors, etc, defined by interrogator Person
al Weather Advisor indicates time periods
where task should be attempted. Utilizes Google
Earth via NOAA license
6
Future work
Verification and bias removal of ensemble set
Multiple variables and joint probabilities More
advanced interrogator Decision support for
routes (point A to point B, via C) Decision
support for areas