9
“Probability forecast use” study Edwin Welles and Jan Verkade February 2012

“Probability forecast use” study Edwin Welles and Jan Verkade February 2012

Embed Size (px)

Citation preview

Page 1: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

“Probability forecast use” study

Edwin Welles and Jan Verkade

February 2012

Page 2: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Theme of present project: probability forecasting

Page 3: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

“How to realise the benefits of probability forecasting?”

• Probability forecasting brings benefits to forecasters and end users

• Having a forecasting system that estimates predictive uncertainty is probably not sufficient to realise these benefits

What needs to be done in addition to having a probability forecast?

Present project aims to preliminary answer this question

By eliciting expertise/judgement from forecasters and end-users

Page 4: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Present project: use of probability forecasts

• Looking at aspects such as:

• visualisation

• communication

• decision-making

• verification

• training

• “downstream” decisionsupport systems

• business procedures

Page 5: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Project participants

• Flood Control 2015 programme (funding)• US National Weather Service, NCRFC• Meuse River Forecasting Centre (RWS, Dienst Limburg)• Water board Noorderzijlvest• Scottish Environment Protection Agency (TBC)• Why Deltares?

• Delft-FEWS (CHPS) related research

• PhD research on predictive hydrological uncertainty

• The Netherlands can benefit from US prediction enterprise

Page 6: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Case studies

• not too large• straightforward forecast – decision – warning – response chain

(not too many stakeholders involved)• largely based on desk research and interviews• desk research: review of current operational procedures• interviews: elicit expertise and/or expert judgement from

forecasters and emergency managers “how would your tasks/responsibilities change if probability forecasts were to be introduced?”

Page 7: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Water Board “Noorderzijlvest”

• Water Board: responsible for maintainingwater levels in polder districts withinacceptable levels (Fully controlled systems, well below MSL)

• 2010 event: flood warning called, but nothing happened• Hydrologist was blamed• Way forward: probability forecasts

Page 8: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Water Board “Noorderzijlvest”

• November 2011: forecasting – warning – response exercise• Lessons:

• interpretation of probability forecasts not an issue

• information overload is

• decision makers: with these forecasts, I don’t have to make my own estimates of the inherent uncertainties

• probability forecasts used to devise scenarios (worst case)

Page 9: “Probability forecast use” study Edwin Welles and Jan Verkade February 2012

November 30, 2011

Study deliverables

• Description of “best practices” of using probability forecasts within FFWRS, from multiple case studies

• Inventory of likely challenges / possible obstacles for effectively using probability forecasts

• Seminar, Report, scientific article• (while we’re at it: collect relevant reports from earlier, related

studies bibliography, links and downloads will be published on the Deltares wiki pages)