Decision Support Systems: Engaging End Users Bill Mahoney Director, Weather Systems

1
Decision Support SystemsEngaging End
UsersBill MahoneyDirector, Weather Systems
Assessment ProgramResearch Applications
Laboratory (RAL)National Center for Atmospheric
Research
2
NCAR RAL Decision Support Systems

• Airline dispatchers
• Air traffic controllers
• Pilots
• Army test range staff
• Road maintenance practitioners
• Water resource managers
• Weather forecasters
• Airline operations personnel
• Airport operations personnel

3
Engaging End Users - Why is this topic important?

• The importance of connecting science to society
  has risen in the last decade.
• Most new research funding is targeted or directed
  to support a societal need.
• Researchers have traditionally not been well
  connected to end users making it difficult for
  society to take advantage of results.
• Learning how to engage end users is critical for
  advancing science and technology.

4
Decision Support Systems
What is a decision support system?

• An automated tool that makes decisions?
• A semi-automated tool?
• A handbook of recommended practices?
• Local newspaper or news program?
• Student assistant?

Answer All the above!
5
DSS as a Tool

• The best way to understand how someone will
  utilize new information is to introduce it into
  their operational environment.
• The way the new information is used will evolve
  (sometimes significantly) over time.

6
Decision Support Systems

• Before one can consider developing or
  implementing a DSS, some important questions need
  to be asked.

7
Decision Support Systems

• First Question
• 1) What problem(s) are you trying to solve?
• This must be asked several different ways before
  a potential solution may emerge.

8
Decision Support Systems

• 2) What is the culture of your organization?
• - Would a DSS be seen as threatening?
• - Does automation pose problems?
• - Are support systems/tools viewed positively?
• 3) What actual decisions could be supported?
• 4) What job categories would benefit most?

9
Decision Support Systems

• 5) What technical capabilities exist?
• - Is there an in-house framework for a DSS?
• gt Networked system (external internal)
• gt Desktop computers available?
• gt Remote communications used (PDAs)
• gt Database of pertinent data
• gt Are probabilistic products utilized?

Remote Systems
10
Decision Support Systems
6) What are the potential benefits? - Safety -
Productivity/Efficiency 7) Who will champion the
technology? - Management vs. staff - Technology
push (by management) or pull (by staff)?
11
Decision Support Systems

• Application Categories
• Strategic Planning (condition prediction)
• Tactical Planning (alert functions)
• Operations Management (productivity)
• Incident Management (notification function)
• Risk Management
• Evaluation of What if? scenarios
• Training Tool (off line assessments)

12
DSS Development Issues

• There are no off-the-shelf plug and play weather
  and climate capabilities that can fully address
  the needs of all user communities. No
  one-size-fits-all solutions.
• A bottoms-up rather than a tops-down
  approach should be used for DSS system
  development.
• Stakeholders need to determine the level of
  sophistication that is required for their
  specific DSS application.

13
Maintenance Decision Support System

• Weather parameters
• Air temperature
• Relative humidity
• Wind speed and direction
• Precipitation type, rate, accumulation
• Road Parameters
• Road temperature
• Bridge temperature
• Bridge frost potential
• Blowing snow potential
• Road contamination chemical concentration

• Treatment Recommendations
• Treatment type
• Treatment amount
• Treatment location

14
Maintenance Decision Support System
National Weather Service Data Supplemental
Weather Models
MM5 WRF Rapid Update Cycle (RUC)
Multiple Forecast Members
NAM Model MAV MOS GFS Model SYNOP METARS Eta MOS
DOT Data
RWIS
Road Condition Treatment Module - Road
Temperature Prediction Model - Chemical
Concentration Algorithms - Rules of Practice for
Anti- and Deicing
Plow route specific treatment recommendations
15
Communicating Uncertainty
Transportation decision makers are beginning to
accept the concept of probabilities in the
presentation of weather information because they
make risk assessments daily. However, they
really dont understand what a 20 chance of
showers means. What is confidence? What is
probability? How do they compare?
16
Communicating Uncertainty(to non-meteorologists)
Conditional Probability of Precipitation Type
Product
17
The Climate Forecast Applications Project

• Bangladesh at confluence of Brahmaputra and
  Ganges Rivers
• Large river basins with limited warning of
  un-gauged upstream river flows
• Product Provide operational upper catchment
  flood-stage discharge and precipitation forecasts
  at differing time-scales

Tom Hobson
18
Short-term 1 to 10 day Discharge Forecasting for
the Ganges and Brahmaputra
Operational ensemble forecasts for the Ganges
(5-day-in-advance) and Brahmaputra
(9-day-in-advance) are shown in the top two
panels, (critical discharge levels black
dash) Confidence intervals (50 and 95) are
shown in the next two panels Above-Danger-Level
probabilities are shown in the lowest two
panels. Notice that the severe July,
2004 upper-catchment Brahmaputra floods were
well-forecasted.
Tom Hobson
19
Operational Seasonal Forecasting of Discharge
-- Using the ECMWF ensemble precipitation
forecasts, discharge ensemble forecasts can be
generated -- Shown here are 1 to
6-month in-advance discharge ensemble forecasts
for 2004 (monthly-averaged observed discharge
is dashed climatological average discharge Is
dotted) issued in May, June, July, August
20
Engaging End Users - Research Needs

• The meteorological community should develop a
  Best Practices guide for engaging end users.
• Weather DSS Testbeds should be established in
  end-user operational environments to test
  advanced products (methods, techniques, etc.)
  and to learn how end-users utilize the new
  information.

21
Engaging End Users - Research Needs

• Testbeds should be used to evaluate
• How to communicate uncertainty
• How to communicate risk
• New presentation options
• How users interpret products
• Identify knowledge gaps in atmospheric science
• Evaluate societal aspects of new capabilities

22
Engaging End Users - Research Needs

• Weather DSS Testbeds for a variety of economic
  sectors could be established. Industry partners
  could help offset costs (in-kind contributions,
  hosting of technology, etc.)
• Weather DSS Testbeds
• Agriculture Aviation
• Surface Transportation Human Health
• Energy Construction
• Water Resources Emergency Management

23
Thank You