for

1
Science

• for
• State Based Air Quality Programs

Harvey Jeffries, University of North Carolina,
Chapel Hill NC
2
A State AQ Agency must be able to

• Determine the cause(s) of highest ozone in their
  state.
• Confirm cause with diagnostic modeling and
  observational analysis
• If doubt exists, understand its impact on
  proposed policies and provide future correction
  operations

3
Cause of High Ozone

• General cause known diversity of the exact
  causes in a given episode is large.
• It is possible to propose the control of the
  wrong pollutant or wrong source sector
• Not a linear relationship b/w precursors and ozone

4
Modeling Cause Effect

• Modeling is often used to develop a Conceptual
  Model of Specific Causes
• Models depend on both formulation and large
  number of specific inputs
• Processes in model can be positive or negative
  and total ozone results of sum

5
Model Quality

• Biggest concern in accepting a model as cause
  is the quality of the inputs
• Second concern in accepting a model as true is
  avoiding compensating errors
• Third concern is assuring that the model is
  sufficiently representative of total set of causes

6
Confirming Cause

• Regulators can not live by modeling alone.
• Can not predict from observations
• Ozone purely secondary--no part emitted.
• Many indirect analyses needed

7
Inventories Key

• Inputs to the model the most important factor
  under agencys control
• The one input that rests almost entirely with
  state agency is local emissions inventories.
• More models!! MOBILE, BIES, NONROAD

8
Documentation of Inputs and Assumptions

• How reliable are inputs and assumptions?
• How much explaining away of model results is
  needed?
• Which modeling disagreements are important to
  decisions?
• Develop in-state infrastructure to provide
  support use consultants as needed.

9
What if ...

• More than one cause can be supported
• Model has big disagreements
• Cant control the real cause

10
Additional Analyses

• Conduct sensitivity studies to determine effect
  uncertainty has on proposed policies
• Apply Process Analysis to modeling results
• Make explicit decision on how to proceed via
  debate, get stakeholder agreement, and document
  it.
• Plan for future research and correction.

11
Vindicate Use

• Build an argument that no one can do it any
  better at the moment.
• Make it clear that uncertainty has been factored
  into decision
• Plan on future work and possible improved
  decisions.