Comments from a global modeling perspective Thomas L' Delworth GFDLNOAA

1
Comments from a global modeling
perspectiveThomas L. DelworthGFDL/NOAA

• How is observational data used most effectively
  in concert with models to improve our
  understanding and prediction of global climate
  change?
• Evaluation of credibility of models
• Quantification of changes in radiative forcing
• Improved understanding of key physical processes
  and their representation in models
• Detection and attribution of change separating
  internal variability from forced response

2

• From this perspective, many important themes are
  stressed
• Continuity in time, completeness in space
• Emphasis on reanalysis, especially oceanic
• Minimization of temporal drift
• Stabilization existing observational capabilities
• Enhancements to operational networks to make them
  suitable for climate purposes
• Length of records, including both data
  archeology and paleo-reconstructions

3
OBSERVATIONAL ESTIMATE
1023 J
SIMULATED
4
Low signal to noise ratio in some components of
the climate system has significant implications
for observational systems
5
Whats missing?

• A more focused set of questions that (new)
  observations would be targeted to address.
• Support for design studies of optimal observing
  systems.
• Support for evaluating the utility of
  initializing models from an observed state for
  predictions on greater than decadal time scale.

6
Final Comments

• Need for enhanced guidelines in evaluating the
  overall priority of different (and competing)
  observational programs. The table on page 136 is
  a good start at providing a set of prioritization
  criteria for observations. This could be coupled
  with more explicit questions that the
  observations are addressing to construct detailed
  priorities.
• 2. There is a need for more emphasis on
  cost/benefit analyses of observing systems how
  much is enough, and for what question? Models can
  help here.

7
Final comments (continued)

• 3. Need for strong(er) emphasis on maintaining
  existing networks which have long observational
  record, and whose utility has been demonstrated.