What level of complexity do we need for models and data for Data Assimilation

1
What level of complexity do we need for models
and data for Data Assimilation?
Answer Open-ended!
2
Data-centered

• Relatively simple passive microwave data like
  brightness temperature are actually ill-suited
  for ice models which already assume constant
  emissivity
• Sea ice motion and ice area can be ingested into
  existing ice models used operationally
• More complicated info such as melt ponds,
  pressure ridge locations, albedo, etc., require
  more complex physics than are in operational
  models

3
Model-centered

• Model physics must be sufficient to handle the
  ingested data without violating physical
  assumptions or physical reality
• For example, a linear free drift dynamics model
  is necessary but not sufficient for assimilating
  ice motion fields

4
Application-centered

• Short-term analyses (1 day) may not need to be
  as physically rigorous
• Longer-term applications should obey all relevant
  laws of physics and conservation
• Models which include an ocean require at least
  that the ocean data assimilated does not make the
  ocean solution go unstable!

5
Errors, Errors, Errors

• Errors in the data, in the model, and in the data
  assimilation technique must be well-understood
• Without a good understanding, there is a
  possibility that the sum of the results could be
  worse than its individual parts

6
Model development

• Ice-ocean model development can and should
  continue to be improved
• More data (types, inventory) should be available
  in the future
• Increasingly complex models should be capable of
  ingesting more types of data