Session S3 - Solar Wind Prediction - Linker, Liu, DeRosa 

[Linker]

Here are the things that came out of our session, as far as observational
inputs that global solar wind models would like from HMI:

1) A better approximation to the radial magnetic field at the photosphere
in active regions.

2) Better approximation for polar fields, perhaps from the transverse
field and/or flux evolution models.

3) Flux evolution maps and perhaps other parameters of flux evolution
provided as a product for models.

4) Mitigation of the time discontinuity on the eastern edge for evolving
maps, e.g., the well-known "I only got the leading polarity of a sunspot"
in the assimilation window. One suggestion I heard from the audience:
try to automatically shrink the assimilation window if flux imbalance
exceeds some threshold.

5) Time cadence of several maps a day is probably adequate for global
modeling.

6) High resolution maps (~1 arc sec) will be necessary for some
applications, but will most likely only be used in active regions.

7) Possible inclusion of far side maps.

8) It is difficult to determine which instrument/observatory is producing
the "best" magnetic maps. Reasons for this include that the models
are suspect in various ways, seeing/other conditions vary at different
sites for a given comparison, the modelers post-process the maps prior
to use, and the model parameters may be biased towards producing the
best results for a given instrument/observatory.  After the session,
Todd and I discussed the possibility of a small workshop to get at
this question.  We would pick a small sample (1-3) of time periods when
observing conditions are deemed good for a set of instruments (e.g.,
Kitt Peak, MDI, Mount Wilson, Wilcox).   The magnetographers would do
their best job to process the data, and there would be analysis of the
similarities/differences.  Then the modelers would produce solutions
for the data from the different instruments and compare with some set
of coronal and solar wind data.

___
[Liu]

1. For WSA solar wind model, is any siginificant influence from
streamer-streamer interaction? What cause failure of this model except
solar transients? (Xuepu and Keiji have been working on streamer-streamer
interaction for a while, trying to answer the first question).

2. SAIC code can apply to high spatial and temporary resolution data and
then provide possible products needed such as heliospheric structure,
solar wind speed and IMF, ... Is it possible to start with MDI data in
near future?

3. Jon mentioned demanding good data (vector magentograms with 180 degree
ambiguity well solved). Is it a possible solution for HMI to average
several sets of vector magentograms to meet such data requirement? If so,
what kind of data can we play around now?

4. We need to test noise-toleration for the codes, especially for the
non-uniform noise that is an issue in the daily update synoptic maps.

5. Last not least, we need to work with SOLIS team to test the suggestions
in Jon's list (like better radial component, better polar field data,
etc), as well as issues above here. How do we start with?