Thanks to Dave MacEwan and Peter Girard for the great ideas.  I believe
Peter's method for determining the sunrise/ sunset times would produce
exactly what I need.  I just started to formulate an outline for the rather
involved code it would require, when I received an email from the Solar
Analyst programmer that he could help us on a special contract to modify
the model to output the sunrise / sunset times directly.  That is the most
straight-forward solution, so we are looking into that.  I didn't get any
responses from anyone saying they had used Solar Analyst.  - Stacey
(original question is below)

-- from Dave MacEwan
Check out the US Naval Observatory's Astronomical Applications Department
web page at http://aa.usno.navy.mil/.  Under "Data Services" there are
applications that will give you tables of sunrise/sunset times and other
stuff.  Maybe you will find this useful.

--from Peter Girard
Have a look at the JavaScript code that supports this web page:
http://www.srrb.noaa.gov/highlights/sunrise/azel.html
It computes the position of the sun for any time and location and includes
the correction for atmospheric refraction.  The mathematics for this come
primarily from "Astronomical Algorithms" by Jean Meuss, worth getting if
you
will need to deal with this in some depth.

In general, this is how I would go about solving the problem:
1)  Use the algorithm to calculate a series of azimuth/altitude
    points for the time around sunrise/sunset.
2)  Using the range of azimuths from step 1, sample the surface
    model along that part of the horizon, generating a second
    set of azimuth/altitude points.
3)  Treating both sets of points as polylines, find where they
    intersect.  This should be the approximate azimuth/altitude
    on the horizon where the sun rises/sets.  Interpolate the
    approximate time when the sun is at this point.
4)  Repeat steps 1-3 using finer sampling to get a more
    precise time.
The performance bottleneck will probably be in the horizon calculations
from
the surface model.  The rest should go quite quickly on a modern machine.

ESRI provides a sample extension called Visibility Tools
(vistool.avx and .apr) that contains a Line of Sight tool.  When you
digitize a line, it symbolizes which parts of the line are visible from the
first point to the second.  If the second point is sufficiently far from
the
first, the last visible point along this line would be the horizon (as long
as it's not the edge of the grid, of course).  The main script in this
extension is "SA: Line of Sight Profile", and it uses the
Grid.CellValueList
request to get the visible points along the line.

Direct sampling of elevations off a grid does not take into account the
curvature of the Earth or atmospheric refraction.  You may need to figure
these into the line of sight analysis if you are dealing with distant
mountains.

--original question.
Hi.  I'm wondering if anyone out there has used Solar Analyst or TopoView
(Hemisoft,Inc).  I have been using it quite extensively and have developed
several scripts for batch processing.  I have been in contact with the
Solar Analyst developers, but have been unsuccessful getting answers to a
couple of basic questions:

How to determine the duration interval starting times automatically -- we
need to know sunrise and sunset times precisely.   Also, how do you get
duration in partial hour output (instead of rounded to whole hours) if your
output is in GRID format (is the model set up for integer grids only?).
If you have any ideas, or just have experience with Solar Analyst or
TopoView, I'd be interested in hearing from you.   Thanks!!

Stacey L. Stark
stark.stacey@epa.gov
GIS Analyst, CSC
Mid-Continent Ecology Division, EPA
6201 Congdon Blvd.
Duluth, MN  55804
(218) 529-5214