About using MSSP-2 interface
-----------------------------
This interface allows users to create one, or a series of,
spectral plots of energetic particle fluxes of one or more
of 15 species (H to Ni) from one or more data sources.

Users may scan the available data as time series plots to
find intervals of special interest by clicking on a link
in the "Data source" column.  This yields
an OMNIWeb-Plus plot/list interface for the selected data set.

Users specify for which data set - species combinations they 
want to generate spectral plots by clicking on the relevant 
matrix elements having enclosed boxes.  (No box means no data
for that source and species combination.)
Relative to uniqueness of color and symbol combinations used
in plotting, limiting one's selections to 6 or less works best, but
is not a hard requirement.

Users specify the time span over which the fluxes to be shown 
in the first spectral plot will be created.  Note also that data
from the end day specified will be included in the flux averaging.

Users may ensure that all plots of a given series of plots
have a common Y-axis range by specifying a minimum and
maximum flux in the "Y scale range" option near the bottom
of the main interface page.  Otherwise IDL will fix the scale
range for each plot according to the data range on that plot.
(Bottom of the range should not be less than 1.0e-6.)

Data are held at hourly and daily time resolutions for various 
data sets.  Users may create, on the fly, spectral plots for fluxes 
at daily or any lower resolution.  Hourly resolution plots of 
H and He fluxes are possible from the companion interface at
http://omniweb.sci.gsfc.nasa.gov/ftpbrowser/flux_spectr_m.html

The horizontal error bars show energy bins, while the vertical 
error bars show standard deviations in flux averages (see 
discussion of averaging below).  Since we plot spectra on a 
log-log scale, we fix a minimum flux value for plotting at 1.0E-06.  
No fluxes < 1.0E-06 are plotted.  However, computed average values 
of fluxes and their standard deviations are listed via the "Listing 
of spectral function points ..." option on the plot pages. 
Having created the first spectral plot, a user may generate similar
plots for contiguous intervals of the same duration by clicking
on the "Previous" or "Next" button of the spectral plot page.
Clicking on the "Home" button returns the user to the default
version of the interface page.

The output page with the spectral plot offers the option to list 
digital values associated with the plots, including energy bin 
minimum and maximum values and geometric means (where spectral 
points are plotted), the average fluxes and their standard 
deviations, and the numbers of individual fluxes in the flux 
averages.

The spectral points for each source for each output plot are
fitted to a power law, F = A * E ** b, where F and E are flux
and energy, and where A and b are the fit parameters.
A and b are determined by a linear least squares fit to
the relation Log(F) = Log(A) + b * Log(E).
The A and b values are also given on the digital output page.
Zero fluxes are omitted from the fitting.  Thus it is recommended
that times of averaging be long enough that each energy bin
has a non-zero average flux to yield meaningful power law fits
across the full energy range

The interface page's "Optional specifications" allow users
to specify just a subset of the energy range of the data to be
displayed and fitted.  This may be useful if the full-energy-range
spectral plots show a nearly straight line over only part of the 
full energy range. 

Computing the lower resolution fluxes -

Ideally, one should determine a long-term flux using the total
number of relevant counts over the interval and the total 
time of counting over the interval.  However, since at least 
some of the data sets we hope to support with this interface 
provide fluxes and flux uncertainties but not all the counts 
and observing times for each such flux, we cannot determine 
longer term fluxes this way.  Accordingly, we take simple 
linear flux averages and their standard deviations to 
determine longer term fluxes and their uncertainties.

Note on plotted vertical error bars

To avoid the down legs of vertical error bars frequently
reaching to the bottom of the plot on our log scale, we fix the
bottom of the vertical error bar at F = Fmin = <F>**2/(<F>+sigma).
Thus the error bar goes up from Fmin to Fmax (=<F>+sigma),
with <F> half way up the error bar on the log scale.  We've used
Fmax/<F> = <F>/Fmin.  The numeric value of the linearly computed 
standard deviation, sigma, is given via the "Listing of spectral 
function points..." option on the plot page.
-------------------------------------------------------------------

This interface and its capabilities were developed by Natalia
Papitashvili and Joe King as part of the Virtual Energetic Particle
Observatory (VEPO) activity, John F. Cooper, Principal Investigator,
carried out within NASA/Goddard's Space Physics Data Facility.
-------------------------------------------------------------------