GLOW version 0.981 release notes, Stan Solomon, 6/2017

Very minor update to v 0.98, implemented to facilitate using GNU gfortran:
   Changed all open statements to use action='read' instead of 'readonly'
   Commented out two lines from nrlmsise00.f; changed 0 to 0.
   Replaced sind and cosd functions in snoem.f90, and general cleanup.
   Corrected rcolum.f90 to call chap with pi-chi when chi>pi/2.
   Added a trap to etrans.f90 to assure no calculation when j=1.
   Zeroed out all arrays in cglow.f90.
   None of these changes should have any effect on results.

-----------------------------------------------------

GLOW version 0.98 release notes, Stan Solomon, 3/2017

See Quickstart.txt for very basic instructions
See Glow.txt for more details
See Glowlicense.txt for the Open Source Academic Research License Agreement

This is now a Fortran-90 code, with MPI and netCDF options
  No more common blocks.
  No more header file.
  Use-associated variables are defined in module cglow.f90
  Most code converted to lower-case free-form style
Example driver program glowbasic provides functionality in previous versions
  Can run one profile at a time or loop
  Single-processor, no MPI required
  No netCDF required either
  You can easily modify this program to suit particular applications
New driver program glowdriver uses TIE-GCM, TIME-GCM or empirical model inputs
  Ultimately this will include WACCM-X as well
  Fully MPI code with user-specified number of processors
  Namelist input from file in.namelist
  netCDF output to a file specified in the namelist input
Can still run with empirical models:
  Default empirical neutral model updated to MSIS-2K
  NOEM empirical nitric oxide model, based on SNOE data, used for empirical runs
  IRI-90 still used for empirical ionosphere

New features in v. 0.98:
  Most of the new things in v. 0.98 are computational, not scientific.
  Move to MPI parallel processing forced the elimination of common blocks.
  Reads netCDF input files from GCMs, and writes global netCDF output files.
  All of this is optional - you can still run the model the "old" way.
    (But old driver programs will have to be updated to use shared variables.)
  Empirical model calls are consolidated in subroutine MZGRID
  Calculates conductivities using Ryan McGranaghan's CONDUCT routine (adapted).
  N2 LBH, O 1356, N 1493, O 1304 added to airglow output (ZETA array).
    (Note 1493 is very approximate since branching rations are poorly known.)
  Radiative recombination and mutual neutralization included in 1356.
  Subroutine BANDS added to allocate LBH among v' levels.
    (Really just a stub now, will ultimately handle vibrational distributions.)

Note that several bugfixes and parameter adjustments were made following the
  "beta" release of v. 0.98 in early January 2017, so do not use that version.

Other significant changes:

   The quartic equation solver formerly used by GCHEM to calculate electron
density has been abandoned because it produced minor errors and could not
be verified.  Instead, a simple iterative method is used, which turns out
to be more accurate and equally fast.  Thanks to Michael Hirsch for identifying
this problem.

   The high-energy electron precipitation example (hexexample) is no longer
provided.  This functionality is built-in to the codes, but if you want to
do a high-energy electron run you will have to increase NBINS in cglow.f90
(e.g., to 343 for energies up to 100 MeV) and extend the altitude grid to
lower altitude by increasing JMAX in the driver program, and adding levels
in mxgrid.f90.

   Conjugate hemisphere photoelectrons are no longer provided in the example
drivers, since they are very minor, triple the run time, and weren't widely
used.  They can easily be re-installed in a driver program, but note that
the magnetic field needs to be updated.  The calculation is fairly approximate
anyway because it doesn't consider exosphere/plasmasphere attenuation of
the electron flux.

   The background ("night time") ionization rate calculated in QBACK was updated
to conform to the forumlation now used in the TIE-GCM, TIME-GCM, WACCM, and
WACCM-X.  It is still pretty approximate, but produces an night ionosphere in
the E and F1 regions that is in reasonable agreement with observations.

   Several unused artifacts were eliminated, but arrays PIA and SESPEC were
retained in anticipation of including proton aurora ionization in the future.

Known issues (most of these carried over from v. 0.97):
  X-rays shortward of 18 A need to be re-examined and updated.
  Magnetic field (GEOMAG, FIELDM) is out of date.  GEOMAG is only used by
    the NOEM model, and isn't too different from the coordinate transform
    used to generate SNOE data on a magnetic grid.  FIELDM only calculates
    the dip angle, which has a minimual effect on results.  But these should
    be updated to time-varying IGRF/Apex coordinates.
  Temporary Y2K fix to SUNCOR.  Also should update to J2000 epoch.
    Should be fine for 1950-2050, but really need to make if valid for either
    yyddd to yyyyddd date format, which will extend the range of validity to
    1900-2100.  (This only affects SZA.)
  O(1S) needs to be re-evaluated (still).  Nightglow recombination source
    still not included.  Also need to include O2 A-bands.
  O(1S) from O2 dissociation (BSO2) is currently hardwired - only works with
    LMAX=123 (although changes shortward of 800 A are OK).
  IRI should be updated.  Usually, IRI is Only used for electron density above
    200 km, and IRI hasn't changed much in the F-region, so it's OK for now.
  Cascade contributions to 7774 and subsequent cascade to 1356 is questionable.
    I reduced the effective 7774 cross section, so 7774 and 1356 are now
    self-consistent, and in reasonable agreement with GUVI data.  
  1493 sources and branching ratios are speculative.  For now, I am presuming
    that it is produced during photodissociative ionization and electron
    impact dissociative ionization of N2.

-------------------------------------------------

Version 0.973 release notes, Stan Solomon, 3/2015

Version 0.973 is an incremental release of GLOW, mostly just the example drivers
  updates example drivers to MSIS-2K
  adds NOEM empirical nitric oxide model, based on SNOE data, to example drivers
  some cleanup of example drivers
  fixs ssflux so it only reads file on first call (or if ISCALE changes).

-------------------------------------------------

Version 0.97 release notes, Stan Solomon, 4/2005

New features:
   Relativistic correction to electron impact cross sections included
   Maxwellian or monoenergetic fluxes generated by MAXT
   Possible to use any solar grid by changing only input files
   Photoabsorption and photoionization cross sections supplied in files
   SSFLUX completely re-written:
      Model parameters supplied in files
      Default is ~1 nm grid (5 nm in FUV)
      Hinteregger model still there (ISCALE=0)
      EUVAC also available (ISCALE=1)
      User grid and input supported by changing input file and LMAX (ISCALE=2)
   Common block CGLOW is unchanged
      Should facilitate upgrade path for existing programs
      But there are now several obsolete artifacts

Issues addressed in v. 0.97:
   Fixed two problems with Auger electron production
   Fixed some small bugs in O(1S) calculation
   Adjusted N(2D)+O rate coefficient to Fell et al. value (6.9e-13)
   Adjusted C III on N2 cross section to fix O2 ionization rate problem
   Removed various artifacts, including EAURI
   Removed unnecessary N(2D) initial guess (now just set to zero)
   Now use standard energy and altitude grid in both day and aurora examples

Known issues:
   X-rays shortward of 18 A need to be re-examined and updated.
   Magnetic field (GEOMAG, FIELDM) is out of date, really need to update to 
      IGRF, but at the resolutions typical here should be OK for now.
   Temporary Y2K fix to SUNCOR.  Should be fine for 1950-2050, but really
      need to change from yyddd to yyyyddd date format, which will enable
      range of validity to extend from 1900-2100.  (This only affects SZA.)
   O(1S) needs to be re-evaluated (still).
   O(1S) from O2 dissociation (BSO2) is currently hardwired - only works with
      LMAX=123 (although changes shortward of 800 A are OK).
   IRI should be updated.
