﻿Cloud-J version 7.3c Notes
Michael Prather 18 Jul 2015

Fast-J began as 7-band, troposphere-only J-value code that could implement
full cloud and aerosol scattering (Wild et al., 2000 JAC). Later, 11 bands in
the ultraviolet were added as absorption-only to include stratospheric J-values
and called Fast-J2 (Bian and Prather, 2002 JAC).  By version 5.3 (circa 2005)
Fast-JX was created as a cross between the two codes and implemented full
scattering in all 18 bands so that the impact of a Pinatubo-like aerosol on
J-values could be calculated. With the implementation of Cloud-J version 7.3
(Neu et al., 2007; Prather, 2015), and the expected implementation of a
short-wave (solar) heating code, the X has been dropped from the Fast-J.


# typical makefile sequence for stand-alone Cloud-J
df -c fjx_cmn_mod.f90
df -c fjx_sub_mod.f90
df -c fjx_init_mod.f90
df -c cld_sub_mod.f90
df    cldj.f90 *.obj


HISTORY of Fast-J

Here is a short history of the Fast-J code as released, including its names
and capabilities.  I apologize for the shifting names, they evolved naturally
but it would have been much better to just have version numbers.


Fast-J
2000 (Wild, J.Atmos.Chem.) Fast-J code is invented, has 7 visible bands
spanning 294 to 850 nm.  Fast-J is intended to produce ONLY tropospheric
J-values.  The detailed paper shows the optimization over wavelength
bands and the comparison of 1-D (plane parallel) scattering codes
for cloud-sized particles carried out to 320 streams (angles)

Fast-J2
2002 (Bian, J.Atmos.Chem.) Requests to use Fast-J for stratospheric J-values
required a more careful wavelength optimization and 11 new bins.  This
wavelength binning has not been redone since Fast-J2 and thus all recent
Fast-J versions have 18 wavelength bins.  To avoid the cost of multiple scattering,
Fast-J2 used only attenuation by absorbing species in bins 1-11 (187 – 291 nm)
and full aerosol-cloud-Rayleigh scattering in bins 12-18 (291 – 850 nm).
It estimated that Rayleigh scattering could be approximated by taking 57% of
the scattering cross sections as absorbing.

Fast-JX
(dates from 2005, no pub, but see Neu, 2007)  In order to treat stratospheric
aerosol scattering (e.g., Mt. Pinatubo) It was necessary to do full scattering
down to 200 nm to get J-O2. Thus, Fast-JX is a cross between Fast-J and Fast-J2
that does full scattering at all wavelengths.
The first major version of Fast-JX was 5.3 (Jun 2005).  Version of Fast-JX
then proceeded through 5.7 and then 6.0 through 6.8 as .f (FORTRAN 77) versions.
A range of improvements involved adding semi-log layers to the tops of clouds
to improve the accuracy and conservation of cloud scattering, various attempts
at speeding up Fast-JX (no major breakthroughs here), added scattering profile
for clouds and aerosols at 200 nm, better interfaces with CTM, updated species
and cross sections for JPL-2010 including 2 or 3 temperatures, treatment of J-VOCs
via pressure-dependent yields, etc. (see Sovde et al, 2011 ACP)

Fast-JX FORTRAN 90 versions
(2013)  A .f90 module format of Fast-JX was developed for implementation in CESM CAM.
Thus Fast-JX versions 7.0, 7.1, and 7.2 were developed.

Fast-JX ver 7.2
(2013-14) Version 7.2, including full stand-alone coding of the Neu 2007 cloud
quadrature, is implemented in the UCI CTM. It also includes small code changes
and a new fjx_spec.dat data set that calculates J's only for >200 nm to allow
NCAR WACCM to use its lookup tables for <200 nm J's needed for altitudes >64 km.

Cloud-J ver 7.3      (should not be used)
(2015, Prather, GMDD) Cloud-J ver 7.3 is a stand-alone version of the cloud
correlation overlap developed in 2015.  Henceforth, the ‘X’ is being dropped from
Fast-J as it is unnecessary.  All new releases of the Fast-J core lying
underneath Cloud-J or of Cloud-J itself will just use ‘-J’.


Cloud-J ver 7.3b      (should not be used)
While in GMD discussion, minor coding errors involving the use of zero index
subscripts were discovered by colleagues and reviewers.  Also the compiler
differences were found regarding character parameters (all had to be exactly
the specified length) to avoid segmentation faults. These are now fixed with
version 7.3b, July 3 edition. The numerical results did not change.
A new version number was not assigned because this latter is a compiler issue.
Version 7.3b adopts a new fjx_spec.dat files, version 7.3c in preparation
for a merger of Cloud-J with the longer wavelength bins in RRTMG
so that solar heating calculations can be made. The two changes are:
    (i) The long wavelength bin #18 of Fast-J is now truncated at 778 nm
        The cross-sections and photons have been scaled simply to give same J's.
        When cross-sections are updated, the new range will be used (fjx_spec7.3b.dat).
    (ii) The solar fluxes (photons and Watts) and the Rayleigh cross-sections
        changed slightly with the new reference data sets and method of averaging.
        Changes in J's and any of the Cloud-J stats are <1% and usually <<1%.

Cloud-J ver 7.3c       (current best version)
While in GMD revision, it was recognized that MAX-COR groups separated by a
clear-sky group (each group is one decorrelation length) should have reduced
correlation with the cloudy layer above by an additional factor of cc.
This was implemented.  It changed the results by <0.1% rms difference and
a maximum difference (single atmosphere out of 640) of of ~1%.
The summary tables and figures in the submitted GMDD manuscript did not change.
The use of fjx_spec.dat (ver 73c).with Cloud-J ver 7.3c is recommended.
