RRTMG_SW What's New:

Comparison results of latest version of RRTMG_SW relative to RRTM_SW.

RRTMG_SW v5.0

Date: March 2020

• Added several new features and fixes: Two new cloud overlap options were added, exponential (icld=4) and exponential-random (icld=5) that apply an exponential transition from maximum to random through model layers. Exponential applies this transition through all layers without regard to the cloud configuration and exponential-random applies this transition through adjacent cloudy layers while randomly correlating cloud layers separated by clear sky. Both methods require calling a new subroutine, get_alpha, to define alpha, the exponential transition parameter, prior to calling the McICA sub-column generator. Alpha requires the specification of a decorrelation length (in meters), and two methods for defining this value are available that are selected using a new input control variable, idcor, that allows the decorrelation length to be specified either as a constant or as a value that varies by latitude and day-of-year using the method of Oreopolous et al. (2012). See the file doc/rrtmg_sw_instructions.txt for more information.


• The three NRLSSI2 solar source component coefficients (quiet sun, facular, and sunspot) added in v4.0 and updated in v4.1 were added to the optional netCDF input data file (/data/rrtmg_sw.nc), and the optional source module rrtmg_sw_read_nc.f90 was modified to input these data if the netCDF input option is selected in the makefile. These data function in the same way as the identical NRLSSI2 solar source coefficients that are stored in the code by default in module rrtmg_sw_k_g.f90.


• A fix was applied to source module rrtmg_sw_read_nc.f90, which optionally reads the input data coefficients from /data/rrtmg_sw.nc if that option is selected in the makefile, to correct the loading of the water vapor foreign continuum data for the upper atmosphere into the model array forrefo. This issue could impact surface downward shortwave fluxes by as much as five percent if the netCDF option was used. The default input coefficient data in rrtmg_sw_k_g.f90 were not affected.


• Revised the calculation of zwo in rrtmg_sw_reftra.f90 to protect against unlikely, but possible divide by zero conditions.

RRTMG_SW v4.1

Date: February 2020

• Added two fixes that impact solar absorption: The three solar source component coefficients (quiet sun, facular, and sunspot) added in v4.0 were regenerated by g-point in all 14 spectral bands from high-resolution data using the latest model and line file resources (LBLRTM_v12.9, LNFL_v3.2, and aer_v_3.7) with Rayleigh scattering deactivated. The effect on results from this change is most significant in band 25 in which the heating rate near 5 hPa is increased by as much as 0.25 K/day and the upward flux and the top of the atmosphere is reduced by up to 1-2 Wm-2. The impact in other bands is minor.


• The reference layer LAYREFFR was corrected from 58 to 42 in band 28 in module rrtmg_sw_taumol.f90 to more accurately represent the absorption in that band. The net effect of this change is to increase the heating rate by up to 0.1 K/day near 1 hPa.

RRTMG_SW v4.0.2

Date: February 2018

• Added several fixes to the solar variability capability added in v4.0. The tables of facular and sunspot indices for the mean 11-year solar cycle were extended from 132 to 134 points to cover the full period from the first day of year 1 to the last day of year 11. The previous set of 132 points (now surrounded by two new end points) represent the middle date of the 12 months of the 11 year cycle (132 months). To use these indices for the mean solar cycle (isolvar=1), the user provides a fraction (solcycfrac) representing the position of the current time within the 11-year cycle from 0.0 (first day of year 1) to 1.0 (last day of year 11).


• In the subroutine version of the code, the optional capability to scale the mean solar cycle was corrected to be limited to input option isolvar=1.


• In the column version of the code, the input of the time-specific facular and sunspot indices through variable solvar was corrected for input option isolvar=2.


• In the subroutine version of the code several instances of variables tmp_a_0 and tmp_b_0 were corrected to tmp_f_0 and tmp_s_0.


• In the subroutine version of the code the handling of optional input variables bndsolvar, indsolvar and solcycfrac was improved.

RRTMG_SW v4.0.1

Date: November 2017

• Removed the "/src/src_link" script that previously linked to the RRTM versions of rrtatm.f and extra.f. Current versions of those files as of this release have been added to this repository in the /src directory. The values of MXLAY (603) and MXMOL (39) were also updated to correspond with those now used in rrtatm.f. Removed the "/src/util_link" script that previously linked to the utility files and relocated these to a new /aer_rt_utils folder that is tied to ~/RD/AER_RT_UTILS/trunk in the svn repository as an external. All provided makefiles have been updated to get the utility files from the new directory.

RRTMG_SW v4.0

Date: December 2016

• Added new solar variability options based on NRLSSI2 model. The solar variability method is selected with input flag isolvar and options include using the internal solar constant consistent with the NRLSSI2 model (1360.85 Wm-2) with no solar cycle variability, applying the mean of Solar Cycles 13-24 to represent averaged facular brightening and sunspot blocking effects over time, or supplying specific facular and sunspot indices to represent solar variability over specific solar cycles over time. Options that use the averaged solar cycle can be scaled to a different solar constant. Additional capability is provided to adjust the amplitude of the averaged solar cycle or to scale the solar constant by spectral band. The Kurucz extraterrestrial solar irradiance specification that was used in RRTMG_SW_v3.91 and earlier remains an available option.

RRTMG_SW v3.9.1

Date: September 2016

• All source code modules were modified to replace the original copyright statement with a new BSD 3-clause software license (at the request of ECMWF). Additional minor changes include: - Changed declaration of variable cinp in subroutine readprof in file rrtmg_sw.1col.f90 from implicit real to character*3. - Fixed integer kind type problem in file rrtmg_sw_read_nc.f90 by replacing all numeral "1" arguments to nf90_get_var to "1_im" to prevent errors on some compilers. - Fixed integer kind type problem in file mcica_random_numbers.f90 by replacing several numeral "1" arguments to the function "mod" with "1_im" at lines 199 and 220 to prevent errors on some compilers.

RRTMG_SW v3.9

Date: December 2013

• Look up tables of extinction coefficients, single scattering albedos, and asymmetry parameters were regenerated and replaced for the Hu and Stamnes (LIQFLAG=1) liquid cloud option to remove small discontinuities in prior versions of these tables. The new tables were derived using Mie calculations at finer spectral resolution than the original method. The updated tables are consistent with those used in the version of this model applied to the CIRC (Continual Intercomparison of Radiation Codes) program. The impact on fluxes is of order 0.5 Wm-2 or less in profiles with liquid clouds.


• In the callable subroutine version of the code, the aerosol input option flag (iaer) was added to the subroutine call list.


• A correction was applied to the ECMWF aerosol input option (iaer=6) to fix the initialization and specification of the aerosol optical properties for this method. The primary aerosol input option of ingesting the aerosol optical properties directly was unaffected.


• Instances of the 'float' intrinsic function in the code were replaced with the 'real' function in several source modules for consistency with the specified real kind type used throughout the code.

RRTMG_SW v3.8

Date: May 2009

• New option added to input absorption coefficient data either through original source statements or through a netCDF data file. Choice of input source is selected in the makefile. This feature was developed and contributed by Patrick Hofmann and Robert Pincus of NOAA.


• The lower bound of allowable liquid droplet effective radius was changed from 1.5 to 2.5 microns for the inflag=2/liqflag=1 cloud input option.

RRTMG_SW v3.7

Date: February 2009

• Modified the value of the single scattering albedo threshold (from 0.9995 to 0.9999995) that selects between the general 2-stream scattering solution (for most SSA values) and a simpler, conservative scattering solution when SSA is close to 1.0. The new more restrictive threshold generates improved results within multi-layer, highly scattering clouds when the SSA is just under one. This change will affect fluxes and will significantly affect heating rates within highly scattering clouds, especially those that extend over multiple layers. Negligible changes also occur in clear sky since the single scattering albedo can be close to this threshold when Rayleigh scattering dominates the shortwave extinction. Use of this modification is strongly recommended.


• Added several code changes to the two-stream solver that prevent potential exception problems under single precision.


• A new flag was installed (idelm) that is used to select between either unscaled 'true' direct and diffuse fluxes (idelm=0) or delta-scaled direct and diffuse fluxes (idelm=1). This flag is read from the input file in the column version of the model. In the code to be used in GCMs, it is set by default to 1 to provide scaled direct and diffuse fluxes. Total fluxes continue to be provided only as delta-scaled quantities.


• Added a new makefile to the build directory for IBM XL platforms.


• Updated the rrtmg_sw_instructions.txt file in the doc directory.

RRTMG_SW v3.61

Date: September 2008

• Modified input of ice particle effective size to require input of value appropriate for the selected ice optics parameterization. Previous scaling that was applied internally to convert ice effective radius to generalized effective size for the Fu parameterization has been removed. Any necessary scaling needed to provide the proper ice particle size must be done before input to the radiation in the GCM version of the code. This is consistent with the input requirement for the column version of the model. Note that the internal ice optics parameterizations are effective over specific ranges of particle size. Results may be different from previous versions if the Fu parameterization is used in the GCM version of the code.

RRTMG_SW v3.6

Date: August 2008

• Added requirement that cloud liquid and ice water paths, or optionally the cloud optical depth, be input as in-cloud averages rather than grid-averaged values in the version of the code intended for use in a GCM. This change requires that these cloud properties be scaled by the cloud fraction on input to convert to in-cloud values if necessary. The column model version of the code continues to expect in-cloud values as before. Oxygen amount was changed from being defined internally to a required input. The forward scattering fraction for clouds was added as an input array. Several other cosmetic modifications do not affect results, but a change in the factor used to convert flux to heating rate from a fixed to a calculated quantity may produce negligible differences in heating rates.

RRTMG_SW v3.5

Date: January 2008

• Corrected calculation of cloud optical property stochastic arrays in McICA sub-column generator when cloud optical properties are input directly. This change will impact calculations with clouds when McICA is used and input parameter inflag is set to 0.

RRTMG_SW v3.4

Date: September 2007

• Modified use of ice particle size variables rei/reicmc to utilize separate variable names (dge/dgesmc) when units of generalized effective size are required for Fu ice parameterization (iceflag=3) option. Results may be minimally affected when ice particle radii exceed 90 microns and the iceflag=3 option is used. Also modified several local arrays in top-level subroutine module to accomodate an extra model layer at the top of the atmosphere if this feature is utilized when using the code in a GCM.

RRTMG_SW v3.3

Date: September 2007

• Modified use of column dimension in some optional output arrays in subroutine version of model, and corrected intent attribute of icld flag. Also added solar constant as input value into subroutine version of model to allow scaling of solar constant from default value. Results are identical to the previous version.

RRTMG_SW v3.2

Date: August 2007

• Includes changes to the makefile for AIX systems and a few minor fixes in the interface for the subroutine version of the model. Results are identical to the previous version.

RRTMG_SW v3.1

Date: August 2007

• Includes several source code format changes that have no affect on results.


• Added exponential lookup table for transmittance, similar to that used in RRTMG_LW, to improve computational performance. Improvement in speed is machine dependent, but it is typically 5-10 percent. Slight changes in results are possible, but they are generally negligible.

RRTMG_SW v3.0

Date: April 2007

• Major source code reformatting to utilize modern FORTRAN90 features such as modules. Older common blocks have been removed from all but I/O related source files to retain I/O compatibility with RRTM.


• Model can be run as a stand-alone column model using the same I/O format as RRTM and LBLRTM, or it can be used as a callable subroutine and implemented within a global or regional model.


• McICA (Monte Carlo Independent Column Approximation) capabililty added for representation of sub-grid cloud variability such as cloud fraction and cloud overlap.


• Includes two options for the random number generator needed for McICA.


• Sample makefiles for several computer platforms are provided.


• Sample input and output files and instructions for the input files are also included in the source package.