aer-rc / rrtmg_lw

RRTM for GCMs, LongWave domain

RRTMG_LW: Longwave Radiative Transfer Model for GCMs

---

Contents

1. Introduction
2. Releases
3. Column Version
4. GCM Version
5. Contact
6. References

Introduction

This package contains the source code and sample makefiles necessary to run the latest version of RRTMG_LW, a correlated k-distribution longwave radiative transfer model developed at AER for application to GCMs. This version of RRTMG_LW has been modified from the standard RRTM_LW distributed by AER to enhance its performance for use within general circulation models. This code has also been modified to utilize updated FORTRAN coding features. Two modes of operation are possible:

1. RRTMG_LW can be run as a column model using the input files and source modules described in the column version section, or
2. it can be implemented as a subroutine into an atmospheric general circulation model or single column model.

The version of RRTMG_LW provided here has been modified from the standard RRTM_LW to enhance performance with little effect on the accuracy. The total number of g-points used has been reduced from 256 to 140. Fluxes are accurate to within 0.5 W m^-2 and cooling rate within 0.1 K day^-1 relative to the standard RRTM_LW, which is itself accurate to within 1 W m^-2 of the data-validated line-by-line radiative transfer model, LBLRTM. Required absorption coefficient input data can be read in either from data stored within the code or from an external netCDF file as selected in the makefile.

This model can also utilize McICA, the Monte-Carlo Independent Column Approximation, to represent sub-grid scale cloud variability such as cloud fraction and cloud overlap. If the McICA option is selected to model a cloudy profile in column mode, then the model will run stochastically, and the output fluxes and heating rates will be an average over 200 samples. In GCM mode, the code will calcualte a single column per profile, and the statistical basis is provided by the spatial and temporal dimensions of the 3-D calculations. Several cloud overlap methods are available for partial cloudiness including maximum-random, exponential, and exponential-random.

The model includes an optional feature to provide simultaneously with a normal forward calculation the change in upward flux with respect to surface temperature for each model level. This option is controlled by the input flag, idrv. Setting this flag to 1 will output dF/dT for total sky and clear sky in GCM mode in new output arrays duflx_dt and duflxc_dt. These can be utilized to approximate the change in upward flux for a change in surface temperature only at time intervals between full radiation calls. In single column mode, setting idrv to 1 requires the extra input of a dT change in surface temperature relative to the input surface temperature, and the provided dT will be applied to the flux derivative to output a modified upward flux profile for that dT change in surface temperature. The default idrv setting of 0 provides the original forward radiative transfer calculation.

For more information on the model, see the Wiki Description Page

Releases

Version 5.0 is the latest version of the model

Releases before Version 5.0 are not publicly available.

RRTMG_LW : Column Version

DOCUMENTATION

The following text files (in the doc directory), along with this README provide information on release updates and on using and running RRTMG_LW:

Filename	Description
release_notes.txt	Code archive update information
rrtmg_lw_instructions.txt	Input instructions for files INPUT_RRTM, IN_CLD_RRTM and IN_AER_RRTM

SOURCE CODE

The following source files (in the src directory) must be used to run RRTMG_LW in stand-alone mode as a column model (the utility files are stored separately in the aer_rt_utils directory):

Filename	Description
rrtmg_lw.1col.f90	RRTMG_LW main module
rrtmg_lw_cldprop.f90	Calculation of cloud optical properties
rrtmg_lw_cldprmc.f90	Calculation of cloud optical properties (McICA)
rrtmg_lw_init.f90	RRTMG_LW initialization routine; reduces g-intervals from 256 to 140
rrtmg_lw_k_g.f90	Absorption coefficient data file
rrtmg_lw_read_nc.f90	Optional absorption coefficient data netCDF input
rrtmg_lw_rtrn.f90	Calculation of clear and cloudy radiative transfer using random cloud overlap
rrtmg_lw_rtrnmr.f90	Calculation of clear and cloudy radiative transfer using maximum-random cloud overlap
rrtmg_lw_rtrnmc.f90	Calculation of clear and cloudy radiative transfer using McICA (with user-selectable overlap method)
rrtmg_lw_setcoef.f90	Set up routine
rrtmg_lw_taumol.f90	Calculation of optical depths and Planck fractions for each spectral band
mcica_random_numbers.f90	Random number generator for McICA
mcica_subcol_gen_lw.1col.f90	Sub-column generator for McICA
rrtatm.f	Process user-defined input data files
extra.f	Process input data files
util_**.f	Utilities (available for multiple platforms)

The following module files (in the modules directory) must be used to run RRTMG_LW in stand-alone mode as a column model (these must be compiled before the source code files):

Filename	Description
parkind.f90	real and integer kind type parameters
parrrtm.f90	main configuration parameters
rrlw_cld.f90	cloud property coefficients
rrlw_con.f90	constants
rrlw_kg**.f90	absorption coefficient arrays for 16 spectral bands
rrlw_ncpar.f90	parameters for netCDF input data option
rrlw_ref.f90	reference atmosphere data arrays
rrlw_tbl.f90	exponential look up table arrays
rrlw_vsn.f90	version number information
rrlw_wvn.f90	spectral band and g-interval array information

INPUT DATA

The following file (in the data directory) is the optional netCDF input file containing absorption coefficient and other input data for the model. The file is used if keyword KGSRC is set for netCDF input in the makefile.

Filename	Description
rrtmg_lw.nc	Optional netCDF input data file

MAKEFILES

The following files (in the build/makefiles directory) can be used to compile RRTMG_LW in stand-alone mode as a column model on various platforms. Link one of these into the build directory to compile.

Filename	Description
make_rrtmg_lw_sgi	Sample makefile for SGI
make_rrtmg_lw_sun	Sample makefile for SUN
make_rrtmg_lw_linux_pgi	Sample makefile for LINUX (PGI compiler)
make_rrtmg_lw_aix_xlf90	Sample makefile for AIX (XLF90 compiler)
make_rrtmg_lw_OS_X_g95	Sample makefile for OS_X (G95 compiler)
make_rrtmg_lw_OS_X_ibm_xl	Sample makefile for OS_X (IBM XL compiler)

SAMPLE INPUT/OUTPUT

Several sample input and output files are included in the run_examples_std_atm directory. Note that user-defined profiles may be used for as many as 200 layers.

Filename	Description
INPUT_RRTM	Required input file for (clear sky) atmospheric specification
IN_CLD_RRTM	Required input file for cloud specification if clouds are present
IN_AER_RRTM	Required input file for aerosol specification if aerosols are present
OUTPUT_RRTM	Main output file for atmospheric fluxes and heating rates
input_rrtm_ICRCCM_sonde	Sample radiosonde-style input profile for clear sky
input_rrtm.MLS-clr	Sample 51 layer mid-latitude summer standard atmosphere for clear sky
input_rrtm.MLS-cld-imca0-icld2	Sample 51 layer mid-latitude summer standard atmosphere with cloud flag turned on and maximum-random cloud overlap selected (without McICA)
input_rrtm.MLS-cld-imca1-icld2	Sample 51 layer mid-latitude summer standard atmosphere with cloud flag turned on and maximum-random cloud overlap selected (with McICA)
input_rrtm.MLS-cld-imca1-icld4-idcor0	Sample 51 layer mid-latitude summer standard atmosphere with cloud flag turned on and exponential cloud overlap and constant decorrelation length selected (with McICA)
input_rrtm.MLS-cld-imca1-icld5-idcor0	Sample 51 layer mid-latitude summer standard atmosphere with cloud flag turned on and exponential-random cloud overlap and constant decorrelation length selected (with McICA)
input_rrtm.MLS-cld-imca1-icld5-idcor1	Sample 51 layer mid-latitude summer standard atmosphere with cloud flag turned on and exponential-random cloud overlap and varying decorrelation length selected (with McICA)
input_rrtm.MLS-clr-aer12	Sample 51 layer mid-latitude summer standard atmosphere with aersol flag set
input_rrtm.MLS-clr-xsec	Sample 51 layer mid-latitude summer standard atmosphere with cross-section input (CFCs, etc.)
input_rrtm.MLS-clr-idrv1	Sample 51 layer mid-latitude summer standard atmosphere with derivative option set to provide modified upward fluxes for the provided change in surface temperature
input_rrtm.MLW-clr	Sample 51 layer mid-latitude winter standard atmosphere
input_rrtm.SAW-clr	Sample 51 layer sub-arctic winter standard atmosphere
input_rrtm.TROP-clr	Sample 51 layer tropical standard atmosphere
in_cld_rrtm-cld5	Sample cloud input file
in_cld_rrtm-cld7	Sample cloud input file
in_aer_rrtm-aer12	Sample aerosol input file
`script.run_std_atm	UNIX script for running the full suite of example cases, which will put the output into similarly named files prefixed with output_rrtm*

INSTRUCTIONS FOR COMPILING AND RUNNING THE COLUMN MODEL

1. In the build directory, link one of the makefiles from the makefile sub-directory into build/make.build. To use the optional netCDF input file, switch the keyword KGSRC in the makefile from dat to nc. Compile the model with make -f make.build
2. Link the executable to, for example, rrtmg_lw in the run_examples_std_atm directory
3. If the optional netCDF input file was selected when compiling, link the file data/rrtmg_lw.nc into the run_examples_std_atm directory.
4. In the run_examples_std_atm directory, run the UNIX script ./script.run_std_atm to run the full suite of example cases. To run a single case, modify INPUT_RRTM following the instructions in doc/rrtmg_lw_instructions.txt, or copy one of the example input_rrtm* files into INPUT_RRTM. If clouds are selected (ICLD > 0), then modify IN_CLD_RRTM or copy one of the in_cld_rrtm* files into IN_CLD_RRTM. If aerosols are selected (IAER > 0), then modify IN_AER_RRTM or set it to the sample file in_aer_rrtm-aer12.
5. In column mode, if McICA is selected (IMCA=1) with partial cloudiness defined, then RRTMG_LW will run the case 200 times to derive adequate statistics, and the average of the 200 samples will be written to the output file, OUTPUT_RRTM.

RRTMG_LW : GCM version

SOURCE CODE:

The following source files (in the src directory) must be used to run RRTMG_LW as a callable subroutine:

Filename	Description
rrtmg_lw_rad.f90	RRTMG_LW main module (with McICA)
rrtmg_lw_rad.nomcica.f90	Optional RRTMG_LW main module (without McICA only)
rrtmg_lw_cldprop.f90	Calculation of cloud optical properties
rrtmg_lw_cldprmc.f90	Calculation of cloud optical properties (McICA)
rrtmg_lw_init.f90	RRTMG_LW initialization routine; reduces g-intervals from 256 to 140; (This has to run only once and should be installed in the GCM initialization section)
rrtmg_lw_k_g.f90	Absorption coefficient data file
rrtmg_lw_read_nc.f90	Optional absorption coefficient data netCDF input
rrtmg_lw_rtrn.f90	Calculation of clear and cloudy radiative transfer using random cloud overlap
rrtmg_lw_rtrnmr.f90	Calculation of clear and cloudy radiative transfer using maximum-random cloud overlap
rrtmg_lw_rtrnmc.f90	Calculation of clear and cloudy radiative transfer using McICA (with selectable overlap method)
rrtmg_lw_setcoef.f90	Set up routine
rrtmg_lw_taumol.f90	Calculation of optical depths and Planck fractions for each spectral band
mcica_random_numbers.f90	Random number generator for McICA
mcica_subcol_gen_lw.f90	Sub-column generator for McICA (must be called in GCM just before call to RRTMG)

NOTE: Only one of rrtmg_lw_k_g.f90 or rrtmg_lw_read_nc.f90 is required.

The following module files (in the modules directory) must be used to run RRTMG_LW as a callable subroutine (these must be compiled before the source code)

Filename	Description
parkind.f90	real and integer kind type parameters
parrrtm.f90	main configuration parameters
rrlw_cld.f90	cloud property coefficients
rrlw_con.f90	constants
rrlw_kg**.f90	absorption coefficient arrays for 16 spectral bands
rrlw_ncpar.f90	parameters for netCDF input data option
rrlw_ref.f90	reference atmosphere data arrays
rrlw_tbl.f90	look up table arrays
rrlw_vsn.f90	version number information
rrlw_wvn.f90	spectral band and g-interval array information

INPUT DATA:

The following file (in the data directory) is the optional netCDF file containing absorption coefficient and other input data for the model. The file is used if source file rrtmg_lw_read_nc.f90 is used in place of rrtmg_lw_k_g.f90 (only one or the other is required).

Filename	Description
rrtmg_lw.nc	Optional netCDF input data file

NOTES ON RUNNING THE GCM (SUBROUTINE) VERSION OF THE CODE:

1. The module rrtmg_lw_init.f90 is the initialization routine that has to be called only once. The call to this subroutine should be moved to the initialization section of the host model if RRTMG_LW is called by a GCM or SCM.
2. The number of model layers and the number of columns to be looped over should be passed into RRTMG_LW through the subroutine call along with the other model profile arrays.
3. To utilize McICA, the sub-column generator (mcica_subcol_gen_lw.f90) must be implemented in the GCM so that it is called just before RRTMG_LW. The cloud overlap method is selected using the input flag, icld. If either exponential (ICLD=4) or exponential-random (ICLD=5) cloud overlap is selected, then the subroutine get_alpha must be called prior to calling mcica_subcol_lw to define the vertical correlation parameter, alpha, needed for those overlap methods. Also for those methods, use the input flag idcor to select the use of either a constant or latitude-varying decorrelation length. If McICA is utilized, this will run only a single statistical sample per model grid box. There are two options for the random number generator used with McICA, which is selected with the variable irnd in mcica_subcol_gen_lw.f90. When using McICA, then the main module is rrtmg_lw_rad.f90. If McICA is not used, then the main module is rrtmg_lw_rad.nomcica.f90 and the cloud overlap method is selected by setting flag icld.

Maintenance and Contact Info

Atmospheric and Environmental Research, 131 Hartwell Avenue, Lexington, MA 02421

Original version: E. J. Mlawer, et al. (AER) Revision for GCMs: Michael J. Iacono (AER)

Contact: Michael J. Iacono (E-mail: [email protected])

References

• AER Radiative Transfer Models Documentation
• Github Wiki
• RRTMG_LW, RRTM_LW
  • Iacono, M.J., J.S. Delamere, E.J. Mlawer, M.W. Shephard, S.A. Clough, and W.D. Collins, Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models, J. Geophys. Res., 113, D13103, doi:10.1029/2008JD009944, 2008.
  • Clough, S.A., M.W. Shephard, E.J. Mlawer, J.S. Delamere, M.J. Iacono, K. Cady-Pereira, S. Boukabara, and P.D. Brown, Atmospheric radiative transfer modeling: a summary of the AER codes, J. Quant., Spectrosc. Radiat. Transfer, 91, 233-244, 2005.
  • Iacono, M.J., J.S. Delamere, E.J. Mlawer, and S.A. Clough, Evaluation of upper tropospheric water vapor in the NCAR Community Climate Model (CCM3) using modeled and observed HIRS radiances. J. Geophys. Res., 108(D2), 4037, doi:10.1029/2002JD002539, 2003.
  • Iacono, M.J., E.J. Mlawer, S.A. Clough, and J.-J. Morcrette, Impact of an improved longwave radiation model, RRTM, on the energy budget and thermodynamic properties of the NCAR Community Climate Model, CCM3, J. Geophys. Res., 105, 14873-14890, 2000.
  • Mlawer, E.J., S.J. Taubman, P.D. Brown, M.J. Iacono, and S.A. Clough: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16663-16682, 1997.
• McICA
  • Pincus, R., H. W. Barker, and J.-J. Morcrette, A fast, flexible, approximation technique for computing radiative transfer in inhomogeneous cloud fields, J. Geophys. Res., 108(D13), 4376, doi:10.1029/2002JD003322, 2003.
• Latitude-Varying Decorrelation Length
  • Oreopoulos, L., D. Lee, Y.C. Sud, and M.J. Suarez, Radiative impacts of cloud heterogeneity and overlap in an atmospheric General Circulation Model, Atmos. Chem. Phys., 12, 9097-9111, doi:10.5194/acp-12-9097-2012, 2012.
• Full list of references

rrtmg_lw's Issues

Cloud overlap computation in mcica_subcol_gen_lw.1col.f90 for icld 4 and 5 are identical

case (4) and case(5) are identical

RRTMG_LW/src/mcica_subcol_gen_lw.1col.f90

Line 577 in 6e8786a

case(5)

possible incorrect variable used

Here

RRTMG_LW/src/mcica_subcol_gen_lw.f90

Line 639 in 6e8786a

do ilev = 2,nlayers

and in some places above the nlayers is used instead of nlay, but there is no such identifier nlayers defined

possible memory access to outside of array

Here:

RRTMG_LW/src/rrtmg_lw_rtrnmr.f90

Line 497 in 6e8786a

if (istcldd(lev).ne.1) then

            if (istcldd(lev).ne.1) then
	        faccmb1d(lev-1) = max(0.,min(cldfrac(lev+1)-cldfrac(lev), &
        	    cldfrac(lev-1)-cldfrac(lev)))
            	faccmb2d(lev-1) = max(0.,min(cldfrac(lev)-cldfrac(lev+1), &
                    cldfrac(lev)-cldfrac(lev-1)))
	    endif

the outer cycle is do lev = nlayers, 1, -1
cldfrac is dimension(:)
when lev =1 , then cldfrac(lev-1) is accessing out of bounds

This occurs when cldfrac(1) is larger than 1.e-6_rb;

Run Examples Jupyter Notebook

write a Jupyter Notebook to walk users through how to use the model

Possibly tag older versions?

A question for the fine folks at AER. Looking at this repo and RRTMG_SW, it looks like the code goes back in time a fair bit. So I was wondering if it could be possible tag the git repo so that it "matches" what would have been in, say, RRTMG_LW v4.91, v4.90, etc. ?

This way one could do git diff v4.91 and see the differences from that version.

possible incorrect assignment

Here

RRTMG_LW/src/mcica_subcol_gen_lw.f90

Line 619 in 6e8786a

CDF(isubcol,i,ilev) = rand_num

CDF(isubcol, i, ilev) = rand_num
element is assigned to a column

where
real(kind=rb), dimension(nsubcol, ncol, nlay) :: CDF
real(kind=rb), dimension(ncol) :: rand_num

Fix links in Wiki!

they still point to RTWeb. whenever we move everything to Github, we should force the links to point to their respective wikis

Migration to v5.1

setcoef bug

setcoef function makes
rat_h2och4 and other rat_xxx arrays.

rat_h2oo3
rat_h2oo3_1
rat_h2on2o
rat_h2on2o_1

arrays and possibly others are filled with NaNs by half after call to setcoef subroutine.
Futher in the code there are no checks for NaN in rat_xxxx arrays.

rat_xxx should be set to some adequate value at initialisation or in the code

RRTM LW cloud optical properties

Hello,

I am curious if I can ask a question regarding RRTM in this GitHub? Specifically, I am trying to model scattering and absorption so I am using the DISORT code option. When I try to run the RRTM LW code with the IN_CLD_RRTM input containing the optical properties of an aerosol over two layers, I receive this error:

******* ERROR >>>>>> QGAUSN--max iteration count

I am able to run the code and receive the output file when I do not use DISORT, but i am very interested in the scattering contribution of my input aerosols. I will attach the file I am using for the optical properties. Thanks!

aod_for_incld_wd11_0p3.txt

possible typos in rrtmg_lw_read_nc.f90

I think there are two typos in rrtmg_lw_read_nc.f90, possibly coming from a replace in the file

../src/rrtmg_lw_read_nc.f90(762): error #6404: This name does not have a type, and must have an explicit type. [IM2]
status(1_im2) = nf90_inq_varid(ncid,"H20ForeignAbsorptionCoefficients",varID)
-----------------^
../src/rrtmg_lw_read_nc.f90(762): error #6975: A kind-param must be a digit-string or a scalar-int-constant-name. [IM2]
status(1_im2) = nf90_inq_varid(ncid,"H20ForeignAbsorptionCoefficients",varID)
-----------------^
../src/rrtmg_lw_read_nc.f90(763): error #6404: This name does not have a type, and must have an explicit type. [IM3]
status(1_im3) = nf90_get_var(ncid, varID, forrefo, &
-----------------^
../src/rrtmg_lw_read_nc.f90(763): error #6975: A kind-param must be a digit-string or a scalar-int-constant-name. [IM3]
status(1_im3) = nf90_get_var(ncid, varID, forrefo, &

I suppose they should be
status(12) and status(13)

How to output spectrally-resolved fluxes?

Hi there!

My name is Andrew and I'm a PhD student in Atmospheric Physics. At the moment, I am using RRTMG_LW and coupling it to a single-column model to explore the climate response.

As part of this research, I would like to be able to access the underlying band fluxes from the RRTMG_LW model (ie. net top of atmosphere flux in each of the spectral bands). I'm having a bit of trouble with this at the moment and was wondering if you could assist? I have also emailed Michael, but thought it best to also raise an issue.

My current approach is to take advantage of the loop over spectral bands in rrtmg_lw_rtrnmc.f90 (line 335). Then for each band, once the loop over g-points has finished (around line 534) and the irradiances have been converted to fluxes, I write the top of atmosphere flux for that band to an output array, toaflux_sr.

======== ~ line 534 ~

! Increment g-point counter
         igc = igc + 1
! Return to continue radiative transfer for all g-channels in present band
         if (igc .le. ngs(iband)) go to 1000

! Process longwave output from band for total and clear streams.
! Calculate upward, downward, and net flux.
         do lev = nlayers, 0, -1
            uflux(lev) = urad(lev)*wtdiff
            dflux(lev) = drad(lev)*wtdiff
            urad(lev) = 0.0_rb
            drad(lev) = 0.0_rb
            totuflux(lev) = totuflux(lev) + uflux(lev) * delwave(iband)
            totdflux(lev) = totdflux(lev) + dflux(lev) * delwave(iband)
            uclfl(lev) = clrurad(lev)*wtdiff
            dclfl(lev) = clrdrad(lev)*wtdiff
            clrurad(lev) = 0.0_rb
            clrdrad(lev) = 0.0_rb
            totuclfl(lev) = totuclfl(lev) + uclfl(lev) * delwave(iband)
            totdclfl(lev) = totdclfl(lev) + dclfl(lev) * delwave(iband)
         enddo

! ANDREW WILLIAMS ADDITION
! Write TOA upward longwave flux in this band to output array
! TOA lev = nlayers
! But first, need to add in correction due to fluxfac
! N.B. Don't need delwave(iband) here!
         toaflux_sr(iband) = toaflux_sr(iband) + uflux(nlayers) * fluxfac 

========

However, so far this approach has generated non-sensical values for the net toa fluxes. For example, in many of the output bands, the net outgoing longwave fluxes are negative (see example output array, below).

>>> OLR_sr
        ([[ 1.44658334e-02,  6.92333806e-03, -7.00591313e-03,
        -1.80155860e-02, -1.01838347e-02, -7.32874781e-03,
        -4.97357311e-03, -2.94247288e-03, -1.57885080e-03,
        -6.30687957e-04, -2.55684173e-04, -2.74535550e-05,
        -6.02877610e-06, -1.95616080e-06, -6.31172205e-07,
        -5.00545830e-08]])

I have look at the RRTMG_LW website and github repository and I wondered if you might be the person to contact with this question? I'm sorry if this is not appropriate, but I would appreciate any help you can provide on this!

Thanks very much!

Cheers,
Andrew

Downward flux in output_rrtm_ICRCCM_sonde

hello,
I have a question regarding one of the examples:
run_examples_std_atm)/output_rrtm_ICRCCM_sonde

When I compare the output of the same example run by RRTM_LW (v3.3) the resulting downward flux and net flux are very different:
RRTM lowest level: 0 973.6 406.2205 310.9624 95.2581065 -1.35171
RRTM_G lwst level: 0 973.6 406.2213 106.5797 299.6415589 -0.75080

This seems to me the only example that gives such a big difference between the two programs, what is the cause?

d_rad0_dt may be unitialized in this function

d_rad0_dt may be used before it is set

RRTMG_LW/src/rrtmg_lw_rtrnmr.f90

Line 672 in 6e8786a

d_radclru_dt = d_rad0_dt

d_rad0_dt is set only in line 657, when idrv .eq. 1

what default value for d_rad0_dt should be used NaN, Zero or whatever?

There are also lots of uninitailized vars used in computations.

What value should they be initialized to?
I have Intel compiler and it supports initializeng vars to zero with /Qinit:zero or to NaN /Qinit:snan

INPUT ALTITUDES NOT IN ASCENDING ORDER when IMMAX<0

Hello,
in my INPUT_RRTM file I'm setting IMMAX<0 (IBMAX is also <0) but I'm getting
INPUT ALTITUDES NOT IN ASCENDING ORDER

I tried to set ZM to 0.0; to ascending altitude as well as to leave the field empty, the PM field is descending.

The same INPUT_RRTM works correctly with RRTM_LW
Do you have any suggestion?

INPUT_RRTM.txt