A Line-By-Line (LBL) radiative transfer option has been developed for MODTRAN6. The MODTRAN6 algorithm solves the LBL radiative transfer equations at arbitrarily fine spectral resolution within 0.1 cm-1 spectral bins, stitching together contiguous bins to build the full band pass spectrum. The advantage of the narrow spectral bin approach is that tail contributions from molecular transitions centered more than 0.05 cm-1 from a bin edge can be pre-calculated analytically. The number of transitions that must be modeled on-the-fly for any given 0.1 cm-1 bin is modest.
The MODTRAN6 LBL calculations can include DISORT thermal and solar multiple scattering. The full suite of MODTRAN 0.1 cm-1 band model and correlated-k outputs are available with the LBL option, allowing for direct comparison among the radiative transfer approaches. The LBL option also stores the LBL fine spectral resolution transmittance and line-of-sight radiance data.
The MT_CKD  H2O continuum was introduced into MODTRAN6, and 0.001 cm-1 MODTRAN6 line-by-line transmittances were validated against transmittances generated with the LBLRTM  model. Residuals primarily arise from the differences in molecular line compilations for specific transitions.
MODTRAN6 computes line-of-sight (LOS) radiances by integrating through a stratified spherical atmosphere, including the effects of light refraction. Multiple scattering contributions are most accurately computed using the plane parallel atmosphere DISORT discrete ordinate model. Many changes have to be made to the standard DISORT distribution to accurately compute the MODTRAN6 scattering contributions. Typically, DISORT generates plane-parallel LOS radiances using a delta-M scaling method to improve characterization of scattering phase functions. MODTRAN6 requires calculation of segment (not full LOS) unscaled radiances defined for its spherical refractive path. DISORT must be modified to (1) ingest MODTRAN refractive path segment data, (2) generate delta-M radiances for this collection of segments, (3) transform these segment radiances into unscaled values, and (4) extract out the necessary scattering components. In the end, DISORT is still solving a plane-parallel atmosphere problem. However, radiances are being defined for path segments that match those derived for the spherical atmosphere.
In earlier versions of MODTRAN, e.g. MODTRAN5.3, DISORT was used to define layer dependent multiple scattering effective source functions. The figure illustrates that this earlier DISORT integration method could lead to significant errors.
Users can choose to entirely avoid editing text input files and instead may run MODTRAN6 through an interactive graphical user interface (GUI). The GUI lets users start from scratch, from an existing JSON input file or a collection of baseline use cases. Once loaded, a model can be run directly from the GUI, or the user may first use a settings window to tweak parameters, add additional cases to the model, or to make major changes. Redefined models can be saved to JSON files so that the models can be re-loaded in later MODTRAN6 runs. After running models, the resulting spectra are displayed in an interactive plotting panel that can be used to explore the data.
Code developers wishing to integrate MODTRAN6 into their own software can now run MODTRAN6 using a library (modtran6.dll or modtran6.so) file with an Application Programming Interface (API). The full set of MODTRAN inputs and outputs are described in the MODTRAN6 Interface Control Document (ICD), MODTRAN6_ICD.docx. Default values are defined in a keywords.json file.Sample use case JSON input file is shown below. Note that including comments (# ...) at the end of lines is not a standard JSON feature, but it is allowed by MODTRAN6.
MODTRAN6 defines a stratified atmosphere via constituent vertical profiles. That basic symmetry is relaxed with the introduction of a local chemical (or gas) cloud option. MODTRAN6 enables the user to define line-of-sight profiles of temperature and molecular gases. These are modeled as perturbations of the ambient atmosphere. MODTRAN6 computes ambient, ambient plus local chemical cloud, and contrast spectral signature components. If one or more of the cloud gases is also an ambient species, the gas amounts are combined, providing a spectrally correlated treatment of the absorption. The basic premise is made that the local chemical cloud does not perturb the diffuse flux impingent on the cloud, but it does perturb segment radiance contributions within the cloud.
The local chemical cloud option can be combined with the MODTRAN6 multiple line-of-sight (MLOS) option to generate cloud images from a single MODTRAN run. MODTRAN6 band model data is provided for the HITRAN line compilation of molecular species. MODTRAN compatible data is also available from Spectral Sciences, Inc. (SSI) for the full suite of PNNL Infrared Spectral Library (IRSL) species. Please contact Spectral Sciences, Inc. if you are interested in purchasing this data.
An Aerosol Toolkit (ATK) provides users the option to easily generate aerosol and/or cloud optical property and profile data, in MODTRAN's SAP (Spectral Aerosol Profile) format, from user input aerosol specifications. The ATK allows users to create data files for "custom" aerosols with multiple components (specifying particle types, spectral refractive index data, particle size distributions, and number density profiles). Single particle optical properties for each aerosol component can be calculated directly using Mie code for spherical particles, or T-Matrix codes for more general axisymmetric particles.
The Toolkit uses Mie and T-Matrix software developed by Michael Mishchenko at the NASA Goddard Institute for Space Studies. The first-principles calculations of aerosol properties are described in text by Mishchenko et al.  The Mie and T-matrix subroutines return both arrays of scattering matrix elements as a function of user-specified scattering angles, as well as angular function expansions of the matrix elements. The ATK also has the potential to output polarized scattering data.
The Atmosphere Generator Toolkit (AGT) is a separate program that supplements MODTRAN with new atmospheres from historical and radiosonde data. It provides MODTRAN6 users the ability to input custom atmospheres into MODTRAN in a straightforward manner. The command-line tool ingests either location (latitude, longitude, date, and time) or radiosonde data. The AGT writes the custom atmosphere in either the form of MODTRAN legacy (tape5) CARD 2C1 inputs, or as text for a MODTRAN6 JSON input file.
The radiosonde AGT input can be either post-processed or raw data. A raw radiosonde typically consists of several thousand lines (or more) of altitude dependent pressures, temperatures and relative humidities, ranging from the near ground level to about 30 km. Altitude gaps or noisy profiles are common. A post-processed radiosonde typically consists of processed measurements at hundreds (not thousands) of altitudes. Typically this data extends to 25-30 km, with fewer missing data points and minimal noise. As examples, NOAA/ERSL and The University of Wyoming both maintain actively updated collections of post-processed radiosondes. File formats are not consistent between data sources; the AGT understands the three most common file formats and regularizes the input stream prior to processing the data for usage within MODTRAN.
Long-term time-averaged global data is available from a database maintained by NOAA in the form of the NCEP/NCAR Reanalysis Monthly Means and Other Derived Variables. This publicly available database contains long-term time-averaged data sets covering the entire globe on a 2.5 degree latitude by 2.5 degree longitude grid at 17 pressure levels that span from ground level to an approximate altitude of 30 km. Temperature and geopotential height data are present at all levels, while relative humidity is only available from the ground to an approximate altitude of 9 km. The long-term monthly data covers 12 months averaged from January 1948 to present. The four times daily data covers covers the globe at 6 hour intervals averaged from 1981-2010.