The MODTRAN band model was developed by Spectral Sciences, Inc. (SSI) in the late 1980's as part of an Air Force sponsored Small Business Innovative Research (SBIR) effort. MODTRAN was originally designed to provide a finer spectral resolution option to the 20 cm-1 Air Force Geophysics Laboratory (now the Air Force Research Laboratory, AFRL) LOWTRAN band model. LOWTRAN essentially mapped arbitrary lines-of-sight into equivalent sea-level horizontal paths for which molecular band models were developed. The MODTRAN band model provided finer spectral resolution by introducing temperature dependent absorption coefficient and line spacing band model parameters. These molecule dependent parameters characterize the distribution of molecular transitions in each spectral bin.
MODTRAN uses a statistical characterization of the distribution of line strengths in a spectral interval to compute spectral band transmittances, radiance, fluxes, etc. This statistical approach leads to inaccuracies, and these depend strongly on the calculation scenario. The claim use to be made that " a good rule of thumb is that transmittance absolute accuracy is generally better than ±0.005, thermal brightness temperature is generally accurate to better than 1K, and radiance accuracy is approximately ±2%." Since MODTRAN now includes a line-by-line (LBL) radiative transfer option, we encourage users to compare MODTRAN statistical and LBL methods for their specific scenario to benchmark residuals. This is not to imply that the MODTRAN LBL calculations are exact. However, the dominant source of MODTRAN error is statistical, and the LBL comparisons give a direct measure of that largest error component.
Recent MODTRAN publications include [Berk et al., 2014; 2015]. The 2014 paper should be used as a general MODTRAN reference, while the 2015 paper focuses on the MODTRAN line-by-line algorithm:
1. A. Berk, P. Conforti, R. Kennett, T. Perkins, F. Hawes, and J. van den Bosch, "MODTRAN6: a major upgrade of the MODTRAN radiative transfer code," Proc. SPIE 9088, Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XX, 90880H (June 13, 2014); doi:10.1117/12.2050433.
2. Alexander Berk, Patrick Conforti, and Fred Hawes, "An accelerated line-by-line option for MODTRAN combining on-the-fly generation of line center absorption with 0.1 cm-1 bins and pre-computed line tails," Proc. SPIE 9471, Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XXI, 947217 (May 21, 2015); doi:10.1117/12.2177444
After downloading the software, Windows users simply click on the mod6setup installation program to install MODTRAN. Linux users are provided a zipped tar file, which needs to be decompressed.
MODTRAN now includes a JAVA-based GUI, which runs on Windows, Linux, and Mac operating systems. The GUI is designed to facilitate setting up MODTRAN input files. A small set of use-cases are provided, for which inputs are initialized. Alternatively, one can create a case within the GUI from scratch by simply entering input selections. The GUI allows the users to save their template input files and/or run MODTRAN from within the GUI. After a run is complete, the user can choose to plot spectral outputs.
The 6 model atmospheres in MODTRAN differ most significantly in their temperature, H2O, and O3 profiles. The temperature profiles are illustrated in Figure 1. Not surprisingly, the Sub-Arctic Winter Atmosphere has the coolest surface temperature; the Mid-Latitude Winter Atmosphere has the next coolest surface temperature. The Tropical and Mid-Latitude Summer Atmospheres have the warmest surface temperatures. At the tropopause, the temperature of the Tropical Atmosphere is the coolest and the temperature of the Sub-Arctic Summer Atmosphere is the warmest. For all 6 atmospheres, a secondary temperature peak occurs near 50 km; here the Sub-Arctic Summer temperature is the warmest, essentially equal to that of the Mid-Latitude Summer. The Sub-Arctic Winter temperature is the coolest at this altitude. The 1976 U.S. Standard Atmosphere temperature profile provides an effective median for the set of profiles.
Figure 2 contains the density profiles for H2O and O3. Total vertical column amounts for the 12 ambient band model species are listed in Table 1; the CO2 mixing ratio used in creating this table was 380 ppmV. It is interesting to note that the 6 atmospheres provide a nice spread of boundary layer and lower tropospheric water densities.
|Mol\Atm||Tropical||Mid Lat Summer||Mid Lat Winter||Sub Arc Summer||Sub Arc Winter||US Standard|
The MODTRAN users' manual provides descriptions of spectral outputs. As an example, Table 2 contains the descriptions of spectral radiance output.
Slit function central frequency in cm-1.
Slit function direct transmittance for the line-of-sight (LOS) path including all sources of molecular and particulate extinction.
Thermal emission in units of W cm-2 sr-1 / cm-1.
The atmospherically scattered path thermal scatter in units of W cm-2 sr-1 / cm-1, i.e., the thermal radiation scattered by the atmosphere directly into the LOS and transmitted to the sensor.
Surface emission directly transmitted to the sensor in units of W cm-2 sr-1 / cm-1. If the LOS terminates at the ground, this term is computed as the product of the Planck surface emission, the directional emissivity, and the path transmittance. If the LOS does not terminate at the ground but a positive temperature is specified for input TPTEMP, SURF_EMIS will contain the transmitted surface emission of a target object. If the LOS does not terminate at the ground and input TPTEMP is zero, then SURF_EMIS is zero.
The single scatter contribution to SOL_SCAT in units of W cm-2 sr-1 / cm-1.
Solar/lunar radiation in units of W cm-2 sr-1 / cm-1 multi-scattered by the atmosphere and directly transmitted to the sensor. This excludes the single scatter component, SING_SCAT.
Ground reflected radiation directly transmitted to the sensor in units of W cm-2 sr-1 / cm-1. It includes reflection of three downward flux components: the direct solar, the diffuse solar, and the diffuse thermal.
The direct solar component of GRND_RFLT, i.e., the radiance in units of W cm-2 sr-1 / cm-1 arising from solar photons which travel along the Sun to ground to sensor path without being scattered or absorbed by the atmosphere.
The total radiance observed by a sensor in units of W cm-2 sr-1 / cm-1. This is computed as the sum of THRML_EM, THRML_SCT, SURF_EMIS, MULT_SCAT, SING_SCAT and GRND_REFL.
The product of the sensor-to-final_altitude-to-Sun transmittance (final_altitude is either the ground, the TOA or H2ALT) and TOA solar irradiance in units of W cm-2 / cm-1. It does not include the surface reflectance. This is included in the output for target insertion applications.
Solar irradiance transmitted to the observer, calculated as the product of the TOA spectral solar irradiance in units of W cm-2 / cm-1 and the Sun to sensor or observer (H1) spectral transmittance.
The negative natural logarithm of the direct transmittance for the LOS path. Since band model transmittances do not obey Beer's Law, the negative natural logarithm of the transmittance should not be associated with an optical depth when a spectral bin has significant spectral structure.
Directional emissivity at ground towards sensor (between 0 and 1 inclusive).
Top-of-the-atmosphere solar irradiance in units of W cm-2 / cm-1.
Brightness temperature in Kelvin, defined as the temperature a blackbody would need to have to emit the TOTAL_RAD radiance.
MODTRAN contains legacy text, comma separated values (*.csv) and ENVI® spectral library (*.sli) output files. The binary *.sli files each have an accompanying header file (*.hdr). The contents of the files <rootname>.* are listed below in Table 3.
Standard output, describing inputs used, path geometry, warnings, spectral data, and more.
Primary MODTRAN spectral output file containing transmittance, radiance, and/or irradiance data.
<rootname>.tp7 spectral transmittance, radiance, and/or irradiance data that has been scanned using a user-specified filter function.
Auxiliary MODTRAN spectral output file containing spectral data for each path segment along observer lines-of-sight.
A two column MODTRAN spectral output file containing a spectral grid in the first column and either transmittance, radiance, or irradiance in the second column.
<rootname>.plt spectral data that has been scanned using a user-specified file.
A MODTRAN spectral output file containing sensor and solar path direct and diffuse transmittances along with spherical albedo data for use in atmospheric correction of down-looking spectral imagery data.
<rootname>.tp7 spectral transmittance, radiance, and/or irradiance data that has been spectrally convolved with user-supplied sensor spectral response functions (SRFs).
Spectral flux data at multiple altitude levels.
Spectral cooling rate data at multiple altitude levels.
Line-of-sight refractive geometry path output.
k-distribution dependent transmittance data as a function of path length.
k-distribution dependent radiance data as a function of path length.
MODTRAN comment, warning, and error messages.
<rootname>.tp7 spectral transmittance, radiance, and/or irradiance data in a comma separated value format.
<rootname>.tp7 spectral transmittance, radiance, and/or irradiance data in a comma separated value format.
Line-by-line spectral transmittance and radiance in a comma separated value format.
<rootname>.tp7 spectral transmittance, radiance, and/or irradiance data in ENVI® spectral library format.
Header for <rootname>.sli.
<rootname>.7sc spectral transmittance, radiance, and/or irradiance data in ENVI® spectral library format.
Header for <rootname>_scan.sli.
Line-by-line spectral transmittance and radiance in ENVI® spectral library format.
Header for <rootname>_highres.sli.
Observer altitude above sea level (km),
Path final altitude above sea level (km),
Observer path zenith angle (deg),
Refracted path slant range (km),
Earth center angle (deg),
Short (0) vs. long (1) path range switch,
Earth radius (km), and
H2ALT to H1ALT zenith angle (deg).
Fig. 4 also illustrates that ambiguity arises when the observer zenith angle, OBSZEN, exceeds 90°, the observer altitude, H1ALT, exceeds the final altitude, H2ALT, and the LOS does not intersect the Earth; these inputs are consistent with both a short and long path, with the long path passing through the tangent height, HTANGENT. The keyword LENN selects which of these 2 paths to be used. Note that the same ambiguity arises for the slant path which is defined by the input set (BCKZEN, H2ALT, H1ALT) when BCKZEN > 90° and H2ALT > H1ALT; this case also requires use of the keyword LENN.
MODTRAN also includes an option to have the line-of-sight specified as a path to space or ground (ITYPE = 3). For these paths, there are only 3 input options: (H1ALT, OBSZEN), (H1ALT, H2ALT), and (H2ALT, BCKZEN). It is important to remember that when the (H1ALT, H2ALT) option is used, the input H2ALT defines tangent height, HTANGENT, not a final path altitude, and that H1ALT must exceed H2ALT. Also, note that input RAD_E is used if provided; otherwise, a default value is selected based on the chosen model atmosphere, by keyword MODEL.
A second set of MODTRAN geometry inputs (shown in Fig. 5) defines the solar (or lunar) geometry. The direction of incident radiation is specified from the perspective of either the sensor (IPARM equal 0, 1, or 2) or the path end point (IPARM equal 10, 11 or 12). Whichever frame of reference is used, the solar direction is ultimately defined in terms of solar zenith and relative solar azimuth angles. The solar zenith angle is the angle between the vertical at the reference point and the refracted path solar direction (yellow ray) at that point. The relative azimuth is the angle between two vertical planes at the reference point, one containing the line-of-sight (green ray in left image; red ray in right image) and the second containing the solar path. If the two solar angles are not entered directly via IPARM equal 2 or 12, the required inputs are the latitude and longitude of the reference point along with TRUEAZ, the true path azimuth (degrees East of North) at the reference point. In addition, the absolute angular location of the Sun from the Earth's perspective must be determined either by directly entering the solar latitude and longitude (IPARM equal 0 or 10) or from temporal data by specifying the day of year using IDAY and Coordinated Universal Time using GMTIME. For all cases, the Earth-to-Sun distance is determined from the day of year, IDAY.