Radiation Code

Purpose

This FORTRAN code has kindly been provided by Hal Woolf and should be referenced as the Woolf VASFRWRD Model Version 1.  

After compilation, the code in file VASFRWRD.F will calculate the radiances that a VAS instrument in orbit would see given a temperature and moisture sounding. The purpose of this code is to show how to compare temperature and moisture analyses to the radiances at the top of the atmosphere measured by the VAS instruments on the GOES satellites.  A dataset of observed clear sky radiances over the continental U.S. and surrounding ocean is available from the University of Wisconsin GOES_VAS_Product_ Set Project.  These VAS clear radiance data can be compared to those derived from the temperature and moisture fields in atmospheric General Circulation Models.   Furthermore, a user can run sensitivity experiments to determine the probable significance of any discrepancies between the observed radiances and those generated from atmospheric profiles simulated in a GCM.   Output from this code has been compared to line by line calculations using state of the art transmittance models.

The subroutine  VASRTW retrieves an approximate atmospheric profile from VAS radiances, using coefficients derived by correlation with a sample of radiosondes.  This may be used to invert the transformation performed by VASFRWRD.  Such a retrieval is inevitably under-determined, and is only one out of infinitely many which are consistent with the stated radiances and the probable errors in measurement.   It is not recommended as a definitive determination of the true state of the atmosphere, but is included because it is used in the analysis  of cloud top pressure for the partially cloudy pixels within view of the satellite. For this purpose it is sufficiently accurate.  


Instructions

Below are the instructions for compiling and running the example program to calculate VAS radiances from a sounding such as that in SOUNDING.TXT.

NOTE: File names are in lower case, following the UNIX convention; they are shown here in upper case for clarity.

This code, VASFRWRD.F, and associated coefficient files, as delivered, will run ONLY on a UNIX platform. To download, compile, and run VASFRWRD perform the following steps:

1. Download a tar file containing the source code for the VASFRWRD.F program, the SOUNDING.TXT and EXAMPLE.TXT text files, and the asscoiated binary coefficient files (.DAT extension). To untar the files enter the command below.

      tar -xvf radiation_code.tar

2. Compile VASFRWRD.F on an UNIX platform using the appropriate command below.

     IBM f77 -qrndsngl -qnomaf -o vasfrwrd vasfrwrd.f
     SGI f77 -32 -O0 -mips2 -static -bytereclen -col72 -o vasfrwrd vasfrwrd.f
     SUN f77 -o vasfrwrd vasfrwrd.f

3. Run the executable VASFRWRD which uses SOUNDING.TXT to produce the file EXAMPLE.OUT. The text of EXAMPLE.OUT should be the same as the provided EXAMPLE.TXT.

VASFRWRD expects temperature and moisture profiles from SOUNDING.TXT. To experiment with other soundings similar files must be created.

Users are encouraged to contribute an e-mail description to gvp@ssec.wisc.edu of their experiences running this code on other platforms.  These will be made available on  the VAS Product Home Page, accessible through the permanent (redirecting) URL:

"http://www.purl.oclc.org/net/U_Wisconsin_SSEC/VAS_Product/index.html"

All level- or pressure-dependent calculations are performed at the levels shown in SOUNDING.TXT. An arbitrary sounding must be interpolated and/or extrapolated to the specified pressures; a standard atmosphere or climatology is sufficient for the highest levels. Temperature must be specified in Kelvin, and moisture (as mixing ratio) in grams/kilogram. Dewpoint temperature can be transformed to mixing ratio via the function "satmix" (contained in VASFRWRD.F) as follows:

 w = satmix(p,d)

where w is mixing ratio, p is pressure (mb), and d is dewpoint (K).

There are 4 additional required variables, presently hard-coded in VASFRWRD.F starting at line 20:

ksat - GOES satellite number: valid values are 4,5,6,7.

zena - zenith angle of the GOES view through the atmosphere to the location of the sounding; used in transmittance calculations. This parameter is included with each VAS clear-radiance set as the Grid {ASaZ}.

kmon - numeric month of the calendar year, used together with latitude   (see below) to obtain ozone mixing ratio from a set of climatological tables.

rlat - latitude in degrees, from +90 (North Pole) to -90 (South Pole).

zena and kmon are also used to select coefficients for statistical retrieval.

VASFRWRD produces vertical profiles of atmospheric transmittance for each of the 12 channels; they are printed in EXAMPLE.TXT. See below for information regarding construction of the transmittance model and related subject matter.

Calculated VAS radiances are displayed in two forms:

spectral radiance .................. unit = mW/m^2-steradian-cm^-1
equivalent blackbody temperature ... unit = K

The calculated equivalent blackbody (or brightness) temperatures are then passed to subroutine "vasrtw" to produce a statistical-regression-based retrieval (described below) of temperature and moisture.

The resulting profiles, also printed in EXAMPLE.TXT, can be compared to the input profiles used to synthesize the VAS radiances.


Transmittance Modeling for VAS

(1) Line-by-line database

Calculations have previously been done, using LBLRTM release 3.26 (22 Jan 96) and HITRAN96, for a set of 32 atmospheres, consisting of the 1976 U.S. Standard and 31 diverse profiles representing a wide range of meteorological conditions, from arctic to tropical. The very high, variable spectral resolution output was reduced to uniform 0.1-wavenumber spacing by simple averaging. Four runs were made: ALL, using the seven "standard" molecular species defined in LBLRTM -- water vapor, carbon dioxide, ozone, nitrous oxide, carbon monoxide, methane, and oxygen; WCO, water vapor continuum only; WNC, water vapor with no continuum; and WVO, water vapor (including continuum) plus ozone. The calculations, which span the spectral range 550 to 2950 cm^-1, have been done for nadir view plus five zenith angles, corresponding to values of secant(theta) = 1.00, 1.25, 1.50, 1.75, 2.00, and 2.25. The values of theta are 0., 36.87, 48.19, 55.15, 60., and 63.61.

(2) Generation of instrument transmittances

The 0.1-wavenumber transmittances for the four absorber configurations described in (1) were convolved with the response functions for the 12 bands or channels of the VAS instruments on GOES-4 through -7. The resulting band transmittances were then subjected to the following operations: wet = wco * wnc ozo = wvo / wet dry = all / wvo where "dry" is a generic term applied to the remaining uniformly mixed gases, primarily carbon dioxide. Defining "dry" and "ozo(ne)" in this manner preserves the validity of the product rule for total transmittance, even though the data do not actually represent monochromatic conditions.

(3) Fast (parameterized regression) model

The transmittance profiles in each band were then used to generate separate sets of regression coefficients (for the constituents dry, ozo, wco, and wnc) employing the PLOD (PressureLayeredOpticalDepth) algorithm [AIRS LAYERS package science notes, S. Hannon and L. Strow, University of Maryland - Baltimore County (UMBC)]. In addition, a set of "band- correction" coefficients was generated to account for the polychromatic character of the bands or channels in calculating radiances and brightness temperatures. These procedures were executed for each of the four GOES-VAS instruments, and the results stored in the BigEndian floating-point binary direct-access files: vasxtbnd.dat vasxtdry.dat vasxtozo.dat vasxtwco.dat vasxtwtl.dat vasxtwts.dat The reader is referred to the routines "pfcvas" and "tauvas" for usage of these files.

Results:

The results for 13 profiles are shown here.  Profiles 1-12 were part of the set of 32 used in generating the correlation coefficients, i.e. they are dependent.  Profile 33 is from  sounding.txt, and is quite independent.  The entries in the table are residuals, the difference in brightness temperature using the Woolf model minus that using LBLRMTM.  The first table shows the comparison for channels 1-6;  the second table for channels 7-12  

GOES-07 SET 2, ZENITH ANGLE 48.19 atm 1-12 dependent, 33 independent
CHANNEL 1 2 3 4 5 6
WAVE-NUM  681.448  691.717  704.717  715.731  750.220 2213.552
1 -.00346 -.00552 -.04311 -.05737 -.00839 .36905
2 -.00308 .00133 -.00168 .00307 -.01985 .00470
3 -.00139 .00409 .01221 .03107 .00342 -.00015
4 -.04024 -.05566 -.05173 -.05557 -.04056 -.11272
5 -.00217 .01089 .03233 .06825 .04643 .11107
6 -.00299 .00574 .01314 .03716 .02634 .16158
7 -.00853 -.00841 -.00330 .02074 .01389 -.08160
8 -.04210 -.03816 .07997 .12158 .05267 -.05663
9 -.04836 -.06532 -.04376 -.04970 -.05298 -.04533
10 -.00636 .00092 .05194 .10712 .06754 -.05966
11 -.00014 .00404 .00929 .02336 .01074 .25259
12 -.00282 .00203 .00757 .01714 .00554 .01813
33 -.00270 -.00218 -.02895 -.02690 -.03003 .15991

     
CHANNEL 7 8 9 10 11 12
WAVE-NUM  789.179  898.584 1376.914 1489.870 2232.314 2532.704
1 -.05664 -.01703 -.28665  -.57704 .35405 .04462
2 -.00719 .00320 -.01831 -1.39276 .00880 -.00046
3 -.00914 .00095 -.03230 -1.22119 -.00044 -.00203
4 -.00662 -.00156 .19528 -.33603 -.11430 -.00490
5  .00134 .00201 -.08461 -.67143 .11630 .00436
6  -.00011 .00114 .02936 -.82130 .14612 .01035
7 .00121 -.00169 .02359 -.58690 -.07533 -.00601
8 .00909 -.00327 -.03529 -.30779 -.09709 -.00494
9 -.01474 -.01358 .02629 -.16933 -.10196 -.00272
10 -.01788 -.01306 -.06340 -.38043 -.05591 -.00290
11 -.00464 -.00241 -.07436 -.42195 .24252 .01666
12 .00172 -.00243 -.00067 -1.82355 .02444 .00064
33 .03284 -.01111 -.25577 -.59581 .16434 .01089

The comparison shows little obvious difference among the dependent and independent profiles, with the residuals typically at most a few tenths K , except for channel 10, where the differences are typically about 1 K.  It should be noted that, in the radiosonde dataset which was used to generate the regressions, the water vapor vapor in the upper troposphere to which channel 10 is primarily sensitive was routinely set to a constant value because the sensor was not trusted to to give reliable results.  


LBLRTM: S.A.Clough, "Radiative transfer model development in support of the Atmospheric Radiation Measurement program," in Proceedings of the Third Atmospheric Radiation Measurement (ARM) Science Team Meeting, CONF-9303112, U.S.Department of Energy, Washington, D.C., 1994

HITRAN:  L.S.Rothman, R.R.Gamache, R.H.Tipping, C.P.Rinsland, M.A.H.Smith, D.C.Benner, V.Molathy Devi, J-M.Flaud, C.Camy-Peyret, A.Perrin, A.Goldman, S.Massie,L.R.Brown, and R.A.Toth, "The HITRAN Molecular Database: Editions of 1991 and 1992," Journal of Quantitative Spectroscopy and Radiative Transfer 48, 469-507, 1992

PLOD:  S.Hannon, L.L.Strow, and W.W.McMillan, "Atmospheric infrared fast transmittance models: A comparison of two approaches," in Proceedings of SPIE Conference 2830, Optical Spectroscopic Techniques and Instrumentation for Atmospheric and Space Research II (12pp.), 1996


Statistical Retrieval using synthetic VAS brightness temperatures

(1) Sounding database

The basis for the statistical retrieval procedure is 14 months (Dec 96 -Jan 98) of radiosonde data from the region 16-60N, 60-130W. The data were subjected to rigorous quality control and grouped in three-month overlapping sets: Dec-Jan-Feb, Jan-Feb-Mar,..., Nov-Dec-Jan. The number of soundingsranges from 11921 in Feb-Mar-Apr to 12931 in Jul-Aug-Sep, with the average around 12400.

(2) Synthesis of radiometric data

The "forward model" (fast transmittance algorithm plus related radiative-transfer code) contained within VASFRWRD.F was applied to each of the 12 sounding-data files, for each VAS instrument (GOES-4 through -7), for the same 6 zenith angles utilized in the generation of the transmittance model (see Transmittance Modeling for VAS).

(3) Regression procedure

The brightness temperatures from (2) were then regressed against the temperature and mixing ratio profiles described in (1), with secant(theta) as an additional "channel", producing the set of files vasregtw.m01,...,vasregtw.m12 where Dec-Jan-Feb is designated m(onth)01, Jan-Feb-Mar is m02,..., and Nov-Dec-Jan is m12. Each file contains coefficient sets for the four satellites. The contents of these 12 files were then reorganized for efficiency of access and combined into the single file vasregtw.dat which is entered from subroutine "vasrtw" according to zenith angle, month, and satellite number.


Harold M. Woolf
Space Science and Engineering Center
University of Wisconsin-Madison
1225 West Dayton Street,
Madison, WI 53706, U.S.A.
hal.woolf@ssec.wisc.edu


Comments to:  gvp@ssec.wisc.edu
Home Page   Top    User Guide