Minutes of progress meeting 7, 17 December 2009, LMU Munich, 8:30 - 16:00

Participants

  • Marc Bouvet (MB), ESA-ESTEC
  • Robert Buras (RB), LMU
  • Claudia Emde (CE), DLR-IPA
  • Arve Kylling (AK), Ålesund
  • Bernhard Mayer (BM), LMU
  • Franz Schreier (FS), DLR-IMF
  • Ulrike Wissmeier (UW), LMU

Agenda

  1. Discussion of ATBDs, verification report and documentation of solvers
  2. Line-by-line tool Py4CATS
  3. Feedback from ESA for libRadtran-2.0-beta
  4. Discussion of recommendations document
  5. General issues

Minutes

Discussion of ATBDs and verification report

Raman scattering

AK presented the progress regarding the Raman scattering implementation. The problem that has been shown at the last PM for Raman scattering in presence of clouds has been solved. Now a good agreement with published results has been found for clear sky cases, cases with aerosol scattering and also for thick cloud cases. There is one disagreement to a result published by Spurr et al. (2008) for Raman scattering in a thin cloud. Here it is not clear which of the models that are right. Final clarification is only possible by a model-by-model intercomparison for which Spurr has been contacted by AK.

Comparisons to measurements by Mario Blumthaler have been performed and there is a good agreement for the few cases where the Raman signature is clearly visible in the measured irradiance spectra.

MB asked which solver includes Raman scattering. AK clarified with reference to the “documentation of solvers” document that it is included in the solver “qdisort”.

MB suggested to include a summary for Raman scattering in the verification report.

Polarization in MYSTIC

CE presented the final ATBDs and the verification report for the implementation of polarization in MYSTIC. The algorithm is fully implemented and successfully validated by comparison with published benchmark results, by comparison with measurements, and by model-by-model intercomparisons. Also some analytical solutions have been verified. Since the last PM one new feature has been included regarding polarization, namely the polarized reflectance matrix which is required to describe polarized surface reflection, for instance in the sun glint region.

Documentation of solvers document and user's guide

AK presented the “documentation of solvers” which consists of a theoretical part describing the radiative transfer equation and various solution approaches, and a more practical part describing the libRadtran solvers and which of them are useful for which application. MB suggested to put the second part also into the user's guide. The first part of the libRadtran user's guide should be rewritten, it should start with the documentation of solvers, it should include a guide of how to setup an input file, some common applications, and finally the references for all options. Actually the first part of the user's guide is outdated and includes applications that are not common for libRadtran users.

Test suite for libRadtran

RB presented the first version of the test suite for libRadtran. This test suite runs a large number of tests (currently 110, will be extended) and compares the results to pre-calculated results from a stable libRadtran version. The test suite runs continuously always using the most recent development version of libRadtran, thus allowing to rapidly discover undesired changes and bugs and to eliminate them as quickly as possible.

Line-by-line tool Py4CATS/lbl2od

FS presented a general introduction to line-by-line and thermal radiative transfer calculations and his line-by-line tools that are now available for libRadtran. The tools are implemented in python. The most important tools are “extract” which extracts lines of selected molecules (and isotopes) for a specified wavenumber range from line parameter databases, e.g. HITRAN or GEISA, and “lbl2od” which computes the line-by-line optical depth due to molecular absorption. The line-by-line optical depth is required for high spectral resolution radiative transfer calculations using libRadtran. The line-by-line tools will be offered as an additional tool for libRadtran on the libRadtran webpage.

Feedback for libRadtran-2.0beta

MB have tested the new libRadtran version (2.0beta) and provided the following feedback:

  • It can be difficult to install various required packages (netcdf, python, gsl) and it would be nice to have an executable for windows.
  • The name “uvspec” should be changed. It is misleading because “uvspec” is used for the full spectrum from the UV to the thermal infrared.
  • The section “some useful tools” in the user's guide should be renamed
  • The MYSTIC documentation which is currently only available as separate README files should be put into the user's guide
  • The “cookbook” on the webpage should be extended (it should include also the setup for MYSTIC)
  • The first image when the GUI starts should be exchanged, it should show a typical libRadtran result
  • An ESA logo could be put in the GUI
  • The GUI is not yet compatible with MYSTIC
  • The GUI is not yet “idiot-proof”
  • Documentation format in GUI could be improved
  • The plots should include units
  • The extraterrestrial spectrum by Thullier which is commonly used at ESA should be added to libRadtran
  • The GUI should have a “stop” button to cancel simulations that take too much time
  • For calculations in the thermal spectral range one has to specify “source thermal”, this can easily be forgotten in the GUI
  • Surface temperature range in GUI is wrong
  • There should be a better correlated-k method because LOWTRAN is not so accurate
  • Pre-calculated line-by-line files for standard atmospheres should be made available on the web-page
  • Instrument response functions (e.g. for MODIS, LandSat, ATSR, MERIS … ) could be included.
  • Solvers that are not good should be removed or they should be hidden
  • The output files should include documentation headers
Discussion of recommendations document

MB gave priorities to the suggestions that have been made in the recommendations document. In the follow-up ESA study the main priority should be to improve the user friendlyness of libRadtran. A lower priority is given to the development of new algorithms.

The following topics are of high priority for ESA:

  • Improve GUI
  • Speed up Raman scattering
  • Include a new 1D polarized RT solver
  • New correlated-k scheme to replace LOWTRAN
  • Cleaning and restructuring of code and documentation
  • Dissemination of libRadtran / organization of a workshop
Review of action items, date of next meeting

Action items:

  1. AK: fix GUI problems
  2. CE: send list of recommendations to MB
  3. AK: include summary for Raman scattering into verification report
  4. AK: include first part of documentation of solvers to user's guide
  5. all developers: rewrite first part of user's guide
  6. BM: include extraterrestrial spectrum by Thullier into libRadtran

The final meeting will take place at the 11th of March 2010 at ESTEC, Noordwijk.

 
minutes_20091217.txt · Last modified: 2009/12/22 10:39 by esaslight     Back to top