Author: V. Eymet
LAboratoire PLAsmas et Conversion de l'Energie (LAPLACE)
Université Paul Sabatier, bat. 3R1
31062 TOULOUSE Cedex 9 - France
KARINE is a collaborative project started by the Laboratoire d'Enérgetique (Toulouse), the Ecole des Mines d'Albi-Carmaux (EMAC) and the Laboratoire de Météorologie Dynamique (Jussieu, Paris). Contributors to the project include Richard Fournier (LE), Stéphane Blanco (LE), Jean-Louis Dufresne (LMD), Amaury de Lataillade (EMAC), Patrice Perez (EMAC), David Joseph (EMAC), Maxime Roger (EMAC), Cyril Caliot (EMAC), Mouna El Hafi (EMAC).
KARINE stands for "K-distribution Atmospheric Radiation: Infrared Net Exchanges". It is a based on a Monte-Carlo algorithm, optimized for fast computations. It will compute atmospheric radiative transfer (radiative budgets, radiative fluxes and all Net Exchange Rates) in the infrared part of the spectrum. KARINE uses a k-distribution spectral model, along with the CK hypothesis for taking into account inhomogeneities.
+ Radiative budgets, fluxes and Net Exchange Rates analysis.
+ Reference results: all results are given with an estimation of a numerical uncertainty -- in fact, the statistical standard deviation. This error can be decreased when the number of statistical realizations is increased.
+ Radiative budgets can be decomposed as follows:
- part of the budget due to net exchanges with the ground
- part of the budget due to net exchanges with space
- part of the budget due to net exchanges with the rest of the atmosphere
+ Every result can be computed for each spectral narrowband.
+ All results are computed analytically -- analytic results are only available for purely absorbing atmospheres. KARINE uses these analytic results as a basis for spectral sampling. Note for version 3.4 and higher: analytic computations have been extented to sensitivities of NERs to temperature and species concentrations.
+ The user can use many options to tweak the behaviour of the Monte-Carlo algorithm, increase the accuracy of the results, etc.
+ Possibility to compute only the analytical results (faster).
+ Possibility to resume an interrupted computation.
+ Visualization scripts are provided.
What it is not:
+Fast: even if the Monte-Carlo algorithm has been optimized for fast computations (see this paper or my thesis), a typical run that will estimate radiative budgets with a ~0.1% error will take around ~12h of CPU time on a modern system (Linux running on a ~2GHz processor). The standard deviation will decrease as the inverse square root of the computation time; it means that getting results 10 times more accurate will require a computation a 100 times longer.
+KARINE will NOT guess the radiative properties of the atmosphere. These have to be included in an input data file (see the documentation).
+This code is not intended to be used for radiative computations coupled with other heat transfer mechanisms. KARINE is a research tool, and the reference results it provides are mean to be used as a basis for precise analysis.
This program has already been used on:
+ fedora core linux
+ mandrake 10.1 linux
+ windows XP with cygwin
+ various flavours of linux redhat, mandrake and debian.
+ Ubuntu, of course.
It should be able to run on most linux and unix platforms with the GNU fortran 77 compiler.
KARINE is distributed under the GNU General Public Licence. See the terms of the GPL.
You should consider saving this file under an empty directory, then decompress it using the following command: 'tar -zxvf karine.tgz'. This will create a new directory named "karine".
No specific installation is required: the user should be able to compile the main executable file using the command 'make all' (used in directory "karine"). Once the compilation is complete, the executable file "karine" should appear.
Before running "karine", the user should create at least one spectral data file. By default, this file has to be created in "karine/data". Use the following quick steps to test the procedure:
* go into "karine/data" (type 'cd data' in directory "karine")
* compile the fortran program "make_data.exe" (type 'f0') -> the executable file "make_data.exe" should be created
* run "make_data.exe" (type './make_data.exe') -> a simple test file "atmos_data.txt" will be created.
The next step is to go back into the main directory ("karine") and use the command './karine' in order to run karine for the first time.
Consider reading the user manual ("karine/Doc/manual.pdf") for more detailed information.
The user manual is available in pdf or postscript:
Download the pdf version (332kb)
Download the postscript version (496kb)
If you have not already done so, here is the last version of karine:
Modification date: 09/06/2009 (version 3.4.5)
Before downloading this software, you should first agree with the terms of the GNU General Public License.
You should also consider reading the installation steps and the documentation files before using the code.
Previous versions (history):
Please send me an email if you wish to get one of those previous versions.
See the changelog file for what has been modified between the various versions.
KARINE_light: since version 3.4 of KARINE, analytic computation include the calculation of analytic sensitivities. Which requires the declaration of huge arrays, and may result in compilation failures for low-memory systems and/or high numbers of atmospheric layers/narrowband spectral intervals/quadrature levels. KARINE_light is based on the current version of KARINE, except that all source code related to analytic sensitivities have been commented (and therefore, analytic sensitivities can not be computed !). The archive can be downloaded here.
MARINE : the "M" stands for "monochromatic". I have extracted the monochromatic Monte-Carlo algorithm upon which KARINE was built, and used it to create MARINE. This tool is intended to be used for classical monochromatic radiative transfer analysis in slab medium. This is an easier tool to use than KARINE.
Visit the presentation page of MARINE.
Analytic computations : the codes provided on this page are based on analytic radiative transfer solutions for purely absorbing, plane-parallel inhomogeneous atmospheres.
Screenshot number 1: this is a 1024*768 image (232kb). You can see the code running in a terminal window (under Cygwin). For those who wonder, the wallpaper has been derived from the animation series "Vandread".
Screenshot number 2: (220kb) shows an example of result visualization (Net Exchange Rate matrix) using the Grads scripts that are provided in the main KARINE archive.
Screenshot number 3: (260kb) shows an example of result visualization (difference between two Net Exchange Rate matrices, integrated over narrowbands index 68 to 80) using the new plot shell script.
Any bug or contribution idea can be submitted to the author (Vincent Eymet) by email