Cligen is a stochastic weather generator which produces daily estimates of precipitation, temperature, dewpoint, wind, and solar radiation for a single geographic point, using monthly parameters (means, SD's, skewness, etc.) derived from the historic measurements. Unlike other climate generators, it produces individual storm parameter estimates, including time to peak, peak intensity, and storm duration, which are required to run the WEPP and the WEPS soil erosion models. Station parameter files to run Cligen for several thousand U. S. sites are available for download from this website: also data and software to build station files for international sites. With the exception of Tmin, Tmax, and Tdew temperatures (changed in January 2004), daily estimates for each parameter are generated independently of the others. With the current random number generator, subsequent runs on the same machine made with identical inputs will produce identical results.
Users of daily simulation models should consider the impacts of Cligen's characteristics on their application. Individual parameter distributions may be expected to reproduce monthly historic distributions quite well. However, if the model in question is sensitive to the daily interactions of two or more of the parameters Cligen produces, Cligen may not be the most appropriate weather generator to use. This is because for a given day, it generates solar radiation, and maximum and minimum temperatures completely independently from precipitation. Experience and common sense tell us that these parameters are NOT independent. In practice this may not be a huge issue, since it is not uncommon for models to be sensitive to one weather parameter on a daily basis, and relatively insensitive to the others, as long as their monthly trends are preserved.
Programs and Source Code
Programs and Source code are available for the following versions:
Cligen Version 4.2 Source Code (Recoded)
Cligen Version 4.2 Source Code (Original)
Stations.zip - All CLIGEN Stations .PAR files. See the readme.html file contained in the zip file for information on how to use these files with Cligen. These files can be used with Cligen versions 4.3 and 5.3.
V42data.zip - All CLIGEN Stations .PAR files compatible with Cligen version 4.2.
Note than when downloading the WEPP Model the Cligen program and data files are already included.
The format of the .PAR input files used in Cligen are described in the document cligenparms.pdf.
How Cligen Works
The following document: CligenDesription.pdf describes the basics of Cligen and the development history.
The revisions to Cligen between versions 4 and 5 are outlined in the slideshow in the document: cligenslds.pdf
Cligen was produced by Arlin Nicks and Gene Gander at the USDA Agricultural Research Service (ARS) lab in Durant, Oklahoma. They made their last significant changes to the model in the mid 1990's before Dr. Nicks' retirement and death in July 1997. David Hall and Dayna Scheele at the USDA Forest Service lab in Moscow, Idaho, acquired code from Dr. Nicks' computer to generate from historical weather data, the monthly station parameter files required to drive Cligen. They mechanically cleaned the existing station data files, and added a large number of new stations for the United States. In the Summer of 1999 Bofu Yu of Griffith University in Australia, expressed concerns that Cligen's rainfall intensity calculations were not operating correctly due to a unit conversion error, and offered revised code to correct the problem. In the Fall of 1999, Charles R. Meyer at the USDA-ARS lab in West Lafayette, Indiana, re-coded the CLIGEN source code to make it understandable, maintainable, and extensible by people who had not written it. He incorporated Dr. Yu's corrections. When performing checks on CLIGEN's uniform random number generator and standard normal generator, Meyer discovered that they were not operating correctly. A paper by Johnson, et. al. (Stochastic Weather Simulation: Overview and Analysis of Two Commonly Used Models; Journal of Applied Meteorology, Vol. 35, October 1996), reports effects which appear to be the result of this problem. This has major implications for any stochastic model like CLIGEN because all its output originates from its random number generator. To correct this problem Meyer introduced a form of "quality control" borrowed from industrial engineering leading to Version 5 of Cligen.