Getting Started with LARS-WG

Canadian Climate Data and Scenarios

LARS-WG is a stochastic weather generator developed by Dr. Mikhail Semenov of Rothamsted Research, UK. A stochastic weather generator is a statistical model which is able to simulate daily weather data based on the observed statistical characteristics of weather at a single site. This type of statistical model can:

  • generate long time series of weather data (typically precipitation, maximum and minimum temperature and solar radiation) suitable for the assessment of agricultural and hydrological risk;
  • provide a means for simulating weather data at locations where no observations have been made, or where the observational record is very short; and
  • serve as a computationally-inexpensive tool which can produce high temporal resolution climate change scenario data. This particular model is able to incorporate changes in climate variability as well as changes in mean climate.

Where can I get LARS-WG?

You can download the software and user manual for free from:
http://www.rothamsted.bbsrc.ac.uk/mas-models/larswg.php

How do I use LARS-WG?

  1. Download the software from the LARS-WG web site into a directory of your choice on your computer. To install the software, simply double-click on the UpdateLARSWG.exe file in this directory and follow the instructions in the installation wizard.

  2. To start LARS-WG 4.0 for the first time, go to the c:\Program Files\LARSWG directory (or to the appropriate directory if you installed it somewhere else). Here you will find a butterfly icon which you can move to a more convenient location (e.g. the START menu, or to the Desktop) for easier access to the program if you so wish. Double-click on this icon to start LARS-WG 4.0. The following window should appear:

    LARS-WG 4.0 window
  3. The first step in using any stochastic weather generator is statistical analysis of observed weather data for the station in question. In LARS-WG 4.0, click on the Analysis menu in the top left-hand corner of the main window and select Site Analysis from the drop-down menu that appears. LARS-WG 4.0 comes complete with two demonstration data sets; one for Debrecen (Hungary) and the other for Rothamsted (UK) and you can familiarize yourself with LARS-WG 4.0 by working with either one of these data sets. Note that the user manual (which has not yet been updated from LARS-WG 3.0) uses Debrecen for the demonstration data set and this site is also used as the default in LARS-WG 4.0. If you wish to use a different site, simply enter the correct directory path and file name details in the Site Analysis window.

    Click on the Run Site Analysis button to implement statistical analysis of the observed weather data set (moving the mouse cursor over the buttons at the bottom of this window will identify the correct button). As part of the analysis process, LARS-WG 4.0 undertakes a simple quality control of the input data and flags values which it considers to be suspect. If this is the case, you have the opportunity to view the possible errors and, if possible, you can go back to the original data set and make the necessary corrections. If you have no way of knowing what the correct data values should be, then LARS-WG 4.0 automatically treats the suspect data as missing values and they are not included in the analysis process. The Site Analysis process produces two parameter files which are required by LARS-WG 4.0 to enable it to generate synthetic weather data for the site in question. These parameter files are stored in the Site Analysis directory listed in the Options menu in the main window. More details about these files can be found in the user manual.

  4. You should check how well LARS-WG 4.0 performs in simulating weather at the site in question by comparing the statistics of the original observed data with those of synthetic data generated by LARS-WG 4.0. To do this, click on the Analysis menu and select QTest from the drop-down menu that appears. Select the name of the site you are examining and then click on the Run Q-test button on the right-hand side of this window (the name of your site will only appear on this menu if you have completed the Analysis of the observed weather data, as outlined in (3) above). Once the QTest is complete, you are given the opportunity to view the results. Explanations of these results are given in both the user manual and by clicking on the Help button on the main window. From these explanations, you can decide whether or not LARS-WG 4.0 performs sufficiently well to continue with its use. If LARS WG 4.0 performs poorly, some reasons why this may have occurred are also given and it may be necessary for you to make corrections to the input data (e.g. remove trends in the data if they exist) and then re-run LARS-WG 4.0.

  5. Once you are happy with the performance of LARS-WG 4.0, you can go on to generate synthetic weather data. Click on Generator on the main window, and then on Site on the drop-down menu that appears. To generate synthetic data you need a scenario file. The default scenario file is called base.sce and is located in the c:\LARSWG\Sitebase directory. The base.sce file is set up so that no changes are applied to the data. You would use this file as it stands if you simply wish to generate a much longer time series which has the same statistical characteristics as the observed data set. However, if you wish to apply changes to the observed data set, e.g. according to a particular climate change scenario, then you need to edit and rename this file. For details of how to make changes to this file for a particular climate change scenario refer to the Help pages or to the user manual (note that the user manual will refer to changes in mean temperature only – this has been updated in LARS-WG 4.0 and the model is now able to use changes in maximum and minimum temperature).

    You can edit the base.sce file either by clicking on the Edit Scenario button on the right-hand side of the Site Scenario window, or by using NotePad or WordPad. Remember to rename the base.sce file when you create a climate change scenario file! If you want to generate daily data for a climate change scenario, make sure that the correct directory path and file name are listed in the Scenario File window on the Site Scenario window. You can also enter the number of years of data you want generated and select a random number seed in this window. Changing the random number seed will result in the generation of a different synthetic data set. For more details about the number of years of data to generate and on the random number seed, see the user manual. Click on the Run Generator button on the right-hand side of the Site Scenario window to generate synthetic weather data for your scenario. The output file containing the synthetic scenario data is located in the c:\LARSWG\Output directory.

Note: The options to generate data for a particular region and for interpolation are disabled in this version of LARS-WG and therefore it can be used only for site analysis.

How do I prepare my own data for use in LARS-WG?

To use LARS-WG 4.0 for a site of your choice you need to prepare two files, one containing the daily weather data and the other details of the site. Create a new directory for your site in the c:\LARSWG\Data directory and put these two files in it. The daily data file should be in column format with each row of the data file representing values for a particular day. LARS-WG 4.0 will work with precipitation data alone, or with precipitation in combination with one, or all, of maximum temperature, minimum temperature, solar radiation or sunshine hours. Since the weather generator is conditioned on the length of wet and dry spells, precipitation must be present for the weather generator to work. The file describing the site location should contain the site name, its latitude, longitude and altitude (if available), the name of the data file (if the data file is not in the same directory, include the full directory path for the data file) and the format of the data (i.e. tags detailing which data is in which column). Have a look at the example data sets for Debrecen and Rothamsted or see the Help pages or the user manual for more details. Then proceed to analyze the data for your site according to the instructions listed above.

Top of Page


Further Reading

A more thorough introduction to using LARS-WG may be found in its user manual which can be downloaded from: http://www.rothamsted.bbsrc.ac.uk/mas-models/larswg.php.

More details about the model structure of LARS-WG, its development and performance can be found in the following references:

Dibike Y.B. and Coulibaly P. (2005): Hydrologic Impact of Climate Change in the Saguenay Watershed: Comparison of Downscaling Methods and Hydrologic Models, Journal of Hydrology 307(1-4): 145-163.

Gachon P., St-Hilaire A., Ouarda T.B.M.J., Nguyen V.T.V., Lin C., Milton J., Chaumont D., Goldstein J., Hessami M., Nguyen T.D., Selva F., Nadeau M., Roy P., Parishkura D., Major N., Choux M., Bourque A. (2005): A first evaluation of the strength and weaknesses of statistical downscaling methods for simulating extremes over various regions of eastern Canada. Sub-component, Climate Change Action Fund (CCAF), Environment Canada, Montreal, Quebec, Canada, 209 pp. (available from the 1st author).

Hayhoe H.N. (2000): Improvements of stochastic weather data generators for diverse climates. Climate Research 14: 75-87.

Katz R.W. (1996): Use of conditional stochastic models to generate climate change scenarios. Climatic Change 32 (3): 237-255.

Khan M.S., Coulibaly P. and Dibike Y., 2006. Uncertainty Analysis of Statistical Downscaling Methods, Journal of Hydrology 319(1-4):357-382.

Qian B.D., Hayhoe H., and Gameda S. (2005): Evaluation of the stochastic weather generators LARS-WG and AAFC-WG for climate change impact studies. Climate Research 29(1): 3-21.

Qian B.D., Gameda S., Hayhoe H., De Jong R., and Bootsma A. (2004): Comparison of LARS-WG and AAFC-WG stochastic weather generators for diverse Canadian climates. Climate Research 26: 175-191.

Racsko P., Szeidl L., and Semenov M.A. (1991): A serial approach to local stochastic weather models. Ecological Modelling 57: 27-41.

Richter G.M. and Semenov M.A. (2005): Modelling impacts of climate change on wheat yields in England and Wales – assessing drought risks. Agricultural Systems 84(1): 77-97.

Semenov M.A. (2006): Using Weather Generators in Crop Modelling, Acta Horticultura 707: 93-100.

Semenov M.A. and Barrow E.M. (1997): Use of a stochastic weather generator in the development of climate change scenarios. Climatic Change 35: 397-414.

Semenov M.A., Brooks R.J., Barrow E.M. and Richardson C.W. (1998): Comparison of the WGEN and LARS-WG stochastic weather generators for diverse climates. Climate Research 10: 95-107.

Semenov M.A. and Brooks R.J. (1999): Spatial interpolation of the LARS-WG stochastic weather generator in Great Britain. Climate Research 11: 137-148.

Wilks D.S. and Wilby R.L. (1999): The weather generation game: a review of stochastic weather models. Progress in Physical Geography 23: 329-357.

Top of Page