Downscaling Tools


Although there is no 'standard' approach to downscaling, (i.e. obtaining finer resolution scenarios of climate change from coarser resolution GCM output), there are two pieces of software currently available which can be used to undertake spatial and temporal downscaling.

A Statistical DownScaling Model, developed by Rob Wilby and Christian Dawson in the UK
A stochastic weather generator developed by Mikhail Semenov and Elaine Barrow, also in the UK.


SDSM permits the spatial downscaling of daily predictor-predictand relationships using multiple linear regression techniques. The predictor variables provide daily information concerning the large-scale state of the atmosphere, whilst the predictand describes conditions at the site scale. The software reduces the task of statistically downscaling daily weather series into a number of discrete processes:

  1. Preliminary screening of potential downscaling predictor variables - identifies those large-scale predictor variables which are significantly correlated with observed station (predictand) data. A number of variables derived from mean sea level pressure fields are included, e.g. air flow strength, meridional and zonal components of air flow, vorticity etc. (see Statistical Downscaling Input, in Download Data section);
  2. Assembly and calibration of statistical downscaling model(s) - the large-scale predictor variables identified in (1) are used in the determination of multiple linear regression relationships between these variables and the local station data. Statistical models may be built on a monthly, seasonal or annual basis. Information regarding the amount of variance explained by the model(s) and the standard error is given in order to determine the viability of spatial downscaling for the variable and site in question;
  3. Synthesis of ensembles of current weather data using observed predictor variables - once statistical downscaling models have been determined they can be verified by using an independent data set of observed predictors. The stochastic component of SDSM allows the generation of up to 100 ensembles of data which have the same statistical characteristics but which vary on a day-to-day basis;
  4. Generation of ensembles of future weather data using GCM-derived predictor variables - provision of the appropriate GCM-derived predictor variables allows the generation of ensembles of future weather data by using the statistical relationships calculated in (2);
  5. Diagnostic testing/analysis of observed data and climate change scenarios - it is possible to calculate the statistical characteristics of both the observed and synthetic data in order for easy comparison and thus determination of the effect of spatial downscaling.

SDSM screenshot

To access predictor data, please visit the CMIP5 or TAR/AR4 Statistical Downscaling Input pages.


LARS-WG, based on the serial approach, is one of the most readily available stochastic weather generators.

LARS-WG screenshot

The software is user-friendly and is supported by a relatively comprehensive help system. The software consists of three main sections:

the first step in the weather generation process is the analysis of the observed station data in order to calculate the weather generator parameters, i.e. the statistical characteristics of the data. LARS-WG requires observations for precipitation and one or all of maximum and minimum temperature and sunshine hours (or solar radiation; if sunshine hours are supplied the data are converted using an algorithm based on that by Rietveld, 1978). The analysis process uses semi-empirical distributions, i.e. frequency distributions calculated from the observed data, for wet and dry series duration, precipitation amount and solar radiation. Maximum and minimum temperature are described using Fourier series. The resulting parameter file is then used in the generation process.
LARS-WG generates synthetic weather data by combining a scenario file containing information about changes in precipitation amount, wet and dry series duration, mean temperature, temperature variability and solar radiation with the parameter files generated in step (1). If LARS-WG is being used to generate synthetic data in order to determine how well the model is simulating observed conditions, or to simulate a longer time series of data for a station with only a short observational record, then the scenario file contains no changes. However, if LARS-WG is being used to generate daily data for a particular scenario of climate change, then the scenario file will contain the appropriate monthly changes.
LARS-WG simplifies the procedure for determining how well it is simulating observed conditions by providing the Qtest option. In this step, the statistical characteristics of the observed data are compared with those of synthetic data generated using the parameters derived from the observed station data. A number of statistical tests, the chi-squared test, Student's t-test and the F-test, are used to determine whether the distributions, mean values and standard deviations, respectively, of the synthetic data are significantly different from those of the original observed data set.

The main use of LARS-WG is in the generation of daily data from monthly climate change scenario information. The advantage of using a stochastic weather generator rather than simply applying the scenario changes to an observed daily time series is that a number of different daily time series representing the scenario can be generated by using a different random number to control the stochastic component of the model. Hence, these time series all have the same statistical characteristics, but they vary on a day-to-day basis. This permits risk analyses to be undertaken.

References for SDSM (tool and predictors)

Barrow, E., B. Maxwell and P. Gachon, 2004: Climate Variability and Change in Canada: Past, Present and Future, Climate Change Impacts Scenarios Project, National Report, Environment Canada, Meteorological Service of Canada, Adaptation Impacts Research Group, Atmospheric and Climate Sciences Directorate publication, Canada, 114 pp, ISBN: 0-662-38497-0.

Choux M., (2005): Development of new predictor variables for the statistical downscaling of precipitation. Degree Master of Engineering, Department of Civil Engineering and Applied Mechanics, McGill University. (Dec. 2005).

Conway, D., Wilby, R.L. and Jones, P.D. (1996): Precipitation and air flow indices over the British Isles. Climate Research 7: 169-183.

Dibike, Y., P. Gachon, A. St-Hilaire, T.B.M.J. Ouarda, and VTV Nguyen, 2007: Uncertainty analysis of statistically downscaled temperature and precipitation regimes in northern Canada. Theoretical and Applied Climatology (in press).

Gachon, P., A. St-Hilaire, T. Ouarda, VTV Nguyen, C. Lin, J. Milton, D. Chaumont, J. Goldstein, M. Hessami, T.D. Nguyen, F. Selva, M. Nadeau, P. Roy, D. Parishkura, N. Major, M. Choux & A. Bourque, 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, Final report, Montréal, Québec, Canada, 209 pp.

Goldstein, J., J. Milton, N. Major, P. Gachon, and D. Parishkura, 2004: Climate extremes indices and their links with future water availability: Case study for summer of 2001, article published in the proceeding of the 57th Annual Conference of the Canadian Water Resources Association. Montréal, Canada, June 16-18 2004, 7pp.

Hassan, H., Aramaki, T., Hanaki, K., Matsuo, T. and Wilby, R.L. (1998): Lake stratification and temperature profiles simulated using downscaled GCM output. Journal of Water Science and Technology 38: 217-226.

Hessami M., T.B.M.J. Ouarda, P. Gachon, A. St-Hilaire, F. Selva and B. Bobée, 2004: Evaluation of statistical downscaling methods over several regions of eastern Canada, article published in the proceeding of the 57th Annual Conference of the Canadian Water Resources Association. Montréal, Québec, Canada. June 16-18, 2004, 9 pp.

Jones, P.D., Hulme, M. and Briffa, K.R. (1993): A comparison of Lamb circulation types with an objective classification scheme. International Journal of Climatology 13: 655-663.

Kalnay, E., Kanamitsu, M., Kistler, R. et al. (1996): The NCEP/NCAR 40-year reanalysis project. Bulletin of the American Meteorological Society 77: 437-471.

Nguyen T., V.T.V. Nguyen, P. Gachon and A. Bourque, 2004a: An assessment of statistical downscaling methods for generating daily precipitation and temperatures extremes in the greater Montréal region, article published in the proceeding of the 57th Annual Conference of the Canadian Water Resources Association. Montréal, Québec, Canada. June 16-18, 2004, 10 pp.

Nguyen VTV, Nguyen TD, Gachon P. 2004b: An Evaluation of Statistical Downscaling Method for Simulating Daily Precipitation and Extreme Temperature Series at a Local Site, 14th Congress of the APD-International Association of Hydraulic Engineering and Research, Hongkong, December 15-18, 2004, pp. 1911-1916.

Nguyen VTV, Nguyen TD, Gachon P., 2006: On the linkage of large-scale climate variability with local characteristics of daily precipitation and temperature extremes: an evaluation of statistical downscaling methods. Advances in Geosciences (WSPC/SPI-B368) 4(16): 1-9.

Nguyen, T-D., V-T-V. Nguyen, and P. Gachon, 2007: A spatial-temporal downscaling approach for construction of intensity-duration-frequency curves in consideration of GCM-based climate change scenarios, in 'Advances in Geosciences, Vol. 6: Hydrological Sciences', N. Park et al. (Eds.), World Scientific Publishing Company, pp. 11-21.

Wilby, R.L. and Dettinger, M.D. (2000): Streamflow changes in the Sierra Nevada, CA, simulated using a statistically downscaled General Circulation Model scenario of climate change. In: Linking Climate Change to Land Surface Change, McLaren, S.J. and Kniveton, D.R. (Eds.), Kluwer Academic Publishers, Netherlands, pp. 91-121.

Wilby, R.L., and Wigley, T.M.L., (2000): Precipitation predictors for downscaling: observed and General Circulation Model relationships. International Journal of Climatology 20: 641-661.

Wilby, R.L., Dawson, C.W. and Barrow, E.M. (2002): SDSM - a decision support tool for the assessment of regional climate change impacts. Environmental and Modelling Software 17: 145-157.

Wilby, R.L., Hassan, H. and Hanaki, K. (1998): Statistical downscaling of hydrometeorological variables using general circulation model output. Journal of Hydrology 205: 1-19.

References for LARS-WG

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

Rietveld, M.R. (1978): A new method for estimating for regression coefficients in the formula relating solar radiation to sunshine. Agricultural and Forest Meteorology 19: 243-252.

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 Barrow, E.M. (2000): Development of climate change scenarios for agricultural applications. In: Climate scenarios for agricultural, forest and ecosystem impacts, Cramer, W., Doherty, R., Hulme, M. & Viner, D. (Eds), ECLAT-2 Workshop Report No. 2, Climatic Research Unit, Norwich, UK, pp. 50-58. Available for download from the ECLAT-2 Project web site.

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.

Date modified: