Statistical downscaling of general circulation model outputs to precipitation

Victoria suffered a severe drought over the period 1998-2007, when the annual average precipitation plunged by about 13% from the long term average. Precipitation is directly related to the availability of water resources in a catchment. Therefore it is useful to predict precipitation, particularly in light of any future climate change, which will help in the management of water resources at the catchment level. General Circulation Models (GCMs) are considered to be the most advanced tools available for simulating the future climate. Due to the coarse spatial resolution, however, GCM outputs cannot be used directly at the catchment scale. To overcome this problem statistical and dynamic downscaling techniques have been developed. Downscaling techniques link the coarse GCM outputs to catchment scale hydroclimatic variables. The present research has focussed on statistically downscaling monthly NCEP/NCAR reanalysis outputs to monthly precipitation at the catchment level. A precipitation station in the operational area of the Grampians Wimmera Mallee Water Corporation (GWMWater) in northwestern Victoria was considered as the case study. Multi-linear regression was used in the development of the downscaling models. This research employed separate downscaling models for each calendar month, with the intention of better capturing the seasonal variations of precipitation. A set of probable predictors were selected following the past literature and hydrology. Data for the probable predictors and precipitation were split into three 20 year time slices; 1950-1969, 1970-1989 and 1990-2010. The probable predictors which displayed the best statistically significant correlations consistently with precipitation over the three time slices and the whole period of the study were selected as potential predictors, for each calendar month. These potential predictors were introduced to the downscaling model one at a time based on the strength of the correlation, over the whole period of the study, until the model performance, in terms of Nash-Sutcliffe Efficiency (NSE), was maximised. This approach ensured the identification of the best potential predictor for each calendar month. In calibration and validation, the model displayed good performances with NSEs of 0.74 and 0.70 respectively. In calibration, the average precipitation was perfectly reproduced by the model and in validation it was slightly over-predicted. However, both in calibration and validation, the model tended to under-predict high precipitations and over-predict near-zero precipitations. A graphical comparison of observed precipitation, downscaling model reproduced precipitation and the Hadley Centre Coupled Model version 3 GCM (HadCM 3) simulated raw precipitation output, revealed that there is large bias in the HadCM 3 precipitation outputs. Therefore, before producing any future precipitation projections with the downscaling model, a bias correction to GCM outputs is prescribed.

[1]  Rowan Fealy,et al.  Comparison of ERA‐40, ERA‐Interim and NCEP/NCAR reanalysis data with observed surface air temperatures over Ireland , 2011 .

[2]  Pao-Shan Yu,et al.  Statistical downscaling of daily precipitation using support vector machines and multivariate analysis , 2010 .

[3]  E. Maurer,et al.  Utility of daily vs. monthly large-scale climate data: an intercomparison of two statistical downscaling methods , 2007 .

[4]  Christian W. Dawson,et al.  Multi-site simulation of precipitation by conditional resampling , 2003 .

[5]  Ram S. Gupta.  Hydrology and hydraulic systems , 1989 .

[6]  J. Nash,et al.  River flow forecasting through conceptual models part I — A discussion of principles☆ , 1970 .

[7]  Neville Nicholls,et al.  Australian Climate and Weather Extremes: Past, Present and Future , 2008 .

[8]  A. Bárdossy,et al.  Statistical downscaling of extremes of daily precipitation and temperature and construction of their future scenarios , 2008 .

[9]  Ashish Sharma,et al.  A nesting model for bias correction of variability at multiple time scales in general circulation model precipitation simulations , 2012 .

[10]  Ravi S. Nanjundiah,et al.  Downscaling precipitation to river basin in India for IPCC SRES scenarios using support vector machine , 2008 .

[11]  Christian W. Dawson,et al.  SDSM - a decision support tool for the assessment of regional climate change impacts , 2002, Environ. Model. Softw..

[12]  I. Smith,et al.  Refining rainfall projections for the Murray Darling Basin of south-east Australia—the effect of sampling model results based on performance , 2010 .

[13]  Z. Li,et al.  Generalization of a statistical downscaling model to provide local climate change projections for Australia , 2009, Environ. Model. Softw..

[14]  D. A. Sachindra,et al.  Least square support vector and multi‐linear regression for statistically downscaling general circulation model outputs to catchment streamflows , 2013 .

[15]  D. A. Sachindra,et al.  Statistical downscaling of general circulation model outputs to precipitation—part 2: bias‐correction and future projections , 2014 .

[16]  Yi Wang,et al.  Validation of the weather generator CLIGEN with daily precipitation data from the Loess Plateau, China , 2007 .

[17]  D. A. Sachindra,et al.  Statistical downscaling of general circulation model outputs to precipitation—part 1: calibration and validation , 2014 .

[18]  V. V. Srinivas,et al.  Downscaling of precipitation for climate change scenarios: A support vector machine approach , 2006 .

[19]  James M. Murphy,et al.  An Evaluation of Statistical and Dynamical Techniques for Downscaling Local Climate , 1999 .

[20]  Daniel S. Wilks,et al.  Use of stochastic weathergenerators for precipitation downscaling , 2010 .

[21]  Andrew Barton,et al.  Coping with Severe Drought: Stories from the Front Line , 2011 .

[22]  Daniel S. Wilks,et al.  Simultaneous stochastic simulation of daily precipitation, temperature and solar radiation at multiple sites in complex terrain , 1999 .

[23]  Budong Qian,et al.  Multisite stochastic weather models for impact studies , 2002 .

[24]  S. Charles,et al.  A comparison of three multi-site statistical downscaling models for daily rainfall in the North China Plain , 2013, Theoretical and Applied Climatology.

[25]  T. McMahon,et al.  El Nino/Southern Oscillation and Australian rainfall, streamflow and drought : Links and potential for forecasting , 1998 .

[26]  R. Reynolds,et al.  The NCEP/NCAR 40-Year Reanalysis Project , 1996, Renewable Energy.

[27]  Ravi S. Nanjundiah,et al.  Role of predictors in downscaling surface temperature to river basin in India for IPCC SRES scenarios using support vector machine , 2009 .

[28]  K. Pearson.  Mathematical Contributions to the Theory of Evolution. III. Regression, Heredity, and Panmixia , 1896 .

[29]  Mark A. Nearing,et al.  Concentrated flow erodibility for physically based erosion models: Temporal variability in disturbed and undisturbed rangelands , 2012 .

[30]  T. Ouarda,et al.  Multisite statistical downscaling model for daily precipitation combined by multivariate multiple linear regression and stochastic weather generator , 2012, Climatic Change.