Author(s): Shaik Rehana; P. P. Mujumdar
Linked Author(s):
Keywords: GCM; Statistical Downscaling; Uncertainty; Entropy Weighting; Canonical Correlation Analysis
Abstract: Climate change impact assessment models developed based on General Circulation Model (GCM) outputs are subjected to a range of uncertainties due to incomplete knowledge, inadequate information and understanding about the geophysical processes of global change. This leads to significant dissimilarity between the projections derived with multiple GCMs and scenarios which can be combinedly represented as climate model uncertainty in the assessment of climate change impacts on hydrological variables. Climate model uncertainty can be characterized as epistemic nature which is due to imperfect knowledge and can be reduced by gaining more information about the system. The present study attempts to quantify the epistemic nature of climate model uncertainty by acquiring additional information using the entropy information theory to assign weights to the ensembles of projections from various GCMs and scenarios. Various hydro-meteorological variables (streamflow, rainfall, maximum and minimum temperatures, relative humidity and wind speed) of Bhadra river basin, in India, are modeled by multimodel/scenario weighted mean using entropy weighting approach. A downscaling method based on Canonical Correlation Analysis (CCA) is applied to model the multivariable hydro-meteorological variables from three GCMs, CGCM2 (Meteorological Research Institute, Japan), MIROC3. 2 medium resolution (Center for Climate System Research, Japan), and GISS model E20/Russell (NASA Goddard Institute for Space Studies, USA) with three green house emission scenarios of A1B, A2 and B1. Entropy weights are assigned to all the GCMs with scenarios based on their performance in modeling the hydro-meteorological variables in the recent past (1992-2005), and deviation of each GCM and scenario from the projected ensemble average for the period of 2020 -2060. The proposed entropy weighting method is applied to the projections obtained from CCA for each hydro -meteorological variable from various GCMs and scenarios. The resulting weighted annual mean streamflow shows a reduction of 6. 2% , with an increase in rainfall as 16. 66 % for Bhadra river basin for 2020 -2060 when compared to the observed period of 1972-1992. The weighted mean minimum and maximum temperature increase is predicted as 1. 61 0C and 3. 64 0C respectively and the weighted mean relative humidity increase as 0. 79 % , the weighted mean wind speed increase as 17. 44 % for 2020 -2060 when compared with the observed period of 1972-1992. Such weighted mean projections of hydro-meteorological variables are particularly useful in the decision making, where the multimodel ensembles have limited use and appropriate single aggregated ensemble resulting best estimates of climate change is required.
Year: 2013