An Examination of Sensitivities in Uncertainty Quantification for Future Climate Projections over India.

Abstract:

Reliable future climate information is a necessary requirement for the scientific and policy making community. Uncertainty due to various sources affects the level of accuracy of climate change projections at different scales, it becomes even complex at regional scale. This study is an attempt to unfold the levels of uncertainty in future climate projections over the Indian Region to add value to the information on mean changes reported in Chaturvedi et al (Curr. Sci.,2012).We examine model projections of temperature and precipitation using output from the CMIP5 database.Using the ‘Reliability Ensemble Averaging’ (REA, Giorgi and Mearns, J. Climate, 2002) and “Upgraded REA” (Xu et al, Clim. Res.2010) methods with some modifications, we examine the uncertainty in projections for Annual, Indian Summer Monsoon (JJA) and Winter (DJF) seasons under the RCP4.5 and RCP8.5 scenarios.

Both methods bring to use the principle of weighting model based projections based on objective model performance criteria – such as biases (both univariate as well as multivariate) in simulating past climate and measures of simulated variability. The sensitivity to these criteriais tested by varying the metrics and weights assigned to them. Sensitivity of metrics to observational uncertainty is also examined at regional, sub-regional and grid point levels.