H41A-0787:
On the Predictability of SSTA Indices Relevant for Rainfall Prediction from CMIP5 Decadal Experiments

Thursday, 18 December 2014
Dipayan Choudhury1,2, Ashish Sharma2, Alexander Sen Gupta3,4, Rajeshwar Mehrotra5 and Bellie Sivakumar2,6, (1)University of New South Wales, ARC Centre of Excellence for Climate System Science, Sydney, NSW, Australia, (2)University of New South Wales, School of Civil and Environmental Engineering, Sydney, NSW, Australia, (3)University of New South Wales, Climate Change Research Centre, Sydney, Australia, (4)University of New South Wales, ARC Centre of Excellence for Climate System Science, Sydney, Australia, (5)University of New South Wales, School of Civil and Environmental Engineering, Sydney, Australia, (6)University of California Davis, Department of Land, Air and Water Resources, Davis, CA, United States
Abstract:
Improved predictions of rainfall on inter-annual to inter-decadal timescales would better enable water planners to design catchment management policies effectively whilst considering water availability. However at present, rainfall can only be forecast with useful skill up to the lead times of few months. With the recent advancement in observational and methodological capabilities like the phase five of the Coupled Model Intercomparison Project (CMIP5), which provides simulations of near term climate (10 - 30 year predictions/ hindcasts) including observation based initializations, it may be possible to attain improved predictability of rainfall on longer timeframes.
Sea surface temperature anomaly (SSTA) climate indices in the tropical Pacific and the Indian Ocean have been shown to be statistically significant predictors of seasonal rainfall in the Indo-Pacific region. This study evaluates the predictability of seven such SSTA climate indices from the decadal hindcast experiments of four state-of-the-art climate models. A Monte Carlo (MC) scheme is applied to assess the lead time till which these CMIP5 models are able to predict, with some skill, the evolution of these indices. The effect of applying a simple lead time dependent bias correction and the models’ capabilities in simulating two specific El Niño events (1982 and 1997), in terms of related indices, is also analysed.
El Niño related indices like the Niño 3, 3.4 and 4 were found to have improved predictability up to 20 months whereas indices for the Indian Ocean Dipole (DMI) was as low as 4 months. Bias correction clearly improves the predictability for all the indices. Further, the models simulate the observed peaks (1982 and 1997) in the Niño 3 and Niño 3.4 indices, in relation to the El Niños in these years, with limited success. No conclusion could, however, be made regarding which model performs better overall. Investigation of rainfall predictability is in progress and would be presented later.