Objectives: To assess the seasonal predictability for monsoon climates using ensembles of model integrations, and with the application of advanced statistical techniques. The origins of internal and/or externally forced predictability will be sought.

Work description:

UREADMY: The reproducibility/predictability of monsoon systems will be estimated using the ensemble of AMIP II simulations at different horizontal and vertical resolutions, prepared by UREADMY and The Met. Office. The DEMETER seasonal prediction ensembles of The Met. Office will be used to assess the skill of seasonal forecasts for the Asian and African monsoon systems, including the incidence of extreme events. The interaction between intraseasonal and interannual variability will be investigated in terms of the potential for seasonal predictability. The real skill and value of monsoon forecasts will be evaluated using concepts from decision theory.

DMI: The assessment of the predictability of the Asian/Indian summer monsoon activity will be done by means of singular value decomposition (SVD) or canonical correlation analysis (CCA) applied to the DEMETER seasonal prediction ensembles of ECMWF. The All India Rainfall or a different monsoon-index such as the vertical wind-shear will be used as predictands and various parameters, which may influence the monsoon activity, such as the SSTs in the western tropical Pacific or the Eurasian snow-cover will be used as predictors. A model-dependent bias correction (developed in the DEMETER project) will be applied. The origins of predictability will be investigated using hindcasts of the monsoon activity based on a variety of statistical forecast models with different combinations of predictors

ECMWF: The predictive skill of the DEMETER coupled integrations will be diagnosed with particular attention to the simulation of hydrological variables in the tropical regions of Africa. The usefulness of the predictions as input to crop yield models will be evaluated.

Deliverables:

D1101: Assessment of the seasonal predictability of monsoon systems using ensemble techniques.

D1102: Development of advanced statistical techniques for evaluating the skill and worth of seasonal forecasts.

D1103: Assessment of the origins of monsoon predictability. Milestones:

Year 1: Assessment of the reproducibility/predictability of monsoon climates in AMIP II simulations.

Year 3: Assessment of the current skill in seasonal prediction ensembles for monsoon climates, including the application of advanced statistical techniques.