1. Scientific Objectives

Addressing issues related to ensemble strategies for the sub-seasonal predictions. Relevant scientific questions include:

• Relative merits of ensemble generation strategies for the sub-seasonal timescale (ens. size, burst/LAF, etc.);

• Influence of ocean and coupled initial perturbations for potential improvements in skill and reliability in the prediction of certain regimes (e.g. MJO, tropical cyclone);

• Over-confident predictions due to the discrepancy between the observed and forecast spread resulting from both random and systematic errors (bias errors);

• Understanding and representing model uncertainty (e.g., stochastic physics) for the sub-seasonal timescale.

2. Proposed Activities

(1) Studying influences of forecast configuration strategies

This sub-project conducts comparative studies for providing a guidance of ensemble strategies (LAF and burst approaches).

(2) Benchmarking a spread-error relationship

Sub-seasonal predictions often suffer from over-confident predictions. Reasons for the over-confident predictions may be result from both random and systematic errors.  Uncertainty measures are not settled for sub-seasonal predictions, while a forecast spread is widely used for medium-range predictions. This sub-project aims to propose quantify uncertainty measures and evaluate uncertainty representations in the current generation of S2S prediction systems.

(3) Investigating impacts of stochastic parameterizations

Stochastic physics schemes have the potential to help to represent time-evolving model uncertainty. This sub-project, jointly with PDEF/WGNE, will propose coordinated experiments for quantifying impacts of Stochastically Perturbed Tendency (SPPT) scheme.

(4) Exploring the impacts of coupled initial and model perturbations

Ensemble techniques for coupled models also need to be explored. There may be also impacts of increasing model complexity such as ocean coupling. 

3. Ongoing Community Research on Ensemble Techniques

Please let us know your research, papers related to the ensemble sub-project!

http://enterprise.xwiki.org/xwiki/bin/view/UserGuide/