Global dynamical forecasting system conditioned to robust initial and boundary forcings: seasonal context

We propose how seasonal climate prediction with the use of an atmospheric general circulation model (AGCM) can be optimized. The AGCM predictive skill is extensively examined under various forecast strategies that mimic truly operational prediction. It is shown that the AGCM predictive skill is found to produce superior results given a suitable sea-surface temperatures (SSTs) as forcing and is subject to an initialization strategy that uses realistic atmosphere and soil moisture states. Evaluation of hindcasts performed with the model further revealed that the AGCM is able to forecast anomalous upper air atmospheric dynamics (circulation) over the tropics up to several months ahead. The AGCM probabilistic forecasts for rainfall and surface air temperatures during the austral summer season are also found to be informative and useful. The contribution of the predicted SST, which is based on a multi-model approach, is shown to be of significant importance for best AGCM results. The AGCM may also benefit from the initial condition interface in the AGCM's configuration which is implicitly considered in the analysis. Notwithstanding, the AGCM's predictive skill does not vary much whether the AGCM is initialized with realistic or climatological soil moisture which is presumably suggestive of the AGCM's internal weakness.