Conventionally, intra-seasonal to interannual climate prediction tends to emphasize forecasting time mean (e.g., weekly, monthly, or seasonal) anomalies. The wisdom behind this strategy is that the inherent predictability time scale of weather as an initial value problem is about 1-2 weeks (the average limit of predictability). Beyond a week, the forecast skills of mean anomalies hinge heavily upon the presence of large-amplitude anomalous “external” forcings (such as SST anomalies associated with ENSO events). The challenge of subseasonal to interannual climate predictions is that atmospheric internal variability in the extratropics often overwhelms the externally forced (teleconnection) variability, particularly over the regions where prominent atmospheric internal modes, such as the NAO, are present at all time scale. As a result, the signal coming from the anomalous external forcing is diluted significantly, leading to indecisive climate forecasts.

In this study, we will explore the possibility of predicting the intra-seasonal to interannual climate probability distribution of “extreme” weather events (or large amplitude synoptical scale anomalies) using a combination of dynamics based numerical model predictions and statistical relations between surface weather events and upper level circulation anomalies. The rationale for this strategy is that although there is little “memory” to speak of for individual weather events beyond the limit of predictability, the statistical behavior of all weather events as a whole in a spatial domain at any given time may still well be dictated by a “lingering memory” stored in the “state of the atmosphere”, which can be measured by an integral of large-amplitude synoptical scale anomalies over a given spatial domain. We will evaluate if the lingering memory in the system can be predicted by the state of art climate general circulation models beyond the 1-2 week predictability limit. Next we will utilize the information of the lingering memory predicted by a dynamical based numerical model to make subseasonal and seasonal climate predictions about the general state of the future climate over a continental-scale domain, such as a mild winter with few storms versus a cold and stormy winter or a hot and dry summer versus a wet and cool summer.