Improvement in predictive skill across timescales: Justification for a seamless approach

Abstract:

The global coupled atmosphere-ocean-land-cryosphere system exhibits a wide range of physical and dynamical phenomena that collectively result in a continuum of temporal and spatial variability. The traditional boundaries between weather and climate are, therefore, somewhat artificial, and the interplay among the different components of the climate system may add memory and act as sources of predictability. In this study, we investigate whether better seasonal predictions lead to higher multi-year predictive skill. As a first approach, we explore the seasonal prediction skill for sea surface temperature anomalies (SSTA) in the Tropical Pacific. In particular, we investigate whether skill in forecasting Tropical Pacific SSTA beyond the spring predictability barrier leads to improved skill in predicting SSTA variability on multi-year timescales in other regions. We use the Community Earth System Model Decadal Prediction Large Ensemble (CESM-DPLE) to address this topic. The CESM-DPLE contains 40 ensemble members for 62 initialization times, from 1954 to 2015. We analyze randomly-selected combinations of ensemble members and calculate the skill in predicting SSTA variability in the Tropical Pacific for each combination. Composites of the most and least skillful predictions for the first summer are calculated and are then used to estimate the predictive skill of global SSTA on multi-year time scales. Preliminary results show that the combinations of ensemble members that yields the most skillful seasonal predictions of Tropical Pacific SSTA improve the predictions of SSTA in the extratropical Pacific for lead times out to four years, particularly along the North American coast, indicating the importance of timescale interactions in improving multi-year predictability. Our work thus highlights the relevance of the seamless approach to Earth System prediction, with important implications for future forecasting systems and approaches.