The Internal Atmospheric Noise and Decadal Predictability of Surface Temperature, Precipitation, and Extremes

There is continuously growing demand for the development of decadal climate prediction systems. Making robust and skillful decadal predictions has clear societal benefit in terms of supporting decision-making processes in agriculture, energy and water management. However, forecasting the climate over decades remains a challenge, partly due to the lack of understanding in the sources and mechanisms of decadal predictability. For example, decadal prediction of precipitation is still at a relatively low level of confidence. In this study, we introduce the interactive ensemble (IE) coupling strategy to quantify how internal atmospheric noise impacts decadal predictability of surface temperature, precipitation and extremes. The design of the IE is to couple multiple realizations of the atmospheric component model to a single realization of the ocean model; only the ensemble mean fluxes of heat, momentum and moisture are coupled to the ocean, allowing an assessment of how reduced noise influences decadal predictability. A suit of coupled model experiments including the control and the IE experiments, are conducted using the Community Climate System Model version 4. The results suggest that the internal atmospheric noise can significantly affect decadal predictability of surface temperature, precipitation and extremes. We argue that the internal atmospheric noise impacts decadal predictability from two main perspectives, namely, local ocean-atmosphere interactions and teleconnection patterns.