Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

Climate Variability & Predictability (CVP) logo

Examining the Predictability of the Tropical Atlantic Variability using Coupled Prediction Models

Understanding the predictability of the tropical Atlantic variability (TAV) is crucial for short-term climate prediction in the Atlantic sector. Two major TAV mechanisms are regional air-sea interaction and remote El Niño/Southern Oscillation (ENSO) influence, both potentially predictable on the seasonal-to-interannual time scales. Mid-latitude seasonal atmospheric anomalies over the North and South Atlantic may also be useful precursors for the tropical anomalies in subsequent seasons. In a dynamical forecast model, these mechanisms and their interactions should be represented realistically and initialized accurately. This study examines the TAV predictability using a coupled ocean-atmosphere general circulation model (CGCM) with realistic ocean/atmosphere initial states. In particular, we would like to understand what kind of initial surface and subsurface anomalies within the Atlantic Ocean can damp or amplify the remote ENSO influences, and vice versa. We will also examine under what conditions the midlatitude anomalies can stimulate major tropical air-sea feedback on seasonal time scales. 

Our previous studies have shown that this CGCM can simulate the major TAV features realistically. For this study, we propose further improvements in its simulation and initialization. We will conduct an empirical CGCM error correction that prescribes observational climatological monthly mean low-cloud amount in the model while allowing the anomalous model sea surface temperature (SST)-low cloud feedback. The mean cloud correction targets specifically the inadequate simulation of low cloud fraction over the southeastern tropical Atlantic, which is a major component of the current CGCM systematic bias and unrealistic annual cycle in this region. The preserved anomalous SST-low cloud feedback is important for the interannual variability. To minimize the CGCM initial shock, a new model initialization technique will be tested. We will conduct a series of coupled “initialization runs”, in which the coupled system is nudged continuously toward the observed climate anomalies derived from the oceanatmosphere analyses. Initial conditions from these runs should be more in balance between the ocean and atmosphere and likely represent the low-frequency signals better than the instantaneous initial ocean-atmosphere states separately generated by uncoupled atmospheric and oceanic data assimilation systems. To take into account the potentially significant uncertainty of current oceanic analysis in the tropical Atlantic Ocean, we will use several different ocean analysis products to generate an ensemble of oceanic states with sufficient spread. Using the improved CGCM, we will conduct a set of hindcast experiments for 1981-2005 to establishing its predictive skill in the tropical Atlantic. Further sensitivity case studies will be conducted using our regional coupling strategy with the observed or climatological SST anomalies prescribed in the tropical Pacific to study the relative roles of the ENSO forcing and regional air-sea interaction. To link this study more directly to operational prediction, we plan to conduct some experiments with the NCEP Climate Forecast System (CFS), which has been successfully installed at COLA.

Scroll to Top