Targeted Observations for Improving Prediction of the NAO Onset

Based on the viewpoint that the North Atlantic Oscillation (NAO) has an intrinsic timescale of approximate two weeks and can be treated as an initial value problem, targeted observations for improving the prediction of the onset of NAO events are investigated by using the conditional nonlinear optimal perturbation (CNOP) method with a quasi-geostrophic model. The results show that flow-dependent sensitive areas for the prediction of NAO onset are mainly located over North Atlantic and its upstream regions. Targeted observations over the main sensitive areas could improve NAO onset prediction in most cases (approximately 75%) due to reduced errors in anomalous eddy vorticity forcing (EVF) projection in the typical NAO mode. Moreover, a flow-independent sensitive area is determined based on the winter climatological flow, which is located over North America and its adjacent ocean. The NAO onset prediction can also be improved by targeted observations over the flow-independent sensitive area, but the skill improvement is somewhat lower than that derived from observations over the flow-dependent sensitive area. The above results indicate that targeted observations over sensitive areas identified by the CNOP method can help to improve the onset prediction of NAO events.