Data guided discovery of dynamic climate dipoles

Pressure dipoles in global climate data capture recurring and persistent, large-scale patterns of pressure and circulation anomalies that span distant geographical areas (teleconnections). In this paper, we present a novel graph based approach called shared reciprocal nearest neighbors that considers only reciprocal positive and negative edges in the shared nearest neighbor graph to find dipoles in pressure data. To show the utility of finding dipoles using our approach, we show that the data driven dynamic climate indices generated from our algorithm always perform better than static indices formed from the fixed locations used by climate scientists in terms of capturing impact on land temperature and precipitation. Another salient point of this approach is that it can generate a single snapshot picture of all the dipole interconnections on the globe in a given dataset making it possible to differentiate between various climate model simulations via data driven dipole analysis. Given the importance of teleconnections in climate and the importance of model simulations in understanding the impact of climate change, this methodology has the potential to provide significant insights.