Title: Detecting climate teleconnections with Granger causality

Abstract: Climate system teleconnections, which are far-away climate responses to perturbations or oscillations, are difficult to quantify, yet understanding them is crucial for improving climate predictability. Here we leverage Granger causality in a novel method of identifying teleconnections. Because Granger causality is explicitly defined as a statistical test between two time series, our method allows for immediate interpretation of causal relationships between any two fields and provides an estimate of the timescale of the teleconnection response. We demonstrate the power of this new method by recovering known seasonal precipitation responses to the sea surface temperature pattern associated with the El Ni\~{n}o Southern Oscillation, with accuracy comparable to previously used correlation-based methods. By adjusting the maximum lag window, Granger causality can evaluate the strength of the teleconnection (the seasonal precipitation response) on different timescales; the lagged correlation method does not show ability to differentiate signals at different lags. We also identify candidates for previously unexplored teleconnection responses, highlighting the improved sensitivity of this method over previously used ones.