remote: Empirical Orthogonal Teleconnections in R

Abstract

In climate science, teleconnection analysis has a long standing history as a means for describing regions that exhibit above average capability of explaining variance over time within a certain spatial domain (e.g., global). The most prominent example of a global coupled ocean-atmosphere teleconnection is the El Niño Southern Oscillation. There are numerous signal decomposition methods for identifying such regions, the most widely used of which are (rotated) empirical orthogonal functions. First introduced by van den Dool, Saha, and Johansson (2000), empirical orthogonal teleconnections (EOT) denote a regression based approach that allows for straight-forward interpretation of the extracted modes. In this paper we present the R implementation of the original algorithm in the remote package. To highlight its usefulness, we provide three examples of potential usecase scenarios for the method including the replication of one of the original examples from van den Dool et al. (2000). Furthermore, we highlight the algorithm’s use for crosscorrelations between two different geographic fields (identifying sea surface temperature drivers for precipitation), as well as statistical downscaling from coarse to fine grids (using Normalized Difference Vegetation Index fields).