IMAGe Brown Bag Seminar - A Flexible Decomposition of Climate Variability into Regional Components and Teleconnections: Insights from a Large Initial Condition Ensemble

05/25/2017 - 12:00pm to 1:00pm
ML - Damon Room

A Flexible Decomposition of Climate Variability into Regional Components and Teleconnections: Insights from a Large Initial Condition Ensemble

Vineel Yetella

University of Colorado, Boulder

We propose a framework that facilitates the decomposition of the predictability and variability of a global climate variable into components arising from regions and their teleconnections. The framework is applicable to a broad range of variability studies since the choice of the subsystems is arbitrary. For example, the framework can be used to decompose the predictability and variability of a global variable into components arising from land and ocean, ENSO and non-ENSO regions, polar and non-polar regions etc., and their teleconnections. As such, the framework enables comparing and contrasting the regional contributions to global predictability and variability, both within a climate model and across climate models. To illustrate the measure, we present results from its application to the decomposition of global surface temperature predictability into components arising from land, ocean and their teleconnection in a large initial condition ensemble (Community Earth System Model 1 - Large Ensemble). The ensemble is initialized at 1920 and run under historical forcing from 1920 - 2005 and RCP 8.5 forcing from 2006 - 2100. The land component of initial-value predictability decays to zero in approximately one month, considerably faster than the ocean component which persists for almost two years. The component arising from the land-ocean teleconnection has a non-negligible contribution to the global initial-value predictability for almost two months. On longer time scales, all components of global temperature variability increase relative to their 1850 climatology in response to anthropogenic forcing, suggesting stronger teleconnections and a greater probability of extremes in the model projection.

(Bring your lunch)