Earth System Predictability Section

Who We Are

Earth System Predictability Section scientists use Earth system models such as NCAR’s CESM and DOE’s E3SM, as well as CMIP multi-model data sets, a hierarchy of models, observations, and machine learning (ML), to gain a predictive and process-level understanding of modes of variability in the Earth system. 

Our Mission

The mission of the Earth System Predictability Section is use Earth system models and machine learning techniques to better understand initialized Earth system predictability of modes of variability involving interactions across spatiotemporal scales (subseasonal to decadal, regional to global) that connect modes of variability to high-impact events and weather regimes.

Research Objectives (ROs): 

Research Objective 1 (RO1) proposes to quantify Earth system predictability occurring on subseasonal to seasonal (S2S) timescales related to modes of variability involving the Madden-Julian Oscillation (MJO) and connections to tropical cyclones and Antarctic atmospheric rivers that could provide predictive skill for seasonal to decadal (S2S) time scale phenomena.

Research Objective 2 (RO2) will evaluate Earth system predictability of modes of variability on seasonal to decadal (S2D) timescales with connections to Pacific Decadal Variability (PDV), Atlantic Multidecadal Variability (AMV), the Atlantic Meridional Overturning Circulation (AMOC), and global monsoon systems.

Research Objective 3 (RO3) will provide improved understanding of Earth system model representation of modes of variability across all timescales by examining processes and mechanisms that contribute to model predictive skill across time and space scales. This work will be informed by model- and observation-based analyses of the influences of mean state biases, and individual external forcing agents such as biomass burning aerosols.

Research Objective 4 (RO4) will develop and apply machine learning (ML) methodologies to provide insights into improving the predictability of modes of variability on all timescales and to complement Earth system model initialized hindcasts for analyses of sources of skill that connect fundamental processes and mechanisms of modes of variability on different timescales.