Subject(s): CX.07

Category: Brief Article

Abstract:

The significance of a modern complexity measure and modeling framework is revealed for climate data and climate models respectively. For climate data, we computed the mutual information between vertical sequences of geopotential heights centered at 500 hPa in the Northern Hemisphere, re-analysis data set for extended winters in the 1948-1998 period. The mutual information filter reveals El Nino induced climate change occurring primarily in the northern Pacific and arctic regions, and climate change in the last 25 years compared to the previous twenty five years occurring primarily in the Arctic. We apply the filter to the Lorenz system to help interpret the findings, and we discuss how this new perspective on climate change can best be utilized for improved climate prediction. For climate models, we replaced a statistical parameterization for momentum transport in the sub-grid portion of a single column climate model with a rigorous statistical inference model known as an epsilon machine. In preliminary, ensemble runs, this replacement caused the model to perform equally or better than with the use of the standard eddy-diffusivity parameterization for momentum transport.

Retrieve Manuscript

Retrieve Previous Revision's Abstract

Submit referee report/comment