QUANTIFYING PREDICTABILITY IN A SIMPLE MODEL WITH COMPLEX FEATURES

Publisher Website
Google Scholar
Abstract
Here, Kaplan–Yorke type maps are utilized as simplified models to assess new strategies for quantifying predictability through information theory. These models give rise to a wide variety of "climate" distributions from nearly Gaussian to highly non-Gaussian. For complex models, it is almost impossible to compute proposed theoretical measures of predictability directly and alternative methods of estimation must be utilized. Due to the simplicity of the proposed model, accurate approximations of predictability can be computed and compared to various estimation techniques. A recently proposed method for finding a lower bound estimate of the predictability is outlined in the context of the model. Estimates of this type are computed and evaluated for a long-term climate prediction scenario. The factors that control the predictability for this scenario are determined using an ensemble of ensembles approach. In addition, the lower bound estimates are used as a means of assessing the utility of a Gaussian approximation strategy.

This publication has 11 references indexed in Scilit: