Model selection: Using information measures from ordinal symbolic analysis to select model sub-grid scale parameterizations

PULIDO M

Valdivia

Conferencia; XIX Conference on Nonequilibrium statistical mechanics and nonlinear physics (Medyfinol 2016); 2016

Medyfinol

The use of information measures as a means of model selection is explored to diagnose model parameterizations and to develop them.  Although the resolved dynamical equations of atmospheric or oceanic global numerical models are well established,  the development and evaluation of parameterizations that represent subgrid-scale effects  pose a big challenge. For climate studies, the parameters or parameterizations are usually selected according to a root mean square error criterion, that measures the differences between the model state evolution and observations along the trajectory. However, systematic model errors pervade the root mean square error measures particularly for parameterization evaluation where we expect different responses for different points in the state space. To overcome these difficulties, here we evaluate ordinal information theory quantifiers (Shannon entropy, statistical complexity) as measures of the model dynamics. This ordinal analysis is conducted using the Bandt-Pompe symbolic data reduction in the signals. It distinguishes different dynamical behaviors and discriminate clearly chaotic from stochastic signals. We examine the proposed ordinal information measures in the two-scale Lorenz'96 system. By comparing the two-scale Lorenz'96 system signals with a one-scale Lorenz'96 system with deterministic and stochastic parameterizations, we show that information measures are able to select the correct model and to distinguish the parameterizations including the degree of stochasticity that result in the closest model dynamics to the two-scale Lorenz'96 system.