Effect of parameter change upon the extra-tropical atmospheric variability

Abstract

Global climate models contain numerous parameters with uncertain values. In the context of climate simulation and prediction, it is relevant to obtain an estimate of the range of climate outcomes given the parameter uncertainty. Instead of randomly perturbing parameters, we determine parameter perturbations from short-term integrations that potentially have a high impact on the climate of the model. For this purpose we consider a dry, spectral quasi-geostrophic, three-level model on the sphere and its tangent linear and adjoint equations. With an empirical forcing, the model produces a fairly realistic simulation of the extra-tropical winter circulation. We allowed perturbations in a 1,449 dimensional parameter space. As a measure of impact on the climate we compute the change in the probability density function of the dominant patterns of variability. We find that the largest climate response in a set of 1,000 simulations with potentially high impact perturbations is much larger than the largest response in a similar set of simulations with randomly picked perturbations. We conclude that parameter sensitivity calculations based on short term integrations contain valuable information about the sensitivity of the model climate to parameter perturbations. The approach is feasible for stateof-the-art climate models provided that the tangent linear and adjoint equations are implemented.

An Optimization Strategy for Identifying Parameter Sensitivity in Atmospheric and Oceanic Models

“…Generally, there are many uncertain parameters used to parameterize unresolved physical processes in atmospheric and oceanic models. The uncertainties of these parameters can add significant uncertainties in the simulation of atmospheric and oceanic models (e.g., Murphy et al 2004;Edwards and Marsh 2005;Brierley et al 2010;Levine-Moolenaar et al 2012). Furthermore, the effects of uncertainties in different parameters on the model simulation and prediction are very diverse (Chu 1999;Orrell 2003;Yin et al 2014).…”