The climate system involves a complex interplay between the ocean and atmosphere. Studies of this interplay typically rely on model simulations in comparison with time series data for some feature of oceanic and/or atmospheric circulation on a regional or larger scale. Such comparisons help to probe the realism of the climate models, and yet, time series data typically involve some kind of time averaging. Little is known about how the averaging procedure may introduce inherent differences between data and models.
In a new paper, LML Fellow Davide Faranda and colleagues address this matter by exploring the dynamical properties of a simplified ocean-atmosphere model. They use dynamical systems theory to measure the dimensionality of the system dynamics, and compare the model results with both high temporal resolution data, and a degraded dataset derived from long-term averages. They show that the general impact of averaging is reflected in changes in the attractor of the system. Not surprisingly, averaging over longer times generally tends to smooth the dynamics, especially beyond an interval of 1 year. However, the decrease in dimension with timescale does not follow a simple monotonic pattern. Time-averaging sometimes has counter-intuitive consequences in increasing the dimension of the dynamics. Overall, they conclude that the inferences drawn from the conceptual model, while presenting differences from the analysis of time-averaged climate indices, can nonetheless provide some valuable insights into the behaviour of real-world climate.
The paper is available at https://www.tandfonline.com/doi/full/10.1080/16000870.2018.1554413