On data-driven induction of the low-frequency variability in a coarse-resolution ocean model

Abstract

This study makes progress towards a data-driven parameterization for mesoscale oceanic eddies. To demonstrate the concept and reveal accompanying caveats, we aimed at replacing a computationally expensive, standard high-resolution ocean model with its inexpensive low-resolution analogue augmented by the parameterization. We considered eddy-resolving and non-eddy-resolving double-gyre ocean circulation models characterized by drastically different solutions due to the nonlinear mesoscale eddy effects. The key step of the proposed approach is to extract from the high-resolution reference solution its eddy field varying in space and time, and then to use this information to improve the low-resolution analogue model.

On dynamically unresolved oceanic mesoscale motions

“…Finally, the proposed framework can be used for data-driven parameterizations (similar to the ideas put forward in Ryzhov et al. 2020), in which the field errors (i.e. eddies) are emulated statistically and, then, via the dynamical model, are used to produce the ECF for augmentation of the employed coarse-grid model.…”

“…Ryzhov et al. 2020) to predict the field error (i.e. eddies) for follow-up parameterization purposes is not new.…”

“…An alternative approach (Ryzhov et al. 2019, 2020) is to emulate eddies statistically and input them into non-eddy-resolving models. Note, that either approach needs information about the unresolved eddy field, hence, defining the latter becomes important and may strongly influence the resulting eddy parameterizations.…”

Data-adaptive harmonic analysis of oceanic waves and turbulent flows

“…Still, even after PCA compression, matrix C can be very large and its direct eigendecomposition is computationally prohibitive. For such cases, Ryzhov et al (2019Ryzhov et al ( , 2020 have utilized frequency-domain DAHD formulation, described next.…”

“…The dataset is obtained by the eddy-resolving oceanic quasigeostrophic (QG) model (Shevchenko and Berloff, 2015;Kondrashov and Berloff, 2015;Ryzhov et al, 2019Ryzhov et al, , 2020 (Fig. 9); blue -modes with the largest λ at each frequency, green -power-law fit ≈ f −5/3 ; magenta -exponential law fit e −2tsf with characteristic time scale τ s ≈ 25 days, black dots -DAHD modes shown in Fig.…”