Fig. 6: Statistical evaluation of the different models.
From: Online calibration of deep learning sub-models for hybrid numerical modeling systems
a Probability density function of the vorticity field (b) Time evolution of the kinetic energy \({E}_{t}=\frac{1}{2} < {\overline{\psi }}_{t}{\overline{\omega }}_{t} > \) and enstrophy \({Z}_{t}=\frac{1}{2} < {\overline{\omega }}_{t}^{2} > \) normalized by the energy and enstrophy of the initial condition (of the filtered DNS) E0 and Z0 respectively. c Time averaged kinetic energy spectra Eν, enstrophy spectra Zν and power spectrum of the SGS term ∣Πν∣. The PDF is computed using a kernel density estimator. The result highlighted in this figure correspond to colored boxes in Fig. 5.
