Spectral model turbulence analysis technique is widely used to derive kinetic energy dissipation rates of turbulent structures (ε) from different in situ measurements in the Earth’s atmosphere. Essence of this method is to fit a model spectrum to measured spectra of velocity or scalar quantity fluctuations and thereby to derive ε only from wavenumber dependence of turbulence spectra. Owing to simplicity of spectral model of Heisenberg (1948) its application dominates in the literature.
Making use of direct numerical simulations (DNS) which are able to resolve turbulence spectra down to smallest scales in dissipation range, we advance the spectral model technique by quantifying uncertainties for two spectral models, the Heisenberg (1948) and the Tatarskii (1971) model, depending on 1) resolution of measurements, 2) stage of turbulence evolution, 3) model used.
We show that model of Tatarskii 1971 can yield more accurate results and reveals higher sensitivity to lowest ε-values.
This study shows that the spectral model technique can reliably derive ε if measured spectra only resolve half decade of power change within viscous (viscous-convective) subrange. In summary we give some practical recommendations how to derive most precise and detailed turbulence dissipation field from in situ measurements depending on their quality.
We also supply program code of the spectral models used in this study in Python, IDL, and Matlab.