I'm trying to reproduce the work in computing turbulent energy spectrum from isotropic turbulence flow field in a box but I have a hard time understanding James' answer about the binning part.

Would you be so kind as to try to explain it to me ? Many thanks !


After a shallow read, I guess that the main point is that you have a limit.

When you expand your velocity ${\bf{u}}(x,y)$ (2D for simplicity) in a Fourier series you will have modes like: $$\exp{(ik_{xn}x)} \qquad\exp{(ik_{ym}y)}$$ Therefore you can write in discrete case with $N+1$ points: $${\bf u}(x,y) = \sum_{nm}{\hat{\bf{u}}_{nm}(k_{nx},k_{my},t)\exp{(ik_{xn}x)} \exp{(ik_{ym}y)}}\qquad n,m \in[-N/2,N/2]$$

Where $\hat{\bf{u}}_{nm}(k_{nx},k_{my},t)$ are the Fourier coefficients that describe $u$ in Fourier space given by wavenumbers $k_{xn}$ and $k_{ym}$ is: $$k_{xn} =\frac{2\pi n}{L}=k_0n \qquad k_{ym} = k_0m$$ The total energy $E_T(t)$ can be computed as: $$E_T(t) = \frac{k_0^2}{2}\sum_{n,m}{\hat{\bf{u}}_{nm}(k_{nx},k_{my},t)\cdot \hat{\bf{u}}^*_{nm}(k_{nx},k_{my},t)}$$

This energy can also be computed with the energy spectrum $E(k,t)$ with $k^2=k_{xn}^2+k_{ym}^2$

$$E_{T}(t) = \sum_{k}E(k,t)\Delta k$$ where $\Delta k$ must be defined with care and $E(k,t)$ is $$E(k,t) = \frac{1}{2}\sum_{k_{xn}^2+k_{ym}^2=k^2}{\hat{\bf{u}}_{nm}(k_{nx},k_{my},t)\cdot \hat{\bf{u}}^*_{nm}(k_{nx},k_{my},t)2\pi k }$$

The last sum is related with the discrete version of a line integral. It can be obtained from what he refers with "binning".

It is clear that values that do not meet with the restriction $k_{xn}^2+k_{ym}^2=k^2$ are discarded ( red zone in James' answer) and therefore you cannot reach values $k_{xN/2}$ or $k_{yN/2}$ and other points that are ourside the circle $k_{xn}^2+k_{ym}^2=k^2$. It can be also shown in the $3D$ case with the sphere: $k_{xn}^2+k_{ym}^2+k_{zl}^2=k^2$.

I am not an expert on this area... I would be grateful if someone sees necessary to correct me in aspects that are not correct.

  • $\begingroup$ Thank you for taking time to answer @HBR but are you sure that the red zone in James' answer matches the values that do not meet the restriction $k^2_{xn}+k^2_{ym}=k^2$ ? $\endgroup$ – PandaMat Jul 31 '17 at 7:09
  • $\begingroup$ Yes. I am sure. $\endgroup$ – HBR Jul 31 '17 at 7:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.