How does one practically compute the kinetic energy spectra from a timeseries of u- and v- Cartesian components of ocean current velocity ($m s^{-1}$) obtained from an Acoustic Doppler Current Profiler (ADCP) moored on the seabed?

Do you:

1) compute the spectra of each u- and v- components and then calculate the resulting kinetic energy from the spectra in the frequency domain? i.e. $KE(f) = 0.5\times(u_{(f)}^2 + v_{(f)}^2)$; where $KE_{(f)}$ is the kinetic energy spectrum per unit frequency and similarly $u_{(f)}$ and $v_{(f)}$ are the u- and v- power spectra?


2) compute the kinetic energy $KE = 0.5\times(u_{(t)}^2 + v_{(t)}^2)$, where $u_{(t)}$ & $v_{(t)}$ are the current velocity timeseries and then calculate the power spectral density (i.e. using Matlab's pwelch) of the resulting KE?

Any guidance with the theory or practical calculation would be very much appreciated.

Thank you.


1 Answer 1


Do you get much different results if you do both ways?

I think if you want the kinetic energy spectrum, use your second method. Some separately compute velocity and KE spectra, e.g., figure 2 of http://journals.ametsoc.org/doi/pdf/10.1175/2009JPO4330.1 (DP Wang et al, JPO, Apr 2010). Comparing velocity and KE spectra, the KE spectra looks similar to the more powerful velocity spectrum, especially at low frequencies.

Others use your first method (e.g., http://journals.ametsoc.org/doi/pdf/10.1175/JPO-D-15-0087.1, Rocha et al, JPO, Feb 2016)

  • $\begingroup$ Thank you for your response John. The results are different by an order of magnitude (method 2 larger), although spectral peaks are at the right frequencies (tidal periodicities). As my data is tidally dominant (M2 & S2, in particular), the other main difference is that with my first method, the strongest peak in energy is at M2, whereas with method 2, the strongest peak is at M4 (double M2). I propose this is because K.E. is no longer a sinusoidal signal but rather a scalar quantity. $\endgroup$
    – Mark
    Commented Jul 7, 2017 at 11:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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