But do you know how to do that first from a theoretical point of view?
This is quite a classical topic in the statistical analysis of turbulence. For homogeneous isotropic turbulence, the integral scale can be associated to the Taylor microscale lambda. That can be computed from the autocorrelation, as explained in many textbooks (e.g., Pope).
You can find a lot of link searching in the web, for example