Fluid Dynamics - Extracting Information from Instantaneous Snapshot

[sgg303]

I have questions about how fluid dynamics may be used in general, particularly flow visualisation and extraction of flow parameters.

I currently use a DNS code to propagate a wake near the fluid surface (fixed), from which I extract snapshots of the surface velocity field at particular integrals which then is used as input to some remote-sensing models to simulate a radar image of the surface. I'm doing this for a simulated ship wake, and so what you see is some of the larger flow structure which is present at the surface. So, taking the instantaneous surface velocity profile, I get a snapshot of that flow/profile, analogous to a photograph.

What I'm interested in is whether that snapshot of the flow/structure can be used to extract (qualitative or quantitative) information about the flow. I have to assume that no time series is available, and so whether/what information about the flow is extractable from that single snapshot image. I also have to assume that there is no other information available, apart from the length scale of the image.

Any ideas?


I’m sorry if this is unclear, but I’m not a fluid dynamicist and so I’m not familiar with the terminology or practices involved with CFD or flow visualisation. If you’ve any questions, please ask.

 

[Valkarie]

The most straightforward approach to extracting information from a single snapshot image is to apply image processing techniques. These techniques can be used to identify features such as eddies, vortices, and other flow structures which may be indicative of the flow dynamics. Additionally, it may be possible to calculate some quantitative metrics such as Reynolds numbers and turbulent intensities based on the size and magnitude of the features in the image. You could also use statistical methods to analyse the velocity field, such as power spectra or correlation functions. These methods allow you to characterise the spatial and temporal properties of the flow over a wide range of scales, and can provide insight into the nature of the flow.Finally, another option is to employ machine learning methods, such as convolutional neural networks. These can be used to detect patterns in the image which are indicative of particular flow parameters, allowing for the extraction of useful flow information from a single snapshot.