Reconstruction of three-dimensional turbulent flow structures using surface measurements for free-surface flows based on a convolutional neural network

Abstract

A model based on a convolutional neural network (CNN) is designed to reconstruct the three-dimensional turbulent flows beneath a free surface using surface measurements, including the surface elevation and surface velocity. Trained on datasets obtained from the direct numerical simulation of turbulent open-channel flows with a deformable free surface, the proposed model can accurately reconstruct the near-surface flow field and capture the characteristic large-scale flow structures away from the surface. The reconstruction performance of the model, measured by metrics such as the normalised mean squared reconstruction errors and scale-specific errors, is considerably better than that of the traditional linear stochastic estimation (LSE) method. We further analyse the saliency maps of the CNN model and the kernels of the LSE model and obtain insights into how the two models utilise surface features to reconstruct subsurface flows. The importance of different surface variables is analysed based on the saliency map of the CNN, which reveals knowledge about the surface–subsurface relations. The CNN is also shown to have a good generalisation capability with respect to the Froude number if a model trained for a flow with a high Froude number is applied to predict flows with lower Froude numbers. The results presented in this work indicate that the CNN is effective regarding the detection of subsurface flow structures and by interpreting the surface–subsurface relations underlying the reconstruction model, the CNN can be a promising tool for assisting with the physical understanding of free-surface turbulence.

Vortex imprints on a free surface as proxy for surface divergence

“…A few studies used wavelets for surface wave analysis (Dolcetti & García Nava 2019), and very recently a wavelet-based surface feature tracking technique was demonstrated in the laboratory by Gakhar, Koseff & Ouellette (2022). Another approach to inferring bulk flow from surface information is machine learning, recently demonstrated by Xuan & Shen (2023); machine learning is a powerful predictive technique whereas the technique presented here is conceptually simple and physically transparent at every step, so the two approaches can be considered complementary.…”

“…There are several straightforward ways by which the method can be further improved to remove typical types of false positives and negatives (some observations are mentioned below). A fully optimised detection algorithm is not our goal here, however; a machine learning approach such as Xuan & Shen (2023) could be a better approach if so. We wish instead to highlight the transparent physics and emphasise how a simple selection based only on two physical properties, lifetime and eccentricity, can be highly effective.…”