Benchmarking Data Assimilation Algorithms for 3D Lagrangian Particle Tracking

Abstract

This paper reports a comparison of three state-of-the-art data assimilation algorithms for 3D Lagrangian particle tracking (LPT): Vortex-In-Cell sharp (VIC#), FlowFit3, and neural-implicit particle advection (NIPA). LPT yields position vectors for a large number of tracer particles in a flow volume from a time-resolved sequence of images. Particle position data, a.k.a. “tracks,” are spatially sparse, and a reconstruction algorithm is commonly employed to estimate dense Eulerian fields from the tracks. Data assimilation algorithms for LPT synthesize the tracks with a set of governing equations (or constraints) to enhance the accuracy of the velocity or vorticity field and infer additional quantities like pressure. We compare the performance of VIC#, FlowFit3, and NIPA with respect to their accuracy and computational cost. Accuracy and cost metrics will be presented in terms of the magnitude of tracking errors, mean particle distance, temporal resolution, and total number of frames. Synthetic and experimental data sets will be evaluated; direct numerical simulations will be used for synthetic tests, and realistic tracking errors will be added to the data.