Vector Field Decompositions Using Multiscale Poisson Kernel.

Area of Science:

• Computational fluid dynamics
• Multiscale analysis
• Vector field analysis

Background:

• Scale-space methods are fundamental for scalar field analysis but less common for vector fields.
• Turbulent flows exhibit multiscale vortex structures that are challenging to analyze with existing techniques.
• Traditional scale-space approaches struggle with overlapping features due to iterative smoothing.

Purpose of the Study:

• To develop a new approach for the multiscale analysis of vector fields.
• To overcome limitations of existing methods in extracting overlapping features.
• To enable the separation of local flow behavior from large-scale effects.

Main Methods:

• Utilized spatial regions of influence as the definition of scale for vector fields.
• Employed the Helmholtz-Hodge decomposition to split vector fields into scale components.
• Applied progressively larger neighborhoods to analyze features at different scales without smoothing.

Main Results:

• Successfully separated small-scale features (e.g., vortices) from large-scale effects (e.g., background flow).
• Demonstrated the technique's effectiveness on large-scale turbulent flow data.
• Extracted multiscale features not identifiable with current state-of-the-art methods.

Conclusions:

• The proposed method offers a robust approach for multiscale vector field analysis.
• Spatial regions of influence provide a natural scale definition for vector fields.
• This technique enhances the understanding of complex flows by resolving multiscale phenomena.