Boiling Flow

Boiling Flow is a data-driven algorithm that generates synthetic time-series of images (of arbitrary duration) by estimating statistical parameters from an input time series of images.

For more information about the algorithm, see [3, 4].

See also the AOModel package on GitHub: aomodel. This package implements a data-driven algorithm using Principal Component Analysis and autoregressive modeling to generate synthetic time series data that matches the spatial and temporal statistics of measured time series data.

The benefits of AOModel compared to Boiling Flow are:

• Synthetic data generated by AOModel has more accurate statistics.

• The model is highly generalizable to different measured data sets.

The drawbacks of AOModel compared to Boiling Flow are:

• The size of synthetic data images generated by AOModel is restricted to the size of the measured data images.

• The parameters of AOModel are much less physically relevant.

Use Boiling Flow if your measured data falls into the boiling flow regime (i.e., high convective data with spatially stationary statistics) and you need to extend the phase screens beyond the size of the measured data images. Alternatively, use Boiling Flow to estimate turbulence parameters from measured phase screen data. Use AOModel if you need to generate synthetic data that closely matches the spatial and temporal statistics of the measured data, or if the measured data is not accurately modeled by boiling flow.

---

Disclaimer: Approved for public release; distribution is unlimited. Public Affairs release approval # AFRL-2025-5580.