Turbulence and attenuation are signal degrading factors that can severely hinder free-space and underwater OAM optical pattern demultiplexing. A variety of state-of-the-art convolutional neural network architectures are explored to identify which, if any, provide optimal performance under these non-ideal environmental conditions. Hyperparameter searches are performed on the architectures to ensure that near-ideal settings are used for training. Architectures are compared in various scenarios and the best performing, with their settings, are provided. We show that from the current state-of-the-art architectures, DenseNet outperforms all others when memory is not a constraint. When memory footprint is a factor, ShuffleNet is shown to performed the best.

This paper is a continuation of previous work that analytically examined the strength of radiation leaving an air-water interface. The approach here is to numerically integrate the radiation transport equation in order to capture the non-linear features of the problem, and also to include more realistic models for the thermal boundary layer. The radiation intensity of the photons emitted from the interface, relative to that of thermal radiation at the temperature of the interface, is defined here as the signal. This signal was computed for constant surface temperature and constant heat flux boundary conditions. As expected, the numerical computations show that the signal increased as the air-water temperature difference increased. The results are shown to form a hierarchy of signal strengths based on the chosen thermal stratification model. However, for both boundary conditions, the numerical results for a linear temperature profile compared very favorably with the simplified analytical linearized model over the thermal wavebands of 3–5 and 8–14 microns. In addition, the linearized model compared favorably with the most realistic models of thermal stratification.

We present a convolutional neural network (CNN) framework for classifying underwater optical turbulence strength from multimodal measurements of laser beams carrying orbital angular momentum propagated through a Rayleigh–Bénard convection tank. Three modalities were evaluated for four beam types ( ℓ ∈{−2,0,2,4}): pupil-plane intensity achieved the highest accuracy (96.1%–98.1%), followed by focal-plane (85.0%–90.0%), mode sorter (78.5%–79.3%), and Gaussian beam phase (79.3%). Classification accuracy was comparable across beam topological charges, consistent with findings from our companion paper which analyzed traditional optical turbulence quantification statistics such as scintillation, beam wander, and mode sorter crosstalk. Dual-input fusion architectures that paired the pupil-plane with a second modality did not improve over pupil-only performance. Unlike classical metrics that require multiple frames, the CNN classifies turbulence from individual frames, offering a path toward real-time sensing.

We present a comprehensive experimental study investigating orbital angular momentum (OAM) beams for sensing underwater optical turbulence strength using multimodal optical measurements. Synchronized measurements of pupil plane intensity, focal plane intensity, and mode-sorter spectra were collected in a Rayleigh–Bénard convection tank across three turbulence strengths, ranging from C _n ² =1.96×10 ⁻¹⁰ to 2.06×10 ⁻⁹ m ^−2/3 . The pupil plane scintillation index provided the best discrimination between turbulence conditions, with comparable performance across OAM modes. Focal plane beam wander showed a monotonic increase with turbulence strength across all OAM modes. The mode-sorter results varied with the OAM mode, with modes ℓ =2 and ℓ =4 yielding greater separation in the mode spreading due to the turbulence. Our results demonstrate that the pupil plane scintillation across all OAM modes tested captures the most relevant information for discriminating between optical turbulence strengths, validating structured light as a versatile probe for environmental sensing.

Editorial: Applications of Infrared Methods in Fluid Mechanics