Ocean Photonics.

Plankton is arguably the most important class of organisms inhabiting the pelagic realm, playing a central role in the biological carbon pump and being essential to maintaining healthy aquatic ecosystems. These microorganisms are sensitive to abiotic environmental conditions and are hence threatened by climate change. As a consequence, monitoring the spatiotemporal variations in their distributions, abundance, and biodiversity is a task of ever-growing priority to assess the health status of our oceans. In situ optical tools unlock high-resolution and non-intrusively surveying of plankton with high throughput. Nonetheless, underwater imaging of sub-millimeter organisms is a technically complex task, and current instrumentation is restricted to fixed and small closerange volumes, requiring the instruments to be vertically dived to produce distributions. Here, I report the development of a novel confocal light detection and ranging (LiDAR) system for underwater volumetric sensing of laser-induced fluorescence. This technology combines a 445 nm focused Gaussian beam with a confocal pinhole aperture to spatially suppress out-of-focus contributions from background light and enable high-resolution imaging through scattering media. The integration of bi-directional scanning with remote focusing allows us to survey the water column remotely and map distributions. In the first stage, I implement a single-channel system for spatially-resolved detection of autofluorescence from sub-millimeter copepods. Furthermore, we investigate the impact of water turbidity on the system’s performance, using signal-to-noise ratio and spatial resolution as metrics. Subsequently, the detection is expanded to 15 wavelength channels with a fast and sensitive custom spectrometer. This improvement allows the acquisition of four-dimensional point clouds that integrate high-resolution morphological data with spectroscopic information at the pixel level. I show multispectral classification of four pelagic species of free-swimming copepods, microalgae, and microplastics in a column of water using supervised machine-learning techniques trained on previously annotated optical signatures. Lastly, I explore the effects of optical aberrations arising from refraction at oblique incidence angles on an index-mismatch air-water interface. I propose
and experimentally demonstrate that a spherical glass dome co-centered with the scanning element can passively rectify these aberrations by preserving rotational symmetry even at wide scanning angles. In this work, I also study the deployment of hyperspectral cameras for automatic
detection and classification of aquatic biota. In particular, I introduce a staring-type hyperspectral camera design using a liquid-crystal tunable filter (LCTF) as the wavelengthselective element, alongside a novel LED-based illumination system with high and uniform irradiance to facilitate faster acquisitions. I explore a particular use case of automatic evaluation of biofouling growth in coated surfaces for the marine shipping industry, using
a neural network for objective estimations of coverage per group of species.