This PDF file contains the front matter associated with SPIE Proceedings Volume 12118, including the Title Page, Copyright Information, table of Contents and Conference Committee list.

The RAN2 sea surface temperature (SST) dataset has been created under the NOAA AVHRR GAC SST Reanalysis 2 (RAN2) project from 40+ years of 4 km Global Area Coverage (GAC) data of the Advanced Very High-Resolution Radiometers (AVHRR/2s and /3s) flown onboard ten NOAA satellites (N07/09/11/12/14/15/16/17/18/19). The data were reprocessed with the NOAA Advanced Clear Sky Processor for Ocean (ACSPO) enterprise SST system. The RAN2 reports two SSTs in the full ~3,000 km AVHRR swath: ‘subskin’ (highly sensitive to true skin SST, while being anchored to in situ depth SST) and ‘depth’ (a closer proxy for in situ data, but less sensitive to true skin SST). Long-term orbital and sensor changes were minimized by daily recalculation of regression coefficients using matchups with drifters and tropical moored buoys, (D+TM), collected within limited time windows centered at the processed day. For N07/09, (D+TM) matchups were sparse and supplemented by ships. The adverse effects of nighttime Sun impingements on the sensor were mitigated by recalculating the AVHRR L1b calibration coefficients, while similar effects of stray light in Earth view were flagged and excluded. Massive cold SST outliers caused by atmospheric contamination following major volcanic eruptions (El Chichon in 1982, and Mt Pinatubo and Mt Hudson in 1991) were filtered out by more conservative cloud screening with the modified ACSPO clear-sky mask. This paper evaluates the performance of the RAN2 relative to the two others available AVHRR GAC SST datasets, NOAA Pathfinder v5.3 (PF) and ESA Climate Change Initiative v2.1 (CCI).

The 2nd Reanalysis (RAN2) of the 4 km Global Area Coverage (GAC) data of the Advanced Very High-Resolution Radiometers (AVHRR/2s and /3s) flown onboard ten NOAA satellites from Sep’1981 – present was performed, and global sea surface temperature (SST) dataset created with the NOAA Advanced Clear-Sky Processor for Oceans (ACSPO) enterprise system. The RAN2 dataset includes two SST products retrieved in a full ~3,000 km AVHRR swath: “Subskin” (highly sensitive to true skin SST) and “Depth” (agreeing much closer with in situ SST, but with a reduced sensitivity). The performance of both RAN2 SST products were improved compared with RAN1, by mitigation of several AVHRR sensor issues. The long-term AVHRR calibration trends were mitigated by daily recalculation of the regression coefficients using matchups with in situ SSTs collected within limited time windows centered at the processed day. Biases with respect to in situ SST were further minimized on a monthly basis by adjustment of the offsets of regression equations based on 31-day moving windows. Massive cold SST outliers, caused by nighttime Sun impingements on the AVHRR’s black body calibration targets, were corrected by interpolating the L1b calibration coefficients between the unaffected parts of the orbit. The Earth view pixels affected by stray sunlight were identified and screened out using the elevated signal in the AVHRR band 2 (0.86 µm). This paper describes the methodology and demonstrates its effects on the RAN2 SST.

Estimation of sea surface temperature (SST) is one of the key products generated from the new generation of geostationary (GEO) sensors such as Advanced Baseline Imager (ABI), onboard the NOAA Geostationary Operational Environmental Satellite “R” series (GOES-R; including G16/17/18), and Advanced Himawari Imager (AHI) onboard Himawari-8 and -9 (H08/09). NOAA generates a consistent line of SST products from these GEO platforms using Non-Linear SST (NLSST) retrieval algorithms implemented in its enterprise Advanced Clear Sky Processor for Ocean (ACPSO) system. The ACSPO NLSST algorithm performs well overall but shows occasional unstable performance under some selected atmospheric conditions, cross-platform biases between G16, G17 and H08 SSTs in the corresponding overlap zones, and increased noise in SST imagery. These issues proved not easy to address in the current complex regression formulation, with many regressors and high degree of multicollinearity among them. This paper performs additional analyses of the ACSPO GEO NLSST algorithm and evaluates contributions of the available IR bands as well as the information content of various predictors. The new NLSST formulation is simpler and more stable and interpretable for understanding and mitigation of the observed SST anomalies. It provides performance comparable with the current more complex ACSPO formulation, and improves SST imagery

Currently, NOAA provides two gridded super collated sea surface temperature (SST) products from low Earth orbit (LEO) satellites (L3S-LEO) using ACSPO system, one from PM (afternoon) and the other from AM (mid-morning) platforms. PM/AM products are split into day/night files, sampling the diurnal cycle at four points (approximately at 1:30am/pm and 1:30am/pm). To meet users’ needs for larger coverage, we are developing a combined SST product which collates all four SSTs into one daily product. We present the ACSPO L3S-daily collation algorithm, where L3S-LEO SSTs are debiased to night-time viewing conditions and combined using an iterative weighting algorithm.

Cyber-physical security is gaining a lot of interest in the last years. In this paper, a focus on security in robots’ coordination is addressed. When multiple robots need to perform a complex task, they need to coordinate between them and in this case implicit and explicit communication can be applied. We considered the problem of unknown area discovery with a collaborative map building among robots. In this case, robots can build autonomously part of the map and they can share their knowledge with other robots. The communication among robots can be guaranteed through network interface but it is important to consider possible security threats that can arise. The following paper proposes the integration security module over well-known simulators such as ROS and GAZEBO with the purpose to support technicians in the evaluation of novel coordination strategies to propose also form the security point of view.

The Ocean plays a critical role in our everyday life and in the future of our planet. Recently, there has been a growing interest in Underwater Acoustic (UWA) sensor networks for a wide range of collaborative applications.In order to allow these applications, the aspects of physical phenomena affecting acoustic channel and restricting the range and bandwidth for the reliable communications cannot be neglected. Our paper discusses about a high level channel model based on Markov Chain approach for the underwater environment. Finite State Markov Model is developed for Packet Error Rate (PER) evaluation in an underwater channel, using the concept of error trace analysis.

In this work, we investigate the use of a radar processing technique to enhance detection performance of a short pulse laser system. One of the challenges experienced by this system is discriminating between the return from an underwater object and the clutter return caused by backscatter in highly-scattering and/or low signal-to-noise ratio (SNR) conditions. Taking inspiration from the radar processing community, we apply the Range-Doppler processing transform to data we collect in our laboratory water tank. This work will present the modified Range-Doppler technique, provide laboratory test tank results, and demonstrate performance improvements achieved using the modified Range-Doppler technique.

Simple, low-cost, rangefinders using parallel beams and a single camera are often used for underwater ranging, but they have limited performance in turbid operating environments such as those associated with coastlines and harbors. A new approach to parallel beam ranging is described which incorporates a combination of simple receiver optics and image processing to significantly reduce the effects of scatter, thereby extending and improving performance. The technique is demonstrated using commercial-off-the-shelf (COTS) cameras and experimental results are presented comparing the new approach to the traditional parallel beam rangefinder. Improvements of multiple attenuation lengths are reported.

The use of autonomous underwater vehicles (AUVs), to potentially carry out underwater exploration missions, is limited due to insufficient onboard battery and data storage capacity. To overcome this problem, underwater docking stations are used to provide the facility of underwater charging and data transfer for AUVs. These docking stations are designed to be installed in the dynamic ocean environment, where the turbidity and low-light conditions are key challenges to hinder the successful docking operation. The vision guidance algorithms based on active or passive markers are typically used to precisely guide the AUV towards the docking station. In this paper, we propose a vision-based guidance method, using lock-in detection, to mitigate the effect of turbidity, and to reject the unwanted light sources or noisy luminaries, simultaneously. The lock-in detection method locks on the blinking frequency of light beacons located at the docking station and successfully vanishes the effect of unwanted light at other frequencies. The proposed method uses two light beacons, emitting at a fixed frequency, installed at the simulated docking station and a single CMOS camera. Proof-of-the-concept experiments are performed to show the validity of the proposed approach. The obtained results show that our method is capable of recognizing the light beacons at different turbidity levels, and it can efficiently reject the unwanted light without using separate image processing for this step of the vision-based guidance algorithm. The effectiveness of the proposed method is validated by calculating the true positive rate of the detection method at each turbidity level.

Optical propagation through turbulence remains a topic of active research and is critically important to the development of novel optical communication systems in both air and water. A widely used tool to study propagation through turbulence are laboratory tanks where optically active turbulence is generated through heating and cooling of the horizontal tank walls, akin to classic Rayleigh-Bénard convection. An important complement to the laboratory setup are numerical simulations that can supplement the sparser laboratory measurements through full fields of temperature and velocity. Such simulations can also provide phase screens for modeling of optical propagation through turbulence. We performed numerical simulations of different configurations of Rayleigh-Bénard turbulence tanks for comparison to other physical and numerical convective tanks. Results then provided the basis for optical modeling and the description of beam wander due to optical turbulence.

Remote sensing modalities can utilize the effects of laser scattering from particulate matter to infer information about environmental conditions present in the optical path. Light contains many degrees of freedom that can be manipulated for sensing. In this work the spatial phase distribution is leveraged through the use of light's orbital angular momentum (OAM). Our sensing method, called optical heterodyne detection of orthogonal OAM modes (OHDOOM), uses the distortion of the optical signal to determine the presence of environmental disturbances. OAM beams are sensitive to optical disturbances that induce phase variations, in turn, spreading power among other OAM modes. A set of experiments are performed using different solutions of particulate matter to create a turbid medium. The experimental results showed that OHDOOM is most likely sensitive to a turbid medium containing particles larger than the wavelength.

Maritime environments with their ever-changing conditions could be hostile to cargo, civilian, and military vessels. In recent years, the adversarial attacks have increased considerably and more marine areas are deemed unsafe than ever before. Detection and recognition of fast moving boats is of particular interest since they project more hostility and possess high degree of maneuverability. However, high-speed boats generate long-lasting ship-wakes. The ship-wakes that are due to unsteady hydrodynamics characteristics of ocean water, naturally formed behind the boats, and can reveal several important features about the boats as well as depicting their spatiotemporal behavioral activities. Through remote sensing applications and by scrutinizing the wake patterns via robust deep learning classifiers, both boats’ features and their hostile or normal behavioral activities can be dependently discriminated. This paper presents methods for simulation of different speedboat behaviors and their associated wake formations using Houdini’s FLIP hydrodynamic particle simulation toolbox. Using this technique, we created different boat activities and obtained their simulated ocean wake formations. From each model, we extracted and formed topographical maps corresponding to the wake formations and employed them as ocean wake layer in our remote sensing simulation software, called IRIS. IRIS simulates large-scale physics-based electromagnetic (EM) environments. Using IRIS-EM techniques, we modeled different physics-based marine boats along with exemplar wake formations. We systematically generated and annotated multi-look, multi-range simulated synthetic aperture radar imagery from our marine test models. In this paper, we compare our hydrodynamic particle modeling approach vs an image-based approach and present a comparison of results of both methods and discuss their trade-offs as candidates for the Synthetic SAR imagery generation and deep learning systems development.

This paper presents experiments using a time of flight (ToF) camera modified to use 525 nm green laser illumination to capture amplitude and depth images of an underwater scene. Experiments in object imaging and ranging were conducted in both clear and turbid water. 3D imaging using flood illumination was successfully performed in clear water and in some turbid water conditions. Ranging using collimated laser beams was performed in turbid water. Several major error sources were observed, including low illumination levels, fixed pattern noise, and backscatter contribution to the phase measurement. To attempt to address these concerns, multiple lasers were used to improve illumination levels and spatial frequency domain filtering was performed to mitigate fixed pattern noise. Additionally, experiments with using multiple modulation frequencies suggested that there may be potential for discriminating backscatter from object reflection.

As one of the classic fields of computer vision, image classification is a topic that has expanded exponentially in terms of usability and accuracy in recent years. With the rapid progression of deep learning, as well as the introduction and advancement of techniques such as convolutional neural networks and vision transformers, image classification has been elevated to levels only theoretical until modern times. This paper presents an improved method of object classification using a combination of vision transformers and multilayer convolutional neural networks with specific application to underwater environments. In comparison to previous underwater object classification algorithms, the proposed network classifies images with higher accuracy, shorter training iterations, and deployable parameters.

We develop a remote hyperspectral (HS) imaging work flow that relays spectral and spatial information of a scene via a minimal amount of encoded samples along with a robust data reconstruction scheme. To fully exploit the redundant and multidimensional structure of HS images, we adopt the canonical polyadic (CP) decomposition of multiway tensors. This approach represents our HS cube in a compressive manner while being naturally suitable for the linear mixing model, commonly used by practitioners to analyze the spectral content of each pixel. Under this low CP rank model we achieve frugal HS sensing by attenuating and encoding the incoming spectrum, thereby faithfully capturing the information with few measurements relative to its ambient dimensions. To further reduce the complexity of HS data, we apply image segmentation techniques to our encoded observations. By clustering the pixels into groups of endmembers with similar structure, we obtain a set of simplified data cubes each well approximated by a low CP rank tensor. To decode the measurements, we apply CP alternating least squares to each set of clustered pixels and combine the outputs to obtain our final HS image. We present several numerical experiments on synthetic and real HS data with various levels of input noise. We demonstrate that the approach outperforms state of the art methods, achieving noise attenuation while reducing the amount of collected data by a factor of 1/14.

Phase light modulator has been used in many underwater applications, such as turbulence mitigation, or laser beam shaping, to improve imaging and communication in the underwater environment. Liquid crystal on silicon (LCOS) or Liquid Crystal Display (LCD)-based phase light modulators are used in these applications. In recent years, Texas Instruments Digital Light Processing division developed a piston-mode Phase Light Modulator MEMS device (TI-PLM). One of the benefits of the device is its capability of supporting a high frame rate (i.e., 5.7kframes/sec). In this paper, we evaluated this TI-PLM device on an optical benchtop. While the main focus was the image quality of the Computer-Generated Hologram (CGH), the results also shed some light on this device’s capability for other PLM applications.

This publication will demonstrate recent advances of a self-referencing homodyne interferometry technique for mitigating atmospheric turbulence. The results will be quantified by using QR codes to document the machine-readable performance gain by using Digital Adaptive Optics when compared to a traditional imaging camera.

Water is the new oil of the 21st century due to increased consumption and demand. High-quality water with free of contamination is vital for human beings and many industries such as oil and gas, petrochemicals, pharmaceuticals, and food. To meet the massive amount of freshwater production, United Arab Emirates (UAE) relies on the energy-intensive Multi-Stage Flash (MSF) and Multi-Effect Distillation (MED) technologies to provide fresh water for various applications. These energy-intensive processes consume a significant share of UAE oil and gas. In general, thermal desalination is an energy-intensive process. UAE is shifting to use Reverse Osmosis (RO) desalination for the freshwater production. To reduce the energy consumptionand the pretreatment stages onsite monitoring of the seawater intake has to be intensively recorded. This study discusses a design of a seawater test station, which has a sensor network to measure the quality parameters of seawater, including the water’s pH.

Water Desalination is the process of desalinating seawater into freshwater. The desalination process is typically done by introducing seawater into the plant by an offshore pipeline. Multiple water properties need to be measured and analyzed to assure the feed seawater is suitable for desalination processing to prevent fouling, scaling, and corrosion of equipment and reduce operational costs. These parameters include seawater temperature, total dissolved oxygen, turbidity, conductivity, total dissolved solids, and pH. This paper will discuss developing and integrating a low-cost, highly scalable sensor subsystem measuring water conductivity in the Arabian Gulf.