Ruoying He

This project aims to achieve measurable and significant improvements in short-to-medium range (1- to 10-day), subseasonal, and long-range (3- to 6-month) prediction skill of sustained continuous operational forecasts of Gulf of Mexico (GoM) dynamics. The outcomes will support maritime industry safety by reducing risks associated with offshore energy exploration and production, fisheries stock assessment and forecasting, and forecasting of hurricanes and other natural hazards. This will be accomplished through implementation of enhancements to established and next-generation publicly available Gulf of Mexico forecast modeling systems operated by the US Navy and NOAA, improvements in model utilization of observational data, and development of specific forecast products targeted toward end users in industry. Building upon prior activities of project team members conducted during UGOS-1, Observing System Simulation Experiment (OSSE) techniques will be applied to quantify the utility of future observing systems and guide optimal and rapid adaptive sampling strategies. Capabilities for assimilating these data will be integrated into enhanced configurations of Navy GoM forecast models and a next-generation NOAA modeling system and the improvements in the forecasts will be assessed during real-time field campaigns using the new sampling strategies. Integral to the GOFFISH Consortium are end users of GoM forecasts including partners from industry, NOAA fisheries, and hurricane researchers. These partnerships will guide design, development, and transition of prediction tools and other model products for application of the forecast models and observations. In particular, statistical and machine learning tools will be developed to complement model forecasts to increase predictability of full water column currents, including deep currents associated with Topographic Rossby Waves and other features that impact critical oil and gas infrastructure. Model forecasts will also be applied to fisheries recruitment forecasting and dynamic management building upon existing collaborations and advances made by project team members, and to coupled ocean-atmosphere-wave modeling systems to enhance predictability of hurricanes, sea state, and interseasonal climate forecasts.

In the southeastern U.S. coastal ocean, the Loop Current/Florida Current/Gulf Stream system unites shelf seas from Louisiana to Florida in the Gulf of Mexico and from Florida to North Carolina along the East Coast. This energetic, deep ocean feature flows along the continental shelf in the entire SECOORA footprint, strongly affecting circulation on adjacent continental shelves and providing a conduit for the transport of nutrients, heat, and marine organisms between the sub-regions of the southeastern U.S. coast. The development of a regional coastal observatory in the service of societal goals elaborated in the national IOOS plans therefore requires a regional-scale approach to incorporate this circulation and transport. SECOORA has created a robust regional coastal ocean observing system (RCOOS) to acquire atmospheric and oceanographic observations from HF radar, coastal and oceanographic stations, and autonomous vehicles. Because coastal and ocean observations cannot be collected everywhere and at all times, carefully calibrated numerical models that integrate and assimilate scattered observations are an essential element of the observatory, to fill data gaps and improve our understanding and prediction of marine environment conditions. To contribute to SECOORA??????????????????s efforts in enhancing and expanding the regional ocean observatory, we propose to implement the next-generation data assimilative, fully-coupled ocean-atmosphere-wave-marine ecosystem prediction system. Such a system i) will predict coastal ocean conditions (including surface wind, air temperature, air pressure, wave height, wave direction, sea level, three-dimensional ocean currents, ocean temperature, and salinity, as well as marine biogeochemical variables such as NO3, NH4, Chl-a, DIC, pCO2, and pH) over the entire SECOORA footprint; ii) can provide detailed sub-regional marine environment predictions though relocatable grid refinement and nesting technology; and iii) update and transmit trustworthy model-data integration and prediction information to end users in a timely fashion.

In the midst of a heightened interest in obtaining high-resolution observations of spatiotemporally evolving climatological phenomena, a corresponding interest in deploying persistent, mobile observational networks has arisen. Such networks ultimately rely on a renewable but stochastic resource, enabling continuous operation, and can adjust their locations to follow atmospheric or oceanic patterns that cannot be captured through a fixed observational network. Ideally, the renewably powered, persistent systems can act as hosts and mobile charging stations to non-renewable fixed-location smaller-scale systems that are used for detailed time series measurements and must periodically rendezvous with an appropriate host. Examples of such networks include fleets of sailing drones, solar-powered autonomous surface vessels, wave gliders, or solar-powered aircraft, which can serve as hosts to undersea gliders, floats, and/or quad copters. The proposed research will create and validate a multi-layer dynamic coverage-based adaptive sampling system for addressing the cyber-physical systems (CPS) control, communication, computing, and energetic challenges associated with the general mobile observational configuration, which application to a complex marine environment described above. These challenges are numerous, significant, and significantly differentiated from existing oceanic and atmospheric observational approaches: 1) A suitable adaptive sampling approach (as opposed to pre-planned missions, which serve as the basis for most observational work conducted by renewably powered drones) must balance consideration of stochastic mobility (whereby instantaneously achievable speed and the reachable domain of each agent depends on the renewable resource being accessed) with the scientific characterization of a stochastic resource. This challenge is heightened when the renewably powered systems act as host agents, requiring the satisfaction of rendezvous constraints within the stochastic resource. 2) Communication constraints prohibit exchange of scientific data between a full fleet of agents at most times (due, for example, to distance constraints and/or underwater bandwidth limitations), requiring the aforementioned adaptive sampling to be performed in a decentralized or distributed manner, under partial observability. 3) Computing limitations on-board each agent, as dictated by energetic costs and space constraints, restrict the level of local computational complexity that can be handled in performing the adaptive sampling.

In response to NSF call for National Artificial Intelligence (AI) Research Institutes, we propose to develop The AI Institute: Artificial Intelligence for Environmental Sciences (AI2ES). AI2ES is a transdisciplinary center that will usher in a new era of Trustworthy AI for environmental science. Its is a synergistic collaboration of seven academic institutions, a federally funded research agency, government collaborators, and the private sector. AI2ES draws upon the deep individual expertise of leaders in artificial intelligence (AI), ES, and risk communication. It creates a unique nexus of collaboration that will deliver the next-generation of trustworthy AI, driven by use-inspired risk prediction for ES. The critical challenge that AI2ES addresses is developing trustworthy AI that will be directly utilized by multiple end-users to mitigate the effects of high-impact weather. AI2ES will uniquely impact the US by developing physically-based trustworthy AI techniques that will directly improve prediction, understanding, and communication of high-impact environmental hazards.

Knowledge of connectivity among deep-sea metapopulations is central to predicting population and community dynamics, as well as guiding conservation and understanding biogeographic patterns. Connectivity among metapopulations may be predicted by bio-physical models, and tested with molecular or geochemical tools. However, bio-physical models of dispersal and connectivity require data on larval biology, such as spawning times and locations, pelagic larval duration, vertical distribution in the water column, and swimming patterns. Reliable information on larval biology for deep-sea chemosynthetic organisms is currently inadequate, so biophysical dispersal models have generally relied on untested biological assumptions. The extent and timing of vertical migrations remain unknown for all deep-sea larvae. It is not known, for example, how much of the larval period is spent drifting near the bottom. Dispersal strategies can have profound effects on predictions of metapopulation connectivity. In this proposal, we will model larval transport of select methane seep animals in the Gulf of Mexico to determine why some species are able to disperse around Florida and colonize seeps on the Western Atlantic Margin, whereas others are not. It is hypothesized that interspecific differences in dispersal depth may explain this major biogeographic pattern. Larvae will be sampled using the newly developed SyPRID plankton sampler deployed on the AUV Sentry, and by trapping larvae in year-long deployments of larval collectors on bottom moorings. The work will focus on seep sites at three depths in the northern Gulf of Mexico and at two depths on the Atlantic Continental Margin. Oxygen isotopes and elemental composition of larval and juvenile mollusk shells will be used to obtain independent information about the depths where larvae drift and the variability in dispersal trajectories. Intellectual Merit Reliable connectivity estimates are increasingly important in marine conservation biology and phylogeography as well as metapopulation ecology, yet reliable biological parameters for biophysical models in the deep sea remain unavailable. This project will advance the entire field by using state-of-the-art observations on larval migration, physiology and movements to inform the bio-physical models. The methods will find application not only with chemosynthetic organisms, but also in many other deep-sea environments, including ones that are currently endangered by human activities such as trawling and mining. Broader Impacts In addition to advancing applied fields such as conservation biology and the siting of marine reserves, results will be disseminated through museum exhibits, school programs, curriculum development and social media. The project will contribute to the development of human resources in oceanography by providing postdocs, graduate students, and many undergraduates the practical skills associated with reproductive and larval work at sea, as well as interdisciplinary training in biological, physical, and geochemical oceanography and ocean modeling. Students, including underrepresented minorities, will go to sea and learn the seldom-taught yet essential skill of sorting plankton and recognizing larval forms using morphological criteria.

The NSF Ocean Observatories Initiative (OOI) Pioneer Coastal Array will be relocated to the Southern Mid-Atlantic Bight (S-MAB) in 2024. The array will be sited in the region between Cape Hatteras, NC, and Norfolk Canyon, VA. Science questions that will be informed by the S-MAB Pioneer Array include 1) What are the roles of land-ocean interactions, shelf-sea/deep-ocean interactions, and intrinsic variability, in driving the export of S-MAB waters into the Slope Sea? 2) How do cross-shelf transports impact regional biogeochemical cycling and marine ecosystem dynamics? Ocean numerical modeling analysis of the complex and dynamic ocean circulation is needed to offer a virtual ocean framework ?????????????????? a ????????????????digital twin??????????????? ocean ?????????????????? that can be used to inform specific array design decisions, such as mooring siting, spacing, and instrument/sensor selection and configuration. This project will 1) characterize variability of regional circulation, 2) define 3-D transport pathways and their variability, 3) assess array design, and 4) develop an observation impact study. At the conclusion of the project, we will partner with the OOI design group to explain how model-based information contributed to the design process, with a view to strengthening future model-based co-design of NSF-supported field programs, including the next Pioneer redeployment five years hence.

Hurricanes are a devastating destructive force to the coastal and inland portions of the eastern United States and states lining the Gulf of Mexico. With ever-increasing numbers of the population moving to these areas, there has been a dramatic increase in risk to both life and property over the last several decades. The destructive force of these storms is not just caused by their high winds, but frequently is due to overland flooding of the coast extending far inland. Even weak hurricanes that remain nearly stationary for several days over land can produce immense destruction as a result of flooding from intense precipitation. This precipitation runs off into streams, rivers, and embayments that are already stressed due to storm surge. The water then breaches banks and floods communities. The goal of this project is to improve forecasting during hurricane events to better predict and improve flooding forecasts so coastal and inland communities will be able to better prepare for, mitigate the effects of, and prevent death and destruction.

This project has the overall goal of achieving a greater understanding of the physical processes that control the circulation in the Gulf of Mexico (GOM), in particular the Loop Current and Loop Current eddy separation dynamics, through the synthesis of historical measurements with advanced data assimilative numerical models. The effort has participation of leading modeling teams with existing gulf-wide data assimilative prediction systems. Three specific tasks will be pursued, including: 1) GOM circulation hindcast in 2009-2011 and 2014-2015; 2) Retrospective forecast of GOM circulation in 2009-2010, and understanding the added values of deep observations; 3) Observing System Simulation Experiments (OSSE), in which each modeling system will be tasked to perform OSSE to quantitatively assess the values of different ocean observing system components to the circulation forecast. Outcomes of this study will include: (1) new knowledge of the GOM circulation dynamics; (2) improvements in ocean forecasting methodology; (3) ensembles from multi-model OSSE experiments that will be used to provide references on observational design criteria, instrumentation locations, sampling intervals before the start of the field campaign in the NAS Loop Current phase 2 program.

The ????????????????Processes driving Exchange At Cape Hatteras??????????????? (PEACH) program is a comprehensive examination of the dynamics of shelf-deep ocean exchange where basin-scale and shelf flows converge. An observational program that combines in situ and remote observing platforms will measure exchange processes at temporal scales from days to years throughout the Cape Hatteras region. Idealized and realistic numerical modeling will complement analysis of the observations and enable exploration of dynamical balances. A. Intellectual Merit. The Cape Hatteras region is a natural laboratory for examining shelf-deep ocean exchange processes. Utilizing an optimized combination of mobile, fixed, and remote sensing platforms, the PEACH observational campaign will resolve relevant shelf-ocean exchange processes across an unprecedented range of spatial and temporal scales. State-of-the-art numerical modeling will compliment the observations. The understanding gained by investigating the wide seasonal range of parameter space will facilitate exploration of how shelf circulation and shelf-deep sea exchange may evolve due to observed and projected longer term shifts in regional and basin-scale circulation, hydrography, and atmospheric forcing. B. Broader Impacts The dynamical understanding of shelf-deep ocean exchange gained through PEACH will be applicable to other regions where shelf and basin-scale currents converge (e.g., the Kuroshio/Oyashio and Brazil Malvinas confluences); both the variability in forcing situations to be encountered during PEACH and the series of idealized simulations allow insights across a wide dynamic range. Though the focus of PEACH will be on physical processes, ancillary measurements collected during the field campaign (e.g., chlorophyll fluorescence, oxygen saturation, and acoustic backscatter from gliders) will allow for future investigations of the role that shelf-deep ocean exchange plays in biogeochemical cycling and ecosystem dynamics in the Hatteras region, an area of high biological diversity that is home to many commercially important species. Predictive capacity gained through PEACH will play a critical role in understanding and mitigating the spread of pollutants from potential future oil and gas extraction in the region. PEACH will include significant outreach and educational efforts. We will develop an exhibit about our work for the Graveyard of the Atlantic Museum in Hatteras, NC, where HF radar sites will be installed. During trips to deploy and recover instrumentation, PIs will give public talks and display instrumentation to educate visitors about the oceanography of the area. If bunks are available during PEACH cruises, we will invite members of the Graveyard of the Atlantic Museum or other organizations to join the cruises to gain insight into oceanographic fieldwork and to share their experiences publicly in realtime. The PIs will also continue existing outreach activities. For example, Todd and Andres will continue their involvement with the Woods Hole Partnership Education Program, a summer program in ocean sciences for undergraduates from underrepresented groups, and Edwards has an ongoing collaboration with the Society of Women Engineers through their Girls Engineer It! Day, a daylong event for girls in grades 6-12 to explore engineering and related fields. PEACH will support three early career scientists (Todd, Andres, Edwards), will train one postdoctoral researcher and four graduate students (NCSU, SkIO, UNC), and will involve undergraduate students in operation of gliders (SkIO, UNC).

We propose to create a regional glider network that will operate as a cooperative effort, with joint deployments/recoveries, piloting, and data management, pooling resources to take advantage of complementary assets (instruments, personnel, and ship access). USF and SkIO will coordinate ship/charter activities, and glider owners will cover cost of supplies. GT and SkIO will develop and test optimal control algorithms for glider position and mission within and independent of model predictions from NCSU and UGA. UNC and USF will coordinate data management, and NCSU and USF will prepare glider data for assimilation into coastal models. Shelf-wide glider surveys will be collected through simultaneous deployment of 3-5 gliders on the west Florida shelf (WFS) and in the South Atlantic Bight (SAB), incorporating lessons learned, efficiencies gained, and input from other SECOORA stakeholders.