(Adaptive Sampling and Prediction)
About the ASAP experiment
The goals of the ASAP experiment are: 1) To find out the most efficient ways to use autonomous ocean vehicles, such as undersea gliders, to study marine processes such as the upwelling of cold water that takes place along the Central California Coast; 2) To use the real-time data gathered by autonomous vehicles and other oceanographic instruments to improve computer models of ocean circulation; and 3) To refine these computer models so that they can reliably predict complex processes such as upwelling-related currents.
In order to achieve these goals, ASAP researchers will be using a technique called "adaptive sampling," in which the paths of undersea gliders are modified each day in order to get the best data on upwelling processes. The scientists will also study how fleets of gliders can travel in different "formations" to cover hundreds of cubic kilometers of constantly evolving ocean.
The ASAP experiment builds on the 2003 AOSN experiment, which also involved ocean monitoring by a diverse fleet of vehicles. However, for the ASAP project, the paths taken by the undersea gliders are much more complex and the gliders are coordinated without human input, using a central computer at Princeton University.
The ASAP experiment will gather data using several different types of sensor-equipped autonomous underwater gliders that can be coordinated automatically (without humans in the loop) into patterns designed to gather useful scientific data in the most efficient way possible. The fleet includes Spray gliders from the Scripps Institution of Oceanography and Slocum gliders from Woods Hole Oceanographic Institution.
The Glider Coordinated Control System (GCCS) is used to control some or all of the gliders so that they follow specific path ("glider coordinated trajectories" or GCTs). The GCTs specify not only the desired tracks for each glider, but also the desired relative motion of gliders on those tracks (e.g., relative spacing, relative direction of motion, etc.). The real-time status of the glider planner, the component of the GCCS that governs glider motion, is updated every minute on the Princeton Glider Planner and Status page (the GCT is also posted there). This page also shows the real-time performance of the coordinated glider fleet, illustrated with movies and stills of glider positions, glider estimated flow, and objective analysis mapping error (a measure of glider sampling performance), which are updated every six hours.
The data from the gliders and from a number of other moving and stationary sensor platforms (including aircraft, satellite, high-frequency radar, moorings, propelled autonomous vehicles, ships) are assimilated in real time into three different ocean models. The HOPS model, the ROMS model and the NCOM model all update their nowcasts and forecasts of ocean fields every six hours.
All observational data and model outputs are made available in near-real time on a central data server at MBARI. Key comparative plots of observational data and model output are updated every morning on the “Summary” pane of the virtual control room. These plots provide a concise and informative synopsis of past, present and predicted future state of the ocean as provided by observations and models.
The glider coordinated trajectories specified for the glider fleet are adapted when warranted by changes in the state of the ocean, changes in uncertainty in ocean models, interruptions or changes in glider operations, etc. The ASAP team determines when and how to adapt the specified GCT and/or to modify operation of other ASAP mobile sensors including propelled autonomous underwater vehicles operated by MBARI, MIT and Cal Poly and aircraft operated by NPS. The decision-making process is carried out every other day using the virtual control room, and team members can participate from any internet-accessible location. Any team member can make an adaptation proposal using the “Proposals” panel. Discussion and voting follows using the “Discuss” and “Voting” panels, respectively. The steps of the decision making process are archived so that future implementations can be made more systematic with increased automation.Site maintained by Michael Godin and Kim Fulton-Bennett
Last updated: Jan. 08, 2016