In making the ten-year leap from prototype to scientific workhorse, low-cost and nimble MERMAID established itself as a vital partner in the seismic exploration of the Earth’s deep interior. Freely-drifting midwater hydrophones will forever fit in a balanced diet that must also contain increased ocean-bottom sensor coverage, among permanent networks of (is)land-based sensors, and possibly even more exotic types of data gathering. Employed in this fashion, MERMAIDs will swarm in the ocean like nodes on land (today’s equivalents to the "Texans" from yesteryear), much as today’s temporary PASSCAL deployments continue to so usefully complement longer-lived installations such as the USArray. Autonomous hydrophones appeal but won’t replace three-component ocean-bottom seismometers (OBS) as a backbone, in the sense of, e.g., the PacificArray, espoused by Hitoshi Kawakatsu at the Earthquake Research Institute of the University of Tokyo. But while those OBS arrays remain dreams of the future, MERMAID rules in the ocean today.

Meanwhile, back in France, Géoazur research engineer Yann Hello and the team led by Olivier Philippe at OSEAN SAS set their sights upon a from-the-ground-up redevelopment of MERMAID that would solve the last of the sticking points: the longevity of its autonomy.

Out went the narrow spar-buoy design of the SOLO and APEX/Rafos floats. In its stead came a 17-inch (43 cm, diameter) glass sphere encapsulating (lithium) batteries, electronics (a low-power acquisition board and a controller), and hydraulic components regulating the transfer of oil from an internal reservoir to the external bladder to achieve neutral buoyancy in the water column. In the current commercial version, MERMAID can maintain acoustic operability down to about 3000 m, a depth determined by the rating of the HTI-96-MIN/HEX hydrophone. Cruising comfortably down to depths of 4000 m, its hydraulics hold down to 5500 m, and the sphere itself withstands pressures up to 7000 m. New models designed to sustain performance to deeper, possibly full-ocean, depths will have upgraded pumps and carry different hydrophones.

Hiroko Sugioka from Kobe University and Masayuki Obayashi from JAMSTEC led an international crew on a test expedition aboard the training ship Fukae-maru during a 24 h deployment in which MERMAID was so lucky as to catch a local magnitude 5.2 event. Recording at 40 Hz with 24-bit precision and surfacing approximately weekly to report a handful of quality-ranked 200 s segments of triggered seismograms endows the third-generation MERMAID model with a lifetime of approximately 5 years.

The breakdown of the time-domain signal into wavelet coefficients over five scale bands serves to identify the likely nature of the signal and to give it a quality rating. It also allows for reducing the data transmission rates. Under typical settings MERMAID only transmits scales 1 through 4 out of 5 wavelet scales, which cuts the data rate by half, for an effective bandwidth out to 20 Hz. With these settings IRIDIUM communication costs add up to about $75 per float per month. At every surfacing MERMAID not only transmits seismogram, but it also receives a command file with instructions, mission parameters and software settings. In this manner, MERMAID, while passively drifting, can be "piloted" to some degree. As and when future versions will be equipped with "landing" capabilities, MERMAID can become a temporary ocean-bottom hydrophone (OBH), which will allow some researchers to shift their focus from the global recording of teleseismic earthquakes to monitoring local and regional events without fear of drifting away from the area of interest, listening for landslides, cracking glaciers, and other such pursuits.

Software and analysis development targets moved to the perfection of post-processing routines of the transmitted data files. Sébastien Bonnieux from the Université de la Côte d’Azur was the architect behind Automaid, a suite of Python tools to handle data recovery, dive reports, system log and command files, suitable for the end user whose main interest is the seismological data analysis. Joel Simon from Princeton University became the force behind the analysis pipeline, mostly written in MATLAB, focusing on earthquake event-catalog matching of the probabilistically triggered but unidentified segments, on making accurate travel-time measurements for individual earthquake phases bolstered by information-theoretical considerations, at multiple wavelet scales, and on their uncertainty quantification.

Jonah Rubin, an undergraduate from the University of Vermont, built the code and the dynamic tools powering this website. Undergraduates at Princeton University, under the current lead developer Peter Mwesigwa, managed to release a first version of an iOS MERMAID tracker app, Adopt-A-Float, written in Objective-C, for use in research and educational settings. All source code is being freely shared under version control on the GitHub platform, adhering to FAIR principles (Findability, Accessibility, Interoperability, and Reuse) as best as possible.

With Yongshun John Chen and his team in the lead at the newly established Department of Ocean Science and Engineering at the Southern University of Science and Technology (SUSTECH) in Shenzhen, China, the French (Géoazur), Japanese (Kobe/JAMSTEC) and US (Princeton) partners joined forces. The volcanic islands of French Polynesia became the focus of the first large-scale science experiment: the ongoing the South Pacific Plume Imaging and Modeling (SPPIM) project. Fifty MERMAIDS so far were launched, with twenty-five additional units in the queue (as of 2021).

Polynesian geography is one of myriad islands, seamounts, hotspot tracks, and large swaths of anomalously elevated oceanic crust known as the South Pacific Superswell. Among all global seismological models there is consensus that broad regions of anomalously slow wave velocities lie at the base of the mantle under the South Pacific: a region known as the Pacific Large Low-Velocity Province (LLVP). Global tomography suggests that a source region in the deep mantle may feed the surface expression of hot spots and swell via conduits of hot uprising rock that may stretch from the core-mantle boundary to the surface. Elucidating the fine-scale nature of this planetary heat-engine plumbing system through regional-scale tomography cannot be accomplished without considerably widening the aperture of seismic arrays beyond the few available ocean-island stations: hence that is the primary scientific target of the SPPIM collaboration for data collection and analysis. Read on about the Fourth Generation.