MIT Lincoln Laboratory hauled a sharper geophone onto drifting Arctic ice to sort whale song from ship hum

A geophone measures vibration in the sea ice rather than pressure in the water, so the instrument and its subject share a medium ^[2]. That ordering matters. A library of fracture signatures, or a classifier that can tell cracking ice from a beluga call from a vessel engine, depends on first knowing what the ice does to a waveform on its way to the sensor ^[4]. Evans puts the propagation question ahead of the separation question, and the 2024 record, gathered with integrated commercial off-the-shelf parts, is why he can pose it with specifics instead of in the abstract ^[3][1].

What that question consumes is hours of ice time, and ice time here means a runway on a drifting floe, put up by the Navy's Arctic Submarine Laboratory for three weeks every second year ^[9][10]. March 2026 was the third such window for Evans and Whelihan, who have been going since 2022 ^[8][20]. Most of it closed on them. Their first trip in slipped by a week when blowing snow halted inbound and outbound flights, and after they landed on March 7 nothing moved for five days ^[12][16]. Six to nine flights a day is the normal mobilization tempo, so five idle days works out to roughly 30 to 45 sorties that did not fly ^[17][19]. Evans says the flags on poles about a hundred feet from the command tent were sometimes not visible ^[18].

The second payload never got that far. Havguard, a Norwegian defense technology startup, supplied a modem that communicates through ice using magnetic fields, on the reasoning that radio-frequency signals are rapidly diminished by seawater ^[13]. The account places the failures in staging rather than in the link itself ^[14]; the magnetic-field approach was never actually put against Arctic ice this year, which makes this a weaker negative result than it looks.

The account describes the deployment and the weather, and reports no findings from the 2026 record ^[22], so whether the higher-fidelity geophone changed any answer remains unknown. The old baseline stands unchallenged too: the record includes no ice-thickness measurement from the camp and no comparison suggesting the 2024 signatures have stopped holding ^[23]. A baseline built this way also has a location problem, since the platform drifts and each record therefore comes from a place that has moved ^[10]. What is on the record is a stated research need and a two-year cadence for meeting it, aimed eventually at cheap sensors left in place to listen without anyone flying in ^[3][8]. The distance between that ambition and a blizzard-gated three-week window is what this program has to close.