On this page
For more than a decade IceCube was the only show in town for high-energy neutrino astronomy. Its cubic kilometer of instrumented Antarctic ice delivered the discovery of the diffuse astrophysical neutrino flux in 2013, the TXS 0506+056 blazar association in 2018, the steady NGC 1068 source in 2022, the tau-neutrino sample in 2024, and the Galactic Plane detection — among many other firsts. But IceCube alone has limits. It sees one hemisphere of sky best, it has a fixed angular resolution set by the ice properties and the 125-meter string spacing, and it cannot easily be upgraded to a different size or configuration.
The next generation of high-energy neutrino astronomy is unfolding through four complementary facilities at four different sites: IceCube-Gen2 at the South Pole, KM3NeT in the Mediterranean Sea, Baikal-GVD in Lake Baikal in Siberia, and P-ONE in the Pacific Ocean off Vancouver Island. Each instruments a different transparent medium with a different geographic position, and each contributes a distinct field of view, angular resolution, and energy reach. Together they will provide all-sky coverage at IceCube-comparable or better sensitivity, with redundancy in flavor reconstruction, source localization, and systematics that no single detector can deliver.
This post is about the four sites, what makes each distinctive, and how their combination defines the high-energy neutrino sky of the coming decade.
Why ice or water at all
A high-energy neutrino interaction produces either a high-energy charged lepton (in charged-current events) or a hadronic-electromagnetic shower (in neutral-current events). Either way, the resulting relativistic charged particles emit Cherenkov light at a fixed angle of about 41 degrees to their direction of motion in a medium with refractive index near 1.33.
Detecting that light over a useful volume requires a medium with three properties. It must be transparent enough at blue-green wavelengths that photons travel 50 to 200 meters before absorption — long enough that an array of optical sensors spaced at tens of meters can detect a single event from multiple angles. It must be dense enough that the target mass per unit volume is meaningful. And it must be available in essentially unlimited supply, because the detector volumes required are enormous: a useful astrophysical detector starts at roughly a cubic kilometer of instrumented medium.
Only two media meet all three criteria at affordable cost. Glacial ice at depth, with the bubble-free structure produced by accumulation under hundreds of meters of overlying snow, has absorption lengths of 50 to 200 meters depending on depth and wavelength. Liquid water of sufficient depth — to suppress atmospheric muon backgrounds — and of sufficient optical clarity, found in deep ocean trenches and a few deep lakes, similarly provides transparency over the relevant wavelengths.
Both have engineering complications. Ice has trapped dust layers from past climate events that scatter light and produce inhomogeneities in the optical properties as a function of depth. Water provides better optical isotropy but requires sea-floor cabling, biofouling-resistant sensor housings, and substantial logistics for deployment and maintenance. The two trade off in ways that motivate building both.
IceCube and IceCube-Gen2: the South Pole
IceCube was built between 2005 and 2010 at the geographic South Pole, with 5,160 digital optical modules deployed on 86 vertical strings frozen into the deep glacial ice between 1,450 and 2,450 meters depth. The instrumented volume is approximately 1 km³. Operating since 2010, the detector has produced essentially all the major results in high-energy neutrino astronomy.
IceCube-Gen2, the planned expansion, would add roughly 120 additional strings spaced 240 meters apart (versus IceCube’s 125 meters) over an expanded footprint of about 8 km³. The wider spacing trades angular resolution for fiducial volume, optimised for the higher-energy events where statistics, not resolution, is the limiting factor. A separate radio-detection sub-array adds sensitivity to the cosmogenic neutrino flux at exa-electronvolt scale. Combined with the IceCube Upgrade — a denser in-fill aimed at GeV-scale atmospheric oscillation and mass ordering — the Gen2 era will give IceCube an extended scientific reach across multiple energy decades.
The South Pole site has unique advantages. The ice itself is the deepest, oldest, and clearest of the candidate media. The Antarctic location places the detector at the bottom of the world, so every neutrino arriving from the Northern Hemisphere has traversed essentially the full Earth, with the strongest possible filtering of atmospheric muon backgrounds. The infrastructure of the Amundsen-Scott South Pole Station provides continuous power and data uplink.
The disadvantages are practical. Deployment is constrained to the austral summer drilling season. The ice’s optical properties are not perfectly uniform — there are dust layers at certain depths that produce localised scattering. And the South Pole’s sky coverage is necessarily Southern-Hemisphere-poor: the Galactic Center and Galactic plane, the most likely places to find galactic neutrino sources, are best seen from a Northern Hemisphere detector.
KM3NeT: the Mediterranean
KM3NeT consists of two distinct detectors at two different sites in the Mediterranean Sea. ARCA (Astroparticle Research with Cosmics in the Abyss) is being deployed off the southern coast of Sicily at about 3,500 meters depth, optimised for high-energy astrophysical neutrinos in the TeV-to-PeV range. ORCA (Oscillation Research with Cosmics in the Abyss) is being deployed off the coast of Toulon at about 2,500 meters depth, optimised for atmospheric neutrinos at GeV energies with the mass ordering as its central goal.
The two facilities together aim for an instrumented volume of about 1 km³ for ARCA and 7 Mt for ORCA. The first ARCA strings have been operating since 2016, with the array under continuous expansion through the late 2020s. ORCA has a smaller footprint with much denser spacing.
The Mediterranean site has several advantages for the high-energy program. The deep saltwater has excellent optical properties at blue-green wavelengths, comparable to or exceeding IceCube’s ice in clarity. The detector location in the Northern Hemisphere puts the Galactic Center in the southern sky, where upward-going events are atmospheric-muon-filtered through the Earth and the source can be searched optimally. The site is also at a different geographic location from IceCube, providing time-of-flight constraints for any future multi-messenger event.
The challenges are different from IceCube’s. Marine biofouling, currents, and bioluminescence from deep-sea organisms all complicate operations. The sensor housings have to withstand 350 bars of hydrostatic pressure indefinitely. Submarine cables must be laid and maintained. But the engineering has worked: KM3NeT’s first results, including the recent KM3-230213A event analyzed in 2023, demonstrate the technology at scale.
Baikal-GVD: Siberia’s deep lake
The Baikal Gigaton Volume Detector is deployed in Lake Baikal, the world’s deepest freshwater lake at over 1,600 meters in the southern basin. Strings of optical modules are deployed from the ice surface in winter, when the lake is frozen and a stable platform exists for deployment activities. The detector cluster structure mirrors KM3NeT — multiple “clusters” of 8 strings each, each cluster about 60 meters in radius and instrumenting depths from 600 to 1,300 meters below the surface.
As of 2025, Baikal-GVD has 12 clusters deployed, with an instrumented volume of about 0.5 km³ — already the second-largest high-energy neutrino detector after IceCube. The plan is to expand to roughly 1 km³ by the end of the decade. The detector has been continuously delivering science output since 2018, with steadily improving sensitivity to the diffuse astrophysical flux.
Baikal-GVD’s advantages are freshwater and the relatively simple winter deployment from the ice surface. Freshwater has different optical properties than seawater, with lower bioluminescence backgrounds and somewhat different absorption spectra. The annual ice cover provides an unusually reliable deployment platform compared to ocean-based sites that require ship-based logistics.
The disadvantages include the relatively shallow lake depth, which limits how much vertical extent the instrumented strings can have, and the geopolitical situation, which has complicated international collaboration.
P-ONE: the Pacific Ocean Neutrino Experiment
The Pacific Ocean Neutrino Experiment is under construction off the coast of Vancouver Island in the deep waters of the Cascadia Basin, at about 2,700 meters depth. The first cluster of strings was deployed in 2024, with the full P-ONE configuration aiming at 70 strings spread over 10 clusters and an instrumented volume of about 3 km³ — comparable to IceCube-Gen2 but in saltwater rather than ice.
P-ONE leverages existing infrastructure from the Ocean Networks Canada cabled ocean observatory. The submarine cabling for power and data is already in place, having been deployed years earlier for oceanographic and seismic monitoring. This is a major cost saving versus a greenfield site.
The Pacific site sits at moderate northern latitude — comparable to KM3NeT but on the opposite side of the world. The combination provides 24-hour coverage of the southern sky regardless of which side of Earth is between the source and detector at any given moment, which matters for monitoring transient sources and for time-of-flight constraints on multi-messenger events.
P-ONE has the youngest detector design and benefits from the lessons of all the earlier facilities. It uses string-based deployment similar to KM3NeT and Baikal-GVD, with refined sensor housings and improved fastener engineering. First physics output is expected in the late 2020s as the cluster count grows.
Complementary fields of view
The four sites map out the high-energy neutrino sky in complementary ways.
For upward-going events — the cleanest astrophysical signal, with atmospheric muon backgrounds filtered by the Earth — each detector sees the hemisphere opposite to itself best:
- IceCube/Gen2 at the South Pole sees the Northern Hemisphere sky.
- KM3NeT (latitude ~36-43°N), Baikal-GVD (~52°N), and P-ONE (~48°N) all see the Southern Hemisphere sky, including the Galactic Center.
The three Northern Hemisphere sites observe overlapping but distinct portions of the southern sky due to their different longitudes and Earth rotation. Combined, they provide a near-continuous coverage of the Galactic Center as the Earth rotates, with multiple detectors observing the same source on overlapping cycles.
For downgoing events at the highest energies — where atmospheric muon backgrounds dominate but cannot be perfectly rejected — IceCube has historically struggled, while water-based sites with finer string spacing can sometimes reach into the downgoing region with better topology cuts.
For the Galactic Plane as observed in 2023, the IceCube detection was a 4.5-sigma observation; a coincident or independent detection by KM3NeT or Baikal-GVD would confirm the result with completely different systematics. The same applies to source identification — point-source associations are stronger when multiple detectors independently locate the same direction.
What the combination enables
The four-detector configuration enables several capabilities that no single detector can match.
Multi-messenger reach is the most direct: a coincident detection of a neutrino with a gamma-ray flare, a gravitational-wave event, or an electromagnetic transient across multiple detectors verifies the association at much higher confidence than any single detector alone. The TXS 0506+056 association, the 2017 GW170817 search, and the 2023 Galactic Plane result all benefited from cross-detector constraints; future detections will benefit much more.
Cross-systematic checks are essential when claiming an extraordinary result. The 2018 ANITA anomalous events generated extensive debate partly because no other detector could check them; future high-energy claims will be cross-checked by multiple independent facilities.
Source localization improves with the array configuration. A neutrino arriving at multiple sites within an angular box can be triangulated more precisely than at any single site, particularly for the cascade-event topology where individual detectors give angular resolution of order 10°.
Flavor composition measurements benefit from independent samples at each detector. IceCube’s tau-neutrino sample, currently the only one in the field, will be supplemented and refined by KM3NeT-ARCA’s complementary sample over the next several years.
Summary
High-energy neutrino astronomy is no longer dominated by a single detector. Four major facilities — IceCube and its planned Gen2 expansion at the South Pole, KM3NeT in the Mediterranean Sea, Baikal-GVD in Lake Baikal, and P-ONE under construction off Vancouver Island — provide complementary fields of view, different transparent media (glacial ice, deep seawater, deep freshwater, deep ocean water), and independent systematic budgets. Combined, they cover the entire high-energy neutrino sky with at least one upward-going view of every source. The next generation of multi-messenger results — point-source identifications, flavor-ratio measurements, transient-event coincidences — will depend on cross-detector confirmations that no single facility could provide on its own. The high-energy neutrino sky of the coming decade will be observed in four-fold parallax.