Before IceCube: learning to build telescopes in water and ice

Nature supplies the target; the instrument has to be built

A neutrino rarely interacts, so catching enough high-energy events takes an enormous target. Instead of manufacturing and containing all that material, researchers could place light sensors in deep water or ice and let nature supply much of the volume.

The volume still had to be turned into an instrument. A neutrino interaction can produce a charged particle, such as a muon, whose Cherenkov light reaches several sensors. From the sensors’ positions and the light’s arrival times, the track direction can be reconstructed. The apparatus is as much a network of clocks and optical measurements as a collection of photomultipliers.

Moisei Markov’s early proposal to use deep natural water helped set this direction. Deep water alone did not make an observatory, though: the sensors had to survive, their clocks had to agree, and the way light travelled and was absorbed had to be measured.

Sources: [1] · [2]

DUMAND took the problem into the Pacific

DUMAND, the Deep Underwater Muon and Neutrino Detector project, pursued an array in the deep ocean off Hawaii. The ocean made an enormous detector volume conceivable. It also forced the collaboration to solve problems of pressure, deployment, power, communication and knowing where each sensor was.

Prototype work demonstrated important technologies, but the proposed large array was not completed. US support ended in 1995 after a difficult development history. The concept was promising and the engineering results were useful, but they had not yet become the planned neutrino observatory.

The work still mattered. It defined practical problems and built expertise that the wider field kept using. What DUMAND left was an engineering legacy; it did not detect a population of astrophysical neutrinos.

Sources: [1] · [3]

Lake Baikal offered another way to build and maintain an array

At Lake Baikal, researchers lowered sensors into deep fresh water. In winter the frozen surface served as a platform for installation and servicing. The Russian-led effort, with contributions including those from DESY in Germany, opened an alternative route to an underwater telescope.

The NT-96 stage ran in 1996 with 96 optical modules on four strings. Its analyses reconstructed upward-going muon candidates consistent with atmospheric-neutrino interactions. The larger NT200 configuration was completed in 1998.

Upward-going tracks mattered because ordinary cosmic-ray muons come from above and cannot simply cross the whole Earth to reach the array from below. A badly reconstructed downward event could still look upward. A neutrino sample therefore needed good timing and background rejection, not just an upward arrow on an event display.

Sources: [4] · [1]

AMANDA had to learn how light moves through the ice

AMANDA placed optical modules in the South Pole ice. Its early, relatively shallow deployments ran into air bubbles trapped in the ice, which scattered light strongly. The light could still arrive but lose the timing information needed to reconstruct a track.

Deeper deployments found ice suitable for reconstruction, although dust layers and scattering still had to be calibrated. With ten strings taking data in 1997, AMANDA collected an atmospheric-neutrino sample that agreed with expectations, which showed it was a working telescope. The detailed result was published later.

Ice solved some mechanical problems of water arrays: sensors frozen in place did not drift with currents. It created others. Access after deployment was limited, and light propagation changed with depth. The medium itself became part of the measurement.

Sources: [5] · [6] · [1]

A working telescope was a milestone, not yet a discovery

Atmospheric neutrinos were a useful test population. They showed that an array could pick out rare interactions and reconstruct them plausibly. Once the goal became finding neutrinos from distant cosmic accelerators, the same population turned into background.

IceCube’s larger scale built on that experience. Its evidence for a high-energy astrophysical population required event selection and a quantitative comparison with the atmospheric background. Reconstructing a neutrino event and showing that it came from a cosmic source remained separate tasks.

This history spans many projects, institutions and environments. In the reconstruction Lab, changing the assumed optical properties gives a small view of the same problem: sensor times mean little without a model of how the light reached them. The teaching array is simplified, but the need to connect hardware, calibration and inference is real.

Sources: [5] · [7] · [8] · [1]

Try it in the Lab

Primary sources & revision

1. Christian Spiering · Towards High-Energy Neutrino Astronomy: A Historical Review (2012)
2. Kamioka Observatory · Super-Kamiokande detector
3. Arthur Roberts · A personal history of the DUMAND project (1992)
4. Baikal Collaboration · Reconstruction of Atmospheric Neutrinos with NT-96 (1997)
5. AMANDA Collaboration · Observation of High Energy Atmospheric Neutrinos (2002)
6. AMANDA Collaboration · Status of the AMANDA Project (IceCube proposal)
7. IceCube Collaboration · Instrumentation and Online Systems (2017; arXiv v3)
8. IceCube Collaboration · Evidence for High-Energy Extraterrestrial Neutrinos (2013)

First published 2026-10-09; last revised 2026-10-09. Original explanatory text and diagrams by Sandbox Physics. Illustrations are schematic; no experimental event records are reproduced here. This is an independent educational publication, not an official Nobel or experiment collaboration publication.