TAMBO: Tau Air-Shower Mountain Observatory

• TAMBO is a deep-valley neutrino observatory that detects Earth-skimming tau neutrinos via upward-going air showers, targeting the 1–100 PeV (up to EeV) energy range.
• It employs a unique one-sided valley deployment where one mountain wall acts as both shield and target, enabling precise, sub-degree directional sensitivity.
• The design leverages modular detector arrays and advanced simulation frameworks to complement large-volume observatories, enhancing multimessenger astrophysics.

TAMBO, the Tau Air-Shower Mountain-Based Observatory, is a deep-valley, mountain-based neutrino observatory designed to detect Earth-skimming $\nu_\tau$ through the upward-going extensive air showers produced when an emerging $\tau$ decays in air. The concept uses one valley wall as target and shield and the opposite wall as an instrumented screen, with the Colca Valley in Peru serving as the leading site model in current studies. Across the concept paper, recent design overview, simulation proceeding, and later telescope overview, TAMBO is presented as a high-purity $\nu_\tau$ observatory in the PeV-to-EeV domain, with particular emphasis on the $\sim 1$–$100~\mathrm{PeV}$ interval, sub-degree pointing, and multimessenger follow-up of astrophysical neutrino sources (Collaboration et al., 10 Jul 2025).

1. Scientific motivation and development

Neutrino astronomy has matured since IceCube’s discovery of an extra-galactic PeV neutrino flux in 2013, yet the high-energy neutrino sky remains only partially mapped. The majority of the diffuse flux has no firmly identified sources; only three have been established with high significance to date—an active galactic nucleus, an active galaxy, and the Milky Way—and even these contribute only a small fraction of the total flux. A central difficulty is that, at sub-PeV energies where current detectors are most sensitive, atmospheric neutrinos overwhelm astrophysical signals by roughly six orders of magnitude. TAMBO is explicitly motivated by this background regime: it targets the multi-PeV–EeV domain, where the cosmic signal strongly dominates atmospheric backgrounds, especially for tau neutrinos, and where neutrino detection can be flavor-selective (Collaboration et al., 10 Jul 2025).

The underlying physics case predates the current design studies. The earlier Andean deep-valley detector concept defined TAMBO as a detector optimized for tau-neutrino astronomy in the $1$–$100~\mathrm{PeV}$ band, stressing that direct $\nu_\tau$ measurement is needed to constrain source flavor ratios, acceleration environments, and possible nonstandard neutrino phenomena. That work also framed TAMBO as a response to the fact that IceCube is excellent for $\nu_\mu$ through through-going tracks, but isolating $\nu_\tau$ has been challenging, with first $\tau$0 candidate “double-cascade” events identified only after about a decade (Romero-Wolf et al., 2020).

Subsequent proceedings refine rather than replace this motivation. A 2023 overview describes TAMBO as a deep-valley array of water Cherenkov detectors and/or plastic scintillator stations on the walls of Peru’s Colca Canyon, purpose-built for high-purity detection of astrophysical $\tau$1 in the $\tau$2–$\tau$3 range (Thompson, 2023). A 2026 telescope overview expands the stated target interval to $\tau$4–$\tau$5 and emphasizes the experiment’s role in addressing point-source identification in the presence of large trials factors and in mapping the largely unconstrained neutrino spectrum above $\tau$6 (Zhelnin et al., 28 May 2026). Read together, these papers present TAMBO as a specialized observatory occupying the parameter space between optical Cherenkov telescopes that are primarily sensitive below a PeV and radio or space-based concepts aimed at the highest energies.

2. Detection principle and deep-valley geometry

TAMBO exploits Earth-skimming $\tau$7 that traverse the Earth’s crust nearly horizontally, interact via charged current near a valley wall, and produce a $\tau$8 lepton that can emerge from rock into the atmosphere. The $\tau$9 then decays, initiating an upward-going extensive air shower detected by a modular particle array deployed on the opposite valley face. The deep-valley geometry boosts acceptance by presenting a large target of rock for $\nu_\tau$0 interactions and a wide open air volume for $\nu_\tau$1 decay and shower development, all in a single line-of-sight arrangement (Collaboration et al., 10 Jul 2025).

The relevant propagation scales are summarized by the standard relations

$\nu_\tau$2

$\nu_\tau$3

and the schematic $\nu_\tau$4 energy-loss law

$\nu_\tau$5

For constant $\nu_\tau$6 and $\nu_\tau$7, this gives

$\nu_\tau$8

The emergence probability requires a $\nu_\tau$9 to interact within an optimal chord depth, produce a $\sim 1$0 with sufficient residual energy, survive energy losses, and exit the rock before decaying; in schematic form,

$\sim 1$1

This probability rises with energy up to multi-PeV and then flattens or decreases when the boosted $\sim 1$2 lifetime pushes decays beyond the instrumented air volume (Collaboration et al., 10 Jul 2025).

Site geometry is central. A candidate TAMBO layout in Peru’s Colca Canyon uses natural baselines with valley depth $\sim 1$3 and median distance between faces $\sim 1$4. Earlier site descriptions emphasize Colca Canyon depths up to $\sim 1$5 and more than $\sim 1$6 of usable mountain faces, with a representative cross-section near $\sim 1$7, $\sim 1$8 giving a $\sim 1$9 wall-to-wall separation at half depth. These figures all serve the same geometric requirement: the shower must develop over several kilometers while remaining visible across the valley (Collaboration et al., 10 Jul 2025).

The deep-valley baseline also sets the directional performance. The design overview gives a timing-based angular estimate

$100~\mathrm{PeV}$0

while the earlier concept paper writes the near-horizontal requirement as $100~\mathrm{PeV}$1, with $100~\mathrm{PeV}$2 and $100~\mathrm{PeV}$3 sufficient for $100~\mathrm{PeV}$4 (Romero-Wolf et al., 2020). Both formulations express the same instrumental logic: kilometer-scale or multi-station baselines convert fast timing into pointing precision.

3. Array architecture and candidate implementations

TAMBO’s instrumentation has evolved from an early large water-Cherenkov concept to a more explicitly modular mixed-technology array. The 2020 design study uses $100~\mathrm{PeV}$5 cubic-meter-sized water-Cherenkov detector modules on a triangular grid with $100~\mathrm{PeV}$6 spacing, installed in lines along a mountain slope. The 2025 design overview instead presents a nominal configuration of $100~\mathrm{PeV}$7 detection units spaced by $100~\mathrm{PeV}$8 on a triangular grid, while the 2025 simulation proceeding uses a $100~\mathrm{PeV}$9-module array to compute apertures but does not specify the exact number of tanks versus scintillators, the module spacing, or the detailed orientation, altitude, or wall-to-wall baseline of the Colca site (Argüelles et al., 11 Jul 2025).

Study	Nominal array	Detector description
(Romero-Wolf et al., 2020)	22,000 tanks, 150 m triangular grid	cubic-meter-sized water-Cherenkov modules on a mountain slope
(Thompson, 2023)	22,000 cubic-meter-sized WCD modules in acceptance studies	water Cherenkov detectors and/or plastic scintillator stations
(Collaboration et al., 10 Jul 2025, Argüelles et al., 11 Jul 2025, Zhelnin et al., 28 May 2026)	5,000 units/modules; later overview gives approximately 50 km²	modular particle-detector array; one-sided valley deployment

Despite the variation in scale, the event observable is stable across the literature. Each unit records the number of incident shower particles and the time of first particle hit. In the simulation proceeding, the readout is modeled as per-module photoelectron yields with a global trigger formed by the distribution of triggered modules, and multiple trigger schemes can be tested post hoc using the same hit maps. The core discrimination variable is the characteristic “upward-going” shower footprint produced when the particle front from a $1$0-induced shower reaches the opposite wall (Argüelles et al., 11 Jul 2025).

The final detector technology is not fixed uniformly across all papers. The design overview describes TAMBO simply as an “array of particle detectors” without fixing the technology mix and notes that options compatible with the concept include segmented scintillator or water-Cherenkov stations with PMTs or SiPMs, emphasizing fast timing and dynamic range for PeV–EeV showers. The simulation proceeding is more concrete and describes an array of water Cherenkov tanks and plastic scintillators deployed in the Colca Canyon. The 2023 status overview adds that prototype TAMBO stations use segmented scintillator bars threaded with wavelength-shifting fibers coupled to SiPM readout (Collaboration et al., 10 Jul 2025).

The one-sided deployment is a defining architectural feature. One valley face is instrumented, while the opposite wall is detector-less and functions simultaneously as passive shield and interaction target. A 2026 overview summarizes this asymmetry as central to the method: the instrumented area and asymmetric “one-sided” deployment are what allow one side to act as shield and target and the other as the imaging plane for the upward-going shower (Zhelnin et al., 28 May 2026).

4. Simulation framework, reconstruction, and calibration

The collaboration’s current end-to-end simulation environment is TAMBOSim, or “T,” a Julia-based framework tailored to Earth-skimming neutrino geometries and steep, inclined detector deployments. The chain begins with initial injection of $1$1 energies at the Earth’s surface sampled from a power law and directions uniform on the unit sphere; if needed, neutrino propagation to the simulation region includes $1$2 regeneration. Vertex sampling follows a ranged-injection philosophy akin to IceCube’s LeptonInjector, accounting for energy-dependent charged-lepton travel between vertex and detector. $1$3 propagation through rock and air is handled with PROPOSAL, including ionization, bremsstrahlung, photonuclear interactions, $1$4 pair production, and LPM/Ter-Mikaelian effects; air showers from $1$5 decays in air are then simulated with CORSIKA8 and propagated to an inclined readout plane representing the opposite canyon wall (Argüelles et al., 11 Jul 2025).

Detector response is computed within TAMBOSim. For particles intersecting detector modules, photoelectron yields are computed per species and energy, and events are flagged as triggers by analyzing the distribution of triggered modules across the array. Event weighting is explicit: Monte Carlo events are assigned “oneweights” encoding the ratio of the desired physical distribution to the injection distribution. Data handling uses Arrow files for structured outputs and Parquet for large EAS particle lists, with stored quantities including event IDs, injection parameters, column depth, oneweights, $1$6 decay flags, detailed energy-loss bookkeeping, and per-module particle-arrival times and multiplicities (Argüelles et al., 11 Jul 2025).

Reconstruction is correspondingly timing-dominated. The design overview describes shower-axis fits from timing planes, impact-point and direction reconstruction, and lateral-distribution-based energy proxies. Figure 1 in that paper illustrates an $1$7 event with colored time stamps and hit multiplicities used to infer shower direction and energy proxy. The simulation proceeding likewise emphasizes shower direction and core reconstruction on steep walls through per-module first-arrival times and particle counts, but also states explicitly that angular, energy, core position, and timing resolutions are not yet reported there and will depend on module layout, calibration, and trigger configurations (Collaboration et al., 10 Jul 2025).

Calibration strategy follows directly from the geometry. Abundant down-going cosmic-ray showers can be used for timing and geometry cross-calibration; light sources or portable calibration modules can align timing and gain across slopes; atmospheric monitoring with LIDAR or meteorological data becomes relevant if optical modalities are included; and cross-checks against external neutrino alerts from IceCube, KM3NeT, or Baikal-GVD can anchor joint reconstructions. A later proceeding adds a prototype reconstruction example in which the reconstructed and true directions differ by about $1$8 in a well-reconstructed $1$9-induced air-shower event, indicating the maturity of the direction-fitting chain without yet constituting a full detector-wide resolution budget (Zhelnin et al., 28 May 2026).

5. Acceptance, event rates, and physics reach

TAMBO’s performance is usually expressed in terms of acceptance or aperture rather than bare effective area, because the field of view is restricted by local topography and Earth-skimming geometry. The 2025 simulation proceeding finds that the aperture of the $100~\mathrm{PeV}$0-module idealized valley is slightly higher across most energies than the Colca-Valley geometry, attributed to the larger average wall-to-wall distance in the idealized case, allowing the shower to develop closer to maximum before reaching the readout wall. In that treatment, the aperture follows approximately an $100~\mathrm{PeV}$1 scaling up to energies where the $100~\mathrm{PeV}$2 decay length becomes comparable to or larger than the valley width, at which point the scaling weakens (Argüelles et al., 11 Jul 2025).

A consistent feature across papers is the comparison with IceCube. One design overview states that above $100~\mathrm{PeV}$3, TAMBO’s $100~\mathrm{PeV}$4 acceptance exceeds IceCube’s $100~\mathrm{PeV}$5 acceptance, while the simulation proceeding places the crossing around $100~\mathrm{PeV}$6 and notes that IceCube’s $100~\mathrm{PeV}$7 aperture becomes negligible above $100~\mathrm{PeV}$8. Both texts also identify special sensitivity near the Glashow resonance at $100~\mathrm{PeV}$9, where $\nu_\tau$0 with subsequent $\nu_\tau$1 enhances the $\nu_\tau$2-shower yield (Collaboration et al., 10 Jul 2025).

The forecast event rate depends on the assumed flux model and detector realization. Using IceCube’s measured diffuse flux, the 2025 design overview forecasts $\nu_\tau$3 extragalactic neutrinos per decade in the nominal $\nu_\tau$4-unit array, rising to $\nu_\tau$5 per decade in optimistic flux scenarios, and characterizes these counts as “cosmic-pure.” A later telescope overview states that the full array is expected to observe approximately $\nu_\tau$6–$\nu_\tau$7 astrophysical neutrino events per year. The earlier $\nu_\tau$8-tank concept, under an IceCube-like extrapolated diffuse flux and democratic flavor ratio at Earth, projected $\nu_\tau$9 $\nu_\mu$0 events in $\nu_\mu$1 years, peaking near $\nu_\mu$2 (Collaboration et al., 10 Jul 2025).

The scientific program extends well beyond counting events. The design overview highlights source discovery, diffuse cosmogenic neutrinos expected above $\nu_\mu$3, tests of neutrino-nucleon cross-sections at supra-PeV energies, precision constraints on astrophysical flavor composition when combined with IceCube, and new probes of air-shower hadronic physics such as the “muon puzzle.” The 2026 telescope overview adds a diffuse-flux mapping role in the $\nu_\mu$4–$\nu_\mu$5 gap and emphasizes that TAMBO’s sensitivity can test diffuse-flux interpretations of the KM3NeT ultra-high-energy event (Zhelnin et al., 28 May 2026).

A recurring misconception is that in-principle sensitivity to the Glashow resonance automatically implies a strong discovery channel. A dedicated study of Glashow-resonance discovery in a TAMBO-like air-shower telescope argues otherwise. For a TAMBO-scale array with geometric area around $\nu_\mu$6, the expected significance is only around $\nu_\mu$7 in $\nu_\mu$8 years for IceCube-like meson-dominated fluxes, although the probability improves to $\nu_\mu$9 if PeV neutrinos mainly originate from neutron decay. The paper attributes the limitation to the suppressed branching ratio $\nu_\tau$0, the fact that the daughter neutrino takes away $\nu_\tau$1 of the energy from the $\nu_\tau$2 decay, and the strong attenuation of Earth-skimming neutrinos at the resonance energy (Huang, 2023). TAMBO therefore has genuine Glashow sensitivity, but that channel should not be interpreted as an automatically dominant discovery mode.

6. Comparisons, synergies, project status, and open challenges

Within the broader landscape of $\nu_\tau$3-neutrino observatories, TAMBO’s distinctive feature is the compact deep-valley geometry. Relative to Trinity, GRAND, POEMMA, and Ashra/NTA, the design overview describes the TAMBO approach as providing a compact, terrain-enabled acceptance at multi-PeV energies, high cosmic purity and sub-degree pointing optimized for point-source discovery and Glashow-resonance sensitivity, and lower cost and easier deployment because the array modules avoid drilling or space-based optics. The same overview estimates a total cost $\nu_\tau$4 of IceCube and emphasizes that staged deployment is feasible because performance scales with the number of installed units (Collaboration et al., 10 Jul 2025).

The complementarity with large-volume Cherenkov detectors is central rather than incidental. IceCube, KM3NeT, and Baikal-GVD have huge volumes and broad sky coverage but limited $\nu_\tau$5 purity and rapidly declining exposure above a few PeV. TAMBO, by contrast, views a restricted horizon band but produces a high-purity sample with sub-degree pointing. This makes it useful as a “viewfinder”: source candidates identified by TAMBO can seed targeted archival or real-time searches in larger-volume detectors, dramatically reducing trials penalties in all-sky analyses. The same logic extends to multimessenger follow-up with $\nu_\tau$6-ray and gravitational-wave facilities (Collaboration et al., 10 Jul 2025).

TAMBO also belongs to a wider family of mountain-based Earth-skimming observatories. Radio, optical, and fluorescence systems such as TAROGE-M, MAGIC, Trinity, and Ashra have explored related geometries, and TAROGE-M in particular demonstrates that a compact horizon-facing mountain array can achieve high duty cycle and reconstruct source directions with $\nu_\tau$7 angular resolution in a mountain-based tau-air-shower context. This suggests that TAMBO’s core geometry is not idiosyncratic but part of a broader experimental class, even though the TAMBO baseline emphasizes particle detectors and precision timing rather than radio or optical imaging (Collaboration et al., 2022).

Project status is defined by design study, simulation refinement, and prototyping rather than deployment. The collaboration has carried out detailed design studies and valley-based simulations with realistic geometry, is actively surveying candidate sites with strong emphasis on social and environmental responsibility, and regards the Colca Valley as the leading option. Prototype scintillator stations are under construction to validate detector performance, DAQ, synchronization, and communications in mountainous conditions. At the same time, several proceedings state explicitly that some engineering parameters remain unsettled, including the final technology mix, trigger definitions, detailed layout optimization, and a quantified systematic-error budget for the full instrument (Argüelles et al., 11 Jul 2025).

The dominant open challenges are well defined. They include $\nu_\tau$8–$\nu_\tau$9 cross-sections and PDFs at PeV–EeV energies, rock density and column-depth modeling across the site, $\tau$00 energy-loss modeling and stochasticity in PROPOSAL, treatment of $\tau$01 polarization effects, hadronic interaction models in CORSIKA8, detector calibration, optical properties and photoelectron-yield modeling where relevant, atmospheric conditions, and the practical problem of reconstructing upward-going showers on steep, inclined walls. These are not peripheral technicalities: they enter directly into $\tau$02, $\tau$03, $\tau$04, trigger efficiency, and therefore the experiment’s aperture and physics interpretation (Argüelles et al., 11 Jul 2025).

Taken together, the published literature presents TAMBO as a specialized neutrino telescope built around a specific geometrical insight: a deep valley can simultaneously supply interaction target mass, passive shielding, air-shower development volume, and a long timing baseline. The result is a concept aimed not at maximal sky coverage, but at a high-purity $\tau$05 sample with strong pointing, immediate multimessenger utility, and sensitivity to astrophysics and fundamental physics in the PeV-to-EeV frontier.