GRAND: Giant Radio Array for Neutrino Detection

• GRAND is a multi-messenger observatory that detects ultra-high-energy neutrinos, cosmic rays, and gamma rays via radio signals from extensive air showers.
• It deploys approximately 200,000 self-triggered antennas over remote, mountainous areas using the earth-skimming tau neutrino technique for robust, large-scale observation.
• The project aims to unravel the origins of UHE cosmic rays and improve precision in energy and directional reconstruction, advancing multi-messenger astrophysics.

The Giant Radio Array for Neutrino Detection (GRAND) is a next-generation, multi-messenger astroparticle observatory under development, designed to detect ultra-high-energy (UHE) neutrinos, cosmic rays, and gamma rays through the measurement of radio signals from extensive air showers (EAS). The project’s primary scientific ambition is to discover the sources and elucidate the propagation mechanisms of UHE cosmic rays—particles reaching energies in excess of 10^17 eV—using a massive, sparsely-instrumented array of self-triggered radio antennas deployed over remote, mountainous landscapes (Martineau-Huynh et al., 2015, Martineau-Huynh et al., 2017, Collaboration et al., 25 Sep 2025). GRAND leverages the earth-skimming tau neutrino channel and advanced radio detection techniques, aiming to achieve unprecedented neutrino sensitivities and precision in reconstructing the properties of cosmic ray–induced air showers.

1. Scientific Objectives and Rationale

The central motivation for GRAND is to resolve the origin of UHE cosmic rays and test the most pessimistic models for cosmogenic neutrinos—those produced by the interactions of UHECRs with the cosmic microwave background. By targeting the detection of neutrinos above $\sim10^{17}$ eV, GRAND aims to pioneer the field of neutrino astronomy at extreme energies and to enable the identification of both diffuse fluxes and point sources. Achieving sensitivities on the order of $3 \times 10^{-11} E^{-2}$ GeV$^{-1}$cm$^{-2}$s$^{-1}$sr$^{-1}$ above $3 \times 10^{16}$ eV or all-flavor limits of $\sim10^{-10}$ GeV cm$^{-2}$ s$^{-1}$ sr$^{-1}$ above $5\times10^{17}$ eV places GRAND among the leading instruments in the field (Martineau-Huynh et al., 2015, Kotera, 2021, Lago, 2021, Neto, 2023, Kotera, 2024). Fundamental questions addressed by GRAND include:

• What are the dominant astrophysical sources and acceleration mechanisms of UHECRs?
• Can the known flux of cosmogenic neutrinos be confirmed or ruled out at the lowest predicted levels?
• Are there anisotropies or point sources of UHE neutrinos that can be directly associated with astrophysical phenomena?
• How do the energy spectrum, mass composition, and angular distribution of UHECRs evolve across the transition from galactic to extragalactic dominance?

2. Detection Principle and Experimental Design

GRAND’s detection strategy is predicated on the radio detection of air showers initiated by the decay of tau leptons produced in the charged-current interactions of tau neutrinos within the Earth’s crust (the "earth-skimming" channel). The tau lepton emerges from the rock or mountain, decays in the atmosphere, and initiates an EAS whose charged component emits a prompt, coherent radio pulse—primarily through the geomagnetic effect (and, to a lesser extent, the Askaryan effect)—in the 50–200 MHz frequency band (Martineau-Huynh et al., 2017, Neto, 2023, Guelfand, 3 Jan 2025, Collaboration et al., 25 Sep 2025).

The GRAND array is designed to:

• Deploy approximately 200,000 self-triggered radio antennas across ~200,000 km², subdivided into ~20 subarrays, each spanning ~10,000 km², at radio-quiet, mountainous sites worldwide (Martineau-Huynh et al., 2017, Lago, 2021, Kotera, 2024).
• Use antennas (“HorizonAntenna” design) elevated at 3.5–5 m with three orthogonal arms for full polarization measurement and sensitivity to near-horizontal events (Guelfand, 3 Jan 2025, Neto, 2023).
• Implement modular, solar-powered detection units with onboard digitization (typically 500 MSPS, 14-bit ADCs), local FPGAs for real-time processing, and WiFi-based communications, permitting autonomous operation (Ma et al., 2023, Guelfand, 3 Jan 2025, Collaboration et al., 25 Sep 2025).
• Target a detection threshold adapted for large radio footprints—especially for very inclined showers (zenith > 70°)—with antenna spacings up to ~1 km to cover vast areas at manageable cost (Martineau-Huynh et al., 2017, Guelfand, 3 Jan 2025).

The detection process is governed by the relation

$$V_{\text{oc}} = \mathbf{L}_\mathrm{eff} \cdot \mathbf{E}$$

where $V_{\text{oc}}$ is the open-circuit voltage at the antenna terminals, $\mathbf{L}_\mathrm{eff}$ is the vector effective length (frequency, zenith, and azimuth dependent), and $\mathbf{E}$ is the incident electric field (Collaboration et al., 25 Sep 2025). This relationship is central to the precise calibration pipeline (as implemented in GRANDlib) and to reconstructing the energy and geometry of detected air showers.

3. Prototyping and Current Status

The path to full deployment is structured through successive prototype phases, each addressing specific design and operational challenges (Guelfand, 3 Jan 2025, Neto, 2023, Collaboration et al., 25 Sep 2025, Martineau-Huynh, 9 Jul 2025). The main testbeds (as of 2025) are:

Prototype	Location	Purpose
GRANDProto300	XiaoDuShan, China	Sparse array (∼200 units in 2024) for efficient, autonomous detection of inclined EAS and validation of background rejection, energy/arrival direction reconstruction, and large-area communications (Ma et al., 2023, Guelfand, 3 Jan 2025).
GRAND@Auger	Malargüe, Argentina (Pierre Auger Observatory)	Repurposed AERA stations; cross-calibration, coincident detection with established particle and radio detectors (Errico et al., 10 Jul 2025, Collaboration et al., 25 Sep 2025).
GRAND@Nançay	Nançay, France	Small-scale hardware and trigger testbench for European groups (Guelfand, 3 Jan 2025, Collaboration et al., 25 Sep 2025).

Early operation has confirmed the successful autonomous detection of EAS and the feasibility of precision calibration using Galactic background noise, with absolute gain calibration at the 10%–50% level for the horizontal and vertical channels, respectively (Errico et al., 10 Jul 2025). Coherent radio signals consistent with cosmic-ray air showers have been detected in both self-triggered and externally triggered modes, and coincident events with the Pierre Auger surface detectors have been reported (Errico et al., 10 Jul 2025, Collaboration et al., 25 Sep 2025).

4. Triggering, Background Rejection, and Data Acquisition

GRAND’s radio detection is challenged by the necessity of efficiently suppressing backgrounds—predominately anthropogenic and atmospheric radio-frequency interference (RFI)—while maintaining high efficiency for genuine air-shower signals:

• Triggering is performed in multiple stages. The first-level trigger (FLT) employs a template-matching approach using simulated air-shower pulse templates convolved with the expected RF response; for SNR > 5, FLT achieves 90% efficiency and >75% background rejection (Correa et al., 6 Jul 2025). Cross-correlation is computed as:

$$\rho_{ij}(\tau) = \int T_{ij}(t) \cdot V_i(t+\tau) dt$$

where $T_{ij}(t)$ is the template and $V_i(t+\tau)$ is the digitized voltage.

• The second-level trigger (SLT) integrates timing and amplitude from multiple antennas; array-level criteria (timing coincidence, amplitude and conic footprint matching, polarization alignment with the geomagnetic v × B direction) increase the robustness of shower candidate selection and suppress accidental triggers from isolated noise bursts (Correa et al., 6 Jul 2025).
• Data from triggered events are streamed via WiFi mesh networks with per-unit rates up to ∼465 kB/s, permitting real-time event building and centralized reconstruction up to the array level (Ma et al., 2023, Guelfand, 3 Jan 2025).

Efforts continue to further reduce false triggers and further increase the purity of candidate lists by integrating machine learning–driven triggers, advanced RFI rejection, and continuous feedback from prototype deployments (Decoene, 2019, Correa et al., 6 Jul 2025).

5. Event Reconstruction and Simulation

Accurate reconstruction of shower geometry, energy, and mass composition is essential for GRAND’s UHECR and neutrino science case:

• Arrival direction and energy are reconstructed using advanced algorithms including planar and spherical wavefront fits, polarization vector analysis, and fits to the Lateral Distribution Function (LDF) and Angular Distribution Function (ADF) (Neto, 2023, Martineau-Huynh, 9 Jul 2025, Ferrière et al., 10 Jul 2025).
• Recent studies employ physics-aware graph neural networks (GNNs) applied to the spatial and time-amplitude pattern of triggered antennas. These models achieve angular resolutions below 0.2°, energy resolutions of 15%–16%, and provide well-calibrated uncertainty estimates:

$$\mathcal{L}_{\mathrm{nll}} = \frac{1}{2} \ln \hat{\sigma}^2 + \frac{(y-\hat{y})^2}{2\hat{\sigma}^2}$$

where $\mathcal{L}_{\mathrm{nll}}$ is the negative log-likelihood loss, $y$ is the true property, $\hat{y}$ the prediction, and $\hat{\sigma}^2$ the model variance (Ferrière et al., 10 Jul 2025).

• Prototypes demonstrate cross-consistency between radio-based and established particle detector reconstructions in arrival direction and energy, validating the radio detection chain and establishing procedures for domain adaptation as deployment scales (Errico et al., 10 Jul 2025, Collaboration et al., 25 Sep 2025).

End-to-end simulation frameworks (notably GRANDlib) incorporate antenna frequency response, atmospheric and topographical effects, and air-shower radio emission physics, underpinning both the instrumental design and the analysis strategies (Collaboration et al., 25 Sep 2025).

6. Scientific Prospects and Impact

GRAND is poised to transform the field of UHE neutrino and cosmic-ray astronomy due to its combination of scale, sensitivity, and timing/angular precision. The ultimate science program encompasses:

• Detection and study of cosmogenic neutrinos, probing all reasonable theoretical flux models and potentially discovering point sources or anisotropies at flux levels unattainable by current optical/particle neutrino telescopes (Martineau-Huynh et al., 2015, Martineau-Huynh et al., 2017, Neto, 2023, Kotera, 2021).
• Precision measurements of the UHECR energy spectrum, composition (mean logarithmic mass ⟨ln A⟩, as inferred from $X_{\mathrm{max}}$), and anisotropies down to dipole amplitudes of $\sim5\times10^{-3}$, particularly in the $10^{16.5}$–$10^{18}$ eV transition regime (Zhang et al., 19 Jun 2025, Martineau-Huynh, 9 Jul 2025).
• Multi-messenger studies combining neutrinos, cosmic rays, and gamma rays, including follow-up of transient phenomena (gamma-ray bursts, tidal disruption events, fast radio bursts) with full-sky coverage using subarrays distributed across both hemispheres (Neto, 2023, Kotera, 2024).
• Potential contributions to radio astronomy (Epoch of Reionization 21-cm line studies, Galactic structure via radio mapping) and probing new physics at energies far above the reach of terrestrial accelerators—such as tests of Lorentz invariance, searches for signatures of exotic interactions, or new propagation effects (Lago, 2021, Guelfand, 3 Jan 2025).

7. Future Roadmap, Challenges, and Outlook

The staged GRAND deployment plan requires phased expansion:

• Ongoing operation of prototypes (GP300, GRAND@Auger, GRAND@Nançay) to validate detection technology, refine DAQ and trigger systems, and test robustness under varied environmental and RFI conditions (Collaboration et al., 25 Sep 2025, Ma et al., 2023, Martineau-Huynh, 9 Jul 2025).
• GRADUAL SCALING to GRAND10k—two arrays of ~10,000 antennas each in both hemispheres—planned for deployment from 2030, and ultimately up to the target 200,000 antenna array for full-sky neutrino and cosmic-ray observation (Kotera, 2024, Martineau-Huynh, 9 Jul 2025).
• Key technical challenges include optimizing per-unit cost, power/communications in remote areas, sustainable maintenance logistics for arrays spanning ~200,000 km², and maintaining precision timing (target <5 ns) across wide baselines (Collaboration et al., 25 Sep 2025, Neto, 2023, Kotera, 2024).
• Ongoing R&D focuses on integrating deep learning for signal discrimination, further automating calibration via the Galactic background, and extending hardware lifetimes in harsh climates (Neto, 2023, Guelfand, 3 Jan 2025, Decoene, 2019).

Successful realization of GRAND will yield a sensitive, robust platform for UHE neutrino and cosmic-ray discovery, enable high-precision multi-messenger astrophysics, and address longstanding questions on the highest-energy phenomena in the Universe (Kotera, 2024, Martineau-Huynh, 9 Jul 2025, Neto, 2023, Martineau-Huynh et al., 2015).