Cosmic-Ray Neutron Sensing (CRNS)

• Cosmic-Ray Neutron Sensing (CRNS) is a non-invasive technique that estimates soil moisture by detecting variations in epithermal neutron flux due to environmental hydrogen.
• It employs calibrated detectors such as ³He, BF₃, and Cherenkov systems to convert neutron counts into quantitative soil moisture measurements.
• CRNS facilitates mobile surveys and incorporates correction algorithms to enable reliable, large-scale mapping for hydrology, precision agriculture, and environmental monitoring.

Cosmic-Ray Neutron Sensing (CRNS) is a non-invasive, landscape-scale soil moisture estimation technique that exploits the energy-dependent moderation of cosmic-ray secondary neutrons by hydrogen nuclei in the environment. It measures the flux of epithermal and fast neutrons above the ground, with the neutron count rate inversely correlated to the volumetric water content in the soil. CRNS has been widely adopted for hydrological observation, precision agriculture, and environmental monitoring due to its unique integration volume and spatial footprint, bridging the scale gap between point sensors and remote sensing platforms (Stowell et al., 2021, Köhli et al., 2018).

1. Physical Principles of CRNS

Primary galactic cosmic rays, predominantly protons, interact with atmospheric nuclei, generating extensive hadronic cascades that produce secondary (spallation) neutrons with energies up to tens of MeV. These neutrons undergo a complex history of scatterings in air and soil, with the probability of moderation—energy loss due to elastic collisions—governed principally by the local abundance of hydrogen-containing substances, most notably soil moisture. Moderation proceeds most efficiently via neutron-proton collisions, so regions with higher water content induce faster thermalization and attenuation of neutron flux.

Epithermal neutrons in the energy window ∼0.5 eV–10 keV are especially sensitive to variations in environmental hydrogen. The population of such neutrons leaking from the soil surface and detected above ground forms the basis for inferring soil moisture. Empirically, the counting rate of epithermal neutrons $N$ is anti-correlated with the volumetric soil water content $\theta$ within a surface footprint of several tens to hundreds of meters (Stowell et al., 2021, Köhli et al., 2018, Schrön et al., 2017).

2. Detector Technologies and Response Functions

Stationary and mobile CRNS systems rely on moderated proportional counters—typically ^3He or ^10BF₃ tubes embedded in hydrogen-rich polyethylene moderators—or, more recently, alternative designs including liquid scintillators and Cherenkov detectors. The system response is modeled by a detector response function $R(E, \theta)$, which quantifies the detection probability of an ambient neutron with energy $E$ and incidence angle $\theta$:

$R(E,\theta) = \epsilon(E) \cdot g(\theta)$

where $\epsilon(E)$ is the intrinsic detection efficiency (fraction captured in the converter gas) and $g(\theta)$ is the angular term normalized to $g(0)=1$, with monotonic decay for increasing $\theta$ and $\theta$0 (Köhli et al., 2018).

Key features of typical CRNS detector response:

Energy E (eV)	R(E)/R_max (^3He)	R(E)/R_max (BF₃)
0.1	0.22	0.20
1.0	1.00	0.88
10	0.96	0.83
10⁵	0.30	0.25

All designs show an epithermal response peak between 1–10 eV and a broad plateau up to 10 keV, with sensitivity rolling off for higher energies as moderators become transparent to fast neutrons. The response functions closely track Bonner-Sphere standards when normalized for moderator thickness (Köhli et al., 2018).

Alternative detection architectures, such as NaCl-doped Water Cherenkov Detectors (WCDs), exploit neutron-capture-induced γ-ray emission in large water volumes and can be tuned for high-resolution, continuous field deployment (Sarmiento-Cano et al., 24 Jan 2026).

3. Soil Moisture Retrieval and Calibration Relations

The neutron count rate $\theta$1 is converted to volumetric soil moisture $\theta$2 using empirically calibrated relationships. A widely adopted semi-empirical form is:

$\theta$3

where $\theta$4 is the reference count rate for a known moisture $\theta$5, estimated during site calibration; $\theta$6 and $\theta$7 are site- and instrument-specific constants incorporating detector response and environmental corrections (Köhli et al., 2018, Stowell et al., 2021, Sarmiento-Cano et al., 24 Jan 2026). For rigorous applications, the full integration over the neutron flux, energy- and angle-dependent detector response, and atmospheric/soil conditions is implemented:

$\theta$8

Additional correction factors account for cosmic-ray intensity, barometric pressure, atmospheric humidity, and, for mobile surveys, road and infrastructure effects (Schrön et al., 2017).

Specialized calibration routines are developed for WCD-based CRNS, involving nonlinear regression over controlled soil moisture scenarios to extract the calibration constants for the specific detector-soil system (Sarmiento-Cano et al., 24 Jan 2026).

4. Mobile CRNS Surveys and Bias Corrections

Mobile CRNS, often deployed as "Cosmic-Ray Neutron Rovers," enables landscape-scale mapping of field soil moisture by traversing large areas on vehicles. The method integrates measured counts over transit intervals using GPS-synchronized sampling, yielding spatially continuous moisture estimates with a typical footprint radius of 100–200 m (Schrön et al., 2017).

Local structures such as roads can introduce significant bias. Monte Carlo neutron transport simulations quantify that overestimation of dryness by up to 40% can occur on impermeable roads. Analytical and empirical correction functions are available, parameterized by road width $\theta$9, road moisture $R(E, \theta)$0, field moisture $R(E, \theta)$1, and lateral offset $R(E, \theta)$2:

$R(E, \theta)$3

where $R(E, \theta)$4 is a composite of fitted functions $R(E, \theta)$5, $R(E, \theta)$6, and $R(E, \theta)$7, with tabulated parameters for operational use. Empirically, contamination from road bias is negligible beyond several meters from the feature, and correction algorithms restore unbiased agreement with field point measurements (Schrön et al., 2017).

5. Experimental Validation and Environmental Sensitivity

Comparative field campaigns with combinations of stationary and mobile neutron detectors, including ^3He proportional counters, moderated/unmoderated arrays, and pulse-shape-discriminating scintillators, confirm consistency between modeled and observed neutron fluxes across environments and elevations, with discrepancies attributed primarily to local variations in ground moisture, elemental composition, and accidental moderation (e.g., snow/ice accumulation) (Woolf et al., 2020).

Observed count-rate reductions of up to ~25% are recorded in the presence of snow or increased soil moisture, and confirmation of pressure (altitude) and, to a lesser extent, geomagnetic latitude dependencies underscores the importance of environmental corrections for quantitative soil moisture retrieval (Woolf et al., 2020).

6. Detector Optimization and Monte Carlo Modelling

Monte Carlo codes such as Geant4 and URANOS are standard for simulating neutron transport and optimizing detector response. Detailed CAD-derived voxel models and modern nuclear cross-section libraries (e.g., ENDF/B-VII.1, JENDL/HE-2007) are used to design and benchmark detector geometries, moderator thickness, and alternative converter media. Large-scale simulations (10⁶+ neutron histories) yield statistical uncertainties well below 1%, with systematic uncertainties dominated by moderator density and deviations in low-energy scattering kernels (Stowell et al., 2021, Köhli et al., 2018).

Cost-effective alternatives such as scintillator- or Cherenkov-based detectors are being explored using similar physics-informed optimization protocols, balancing sensitivity, footprint, and deployment cost for agricultural scale applications (Stowell et al., 2021, Sarmiento-Cano et al., 24 Jan 2026).

7. Current Challenges and Future Perspectives

Primary limitations for widespread CRNS adoption include detector cost (especially for ^3He and large-PMT or stainless-steel detectors), the need for accurate local calibration (especially in heterogeneous or structured landscapes), and the development of robust, real-time correction algorithms for non-uniform environments or networked sensor deployments. Future research is focusing on reducing detector costs (e.g., SiPM arrays, plastic moderators), multi-parameter inversion (retrieving both $R(E, \theta)$8 and soil composition), and integration with precision agriculture systems for closed-loop irrigation control (Sarmiento-Cano et al., 24 Jan 2026).

Algorithmic refinement, especially Bayesian deconvolution for environmental covariates, and large-scale field deployment studies are pivotal for achieving reliable, high-resolution soil moisture mapping at operational scale. Continuous validation against independent ground-truth data, incorporation of local atmospheric and hydrological variability, and ongoing improvement of neutron transport and detector models will define the next phase of CRNS development (Köhli et al., 2018, Stowell et al., 2021, Schrön et al., 2017).