Kelso Dunes in California’s Mojave National Preserve produce a low sound that can be heard and felt when sand avalanches down a slope. The natural source is observable. The scientific disagreement concerns how moving grains and the dune transmit and sustain a note. Original field research and a published exchange show why identifying the source of a mystery does not always settle every part of its mechanism.
A dune field made by water, wind and mountains
The National Park Service traces the sand supply to sediments carried by the Mojave River from the San Bernardino Mountains into former lakes. As those deposits became exposed in a drier landscape, wind redistributed them. The Granite and Providence Mountains influence local airflow, helping sand accumulate repeatedly in the same broad area. This is a landscape with a geological history, rather than an isolated mound that appeared without an identifiable source.
The park’s account places most accumulation at roughly 2,400–9,000 years ago. It also distinguishes moving individual crests from the stability of the field as a whole: seasonal winds can alter local shapes without shifting the entire dune field continuously across the desert. These observations are relevant context for the sound because a dune is an arrangement of grains and layers, not a uniform solid object.
NPS describes Kelso as one of the North American dune fields known for booming, with a deep rumble and bodily vibration associated with avalanching sand. Its separate place account, updated in April 2024, also records the low-pitched sound. Neither description promises that a visitor will hear it on any particular day. The historical phenomenon is the subject of this case; a current park description is not being treated as a newly dated incident.
What the 2007 investigation measured
Nathalie Vriend and colleagues’ 2007 paper includes Kelso among four field locations visited over five summers. Their detailed microphone, geophone and seismic-refraction examples chiefly concern Dumont Dunes. Human-triggered avalanches supplied sound; hammer impulses were used for the separate seismic survey. Keeping those places and procedures distinct prevents a Dumont measurement from becoming an invented Kelso observation.
The team proposed that a shallow dry-sand layer acted as a waveguide, with its structure selecting the sustained note. A winter return and comparisons between dunes tested conditions for booming. They challenged a simple relationship between average grain diameter and frequency. The paper’s publication year dates the report, not every field visit. Exact Kelso visit dates are not assigned here.
A published objection to the waveguide explanation
Bruno Andreotti, Laurent Bonneau and Éric Clément challenged the analysis in a 2008 comment. They argued that waves in the near-surface granular material were dispersive surface modes, and that interpreting surface seismograph arrivals as ordinary bulk propagation could produce an incorrect picture of layering and resonance. They also challenged the claimed agreement between calculated resonance and measured booming frequency.
Their alternative linked the emitted note to grain collisions and shear during a steady avalanche, with acoustic waves helping synchronise the motion. A major qualification was how an avalanche was driven: a pulsed or uneven human push, they argued, was not equivalent to a steady natural slide. The comparison data included work in Morocco. Those experiments were part of the wider physical debate, not observations made at Kelso.
The reply and the importance of definitions
Vriend and colleagues replied that curved ray paths did not rule out their resonance model. They separated hammer-generated survey signals from avalanche-generated booming and maintained that the sustained frequency was independent of the source once booming began. They also defended analysing the full distribution of measurements rather than selecting or averaging points to obtain a grain-size relationship.
The reply distinguishes short bursts from sustained booming and notes uncertainty in calculated frequencies, particularly uncertainty in the depth of the proposed waveguide. This exchange therefore concerns measurable choices: which signals belong in a comparison, what wave velocity a method measures, and how uncertainty affects a predicted note. The fact that both accounts invoke ordinary physics does not make their explanations interchangeable.
Further work and an explicitly tentative proposal
Zhen-Ting Wang’s 2013 preprint proposed resonance of grain chains in a solid-like layer only centimetres thick. Its numerical model treated a chain using elastic mechanics and compared its possible frequencies with observed booming. Wang explicitly acknowledged insufficient experimental evidence for the proposal. The preprint is useful as an original, openly qualified hypothesis; it is not a demonstration that a particular layer at Kelso had the calculated properties.
The published abstract of Vriend, Hunt and Clayton’s 2015 paper distinguishes short broadband “burps” from sustained booming. It reports dispersive nonlinear surface-wave behaviour for the former and a linear compressional-wave interpretation for the latter, supporting the authors’ earlier trapping model. The full paper has not been reviewed for this case, so its abstract is used only for that stated distinction, without adding detailed experimental claims or assigning its measurements to Kelso.
What is explained, and what remains disputed
The outcome is Disputed, specifically for the mechanism selecting and sustaining the sound. There is no need to invent an unseen source for the reported rumble: avalanching sand is directly associated with it. The reviewed scientific disagreement is about the physical coupling and frequency selection. “Disputed” here does not mean that a supernatural explanation has comparable support.
A useful distinction is between asking what starts a sound and asking why it has a particular note or duration. A proposed answer can match a frequency while still needing tests of the material arrangement it assumes. Conversely, disagreement about a model does not erase a recorded sound. The competing papers should be read as attempts to separate those questions, with their methods and geographical limits intact.
Location and access
The map pin is an approximate Kelso Dunes area reference, rounded from the depicted-place coordinates on the accompanying photograph’s Commons record. It is not a measured experimental station, the trailhead or a summit route. Kelso Dunes are in California; Big Dune in Nevada and other sites discussed in the research are comparison locations.
NPS describes a strenuous three-mile return hike with approximately 600 feet of elevation gain, no water at the trailhead and no shade. It advises against summer hiking because of extreme heat. The dunes also support a fragile ecosystem. Use the park’s current advice and conditions; this historical case is not a guarantee of safe access or an instruction to trigger an avalanche.
Original sources and related reading
Vriend and colleagues, 2007, six-page field paper; Andreotti, Bonneau and Clément’s comment, published 2008; Vriend and colleagues’ 2008 reply; Wang’s 2013 preprint; 2015 paper record and abstract; NPS place account and geology and access account. Checked 13 November 2026.
For a different California landform mystery studied with direct observations, see Racetrack Playa’s sailing stones. Its explanation concerns ice, wind and moving rocks, not the acoustic mechanism considered here.
Visit Kelso Dunes
Public wilderness in Mojave National Preserve; strenuous unshaded hike with no trailhead water. Check NPS conditions. Pin is an approximate dune-area reference, not the trailhead.
Official visitor information → · Access information checked 2026-11-13. Check current opening and booking details before travelling.
Share an experience here →Experiences at Kelso Dunes
First-person accounts shared by visitors and reviewed before publication. They are separate from the researched case above.
No visitor reports have been published here yet.