Physics-based sonar performance, transmission loss, and source-level modeling — anywhere on Earth, live and up to 10 days ahead. Real bathymetry, forecast sound-speed fields, hydrophone-validated.
For ocean science, offshore energy, environmental impact assessment, ports & shipping — cloud, on-premises, or air-gapped.
> Acoustic prediction loop
Real Clairwave outputs across the world's key waters — each capability, shown working.
Clairwave runs a complete underwater-acoustic prediction loop — from vessel position and source level, through range-dependent propagation over real bathymetry, to received level and listening range — benchmarked against recorded hydrophone data.
Live positions, speed, type, and dimensions for every vessel in view — plus an AIS time machine to scrub back and replay any historical window and run acoustics on past traffic.
Wales-Heitmeyer spectral model estimates broadband source level per vessel from ship class, speed, length, and draft.
RAM parabolic equation solver with real GEBCO bathymetry computes range-dependent propagation loss per vessel bearing.
Pick the physics for the problem: RAM parabolic equation, Bellhop ray tracing, OAST wavenumber integration, or CUDA GPU solvers — matched to frequency, range, and bathymetry.
Map where hydrophones, recorders and listening stations can hear a source — a passing ship or a calling whale — with signal-excess maps over real bathymetry and forecast ambient noise.
Interactive terrain, vessel markers, and volumetric propagation overlays — run in any browser, no install.
Clairwave couples 10-day forecast ocean conditions (GIOPS) with vessel transit prediction: draw a track, project a vessel forward in time, and run tomorrow's acoustics today — future position and future sound-speed field, together in one simulation.
Hydrophones measure total noise. They can't tell you which vessel is responsible, how much noise it would make at a different speed, or what will happen when the next ship arrives.
AIS-linked source levels decompose the soundscape into individual vessel contributions — who, how much, and why.
Model the acoustic field before a vessel arrives. Evaluate sensor placement, speed, and depth scenarios in seconds — not after the fact.
Built on peer-reviewed propagation codes (RAM, Bellhop, OAST) and benchmarked against recorded hydrophone measurements — transparent, reproducible physics you can audit.
Start free. Upgrade to Premium for full-range modeling, GPU solvers, live AIS, and cross-device profiles — or go Enterprise for on-premises deployment, training, and support.
Explore the platform.
Full modeling power.
Sign in, then upgrade in-app — secure checkout via Stripe.
Get PremiumDeployed, trained, supported.
Built for academic institutions, research labs, marine consultants, offshore-energy developers, fisheries & aquaculture, port authorities, and government science agencies — environmental regulators and hydrographic offices.
📅 Book a demoClairwave is an open MCP server (Model Context Protocol). Connect it once and Claude, ChatGPT or Grok can run propagation physics, pull sound-speed profiles and seabed parameters, look up live ships and estimate their noise — and hand you a link that opens the result in the platform. No account, no API key.
In ChatGPT, add the Clairwave plugin. Elsewhere, paste this URL as a custom connector.
https://www.clairwave.com/mcp
{"type":"mcp","server_url":"https://www.clairwave.com/mcp","server_label":"clairwave"}Listed in the MCP Registry as io.github.clairwave/clairwave. Open source (MIT): github.com/clairwave/clairwave-mcp
Thirteen tools; every location accepts a place name. Each prompt opens pre-filled in the assistant you pick (connect Clairwave there first).
Bathymetry, seasonal sound speed, seabed, RAM and Bellhop solvers, live AIS and vessel noise models — each answer carries its run id and a link that opens the run visualized here.
New to Clairwave? Start with the illustrated guide, or find a quick answer below.
Fifty illustrated pages: signing in, a tour of the screen, the globe, flat map and terrain, your first simulation in ten minutes, and connecting your AI assistant. Free to read, no account needed.
Premium users download the complete User Manual (about 240 pages, every control, five worked examples) from the Account panel in the app.
Clairwave predicts how sound travels underwater, anywhere on Earth, in your browser. You choose a place, a sound source and a frequency, and Clairwave returns a 3-D picture of how loud that sound is at every range, depth and bearing. It also estimates the noise of real ships, the ambient noise from wind and rain, and how far away a hydrophone or sonar can pick up a sound.
No. Clairwave runs in a current Microsoft Edge, Google Chrome, Mozilla Firefox or Apple Safari with WebGL 2. It works on integrated graphics, and a dedicated graphics card keeps the 3-D view smooth. Phones and tablets get a simplified layout. If the globe stays black, turn on hardware acceleration in your browser settings.
No. Click Launch Clairwave, then Continue as guest to try it straight away. A free account keeps your sign-in across visits. Premium features need an account.
Free and guest users run the RAM, Bellhop and OAST solvers out to 10 km at 200 Hz, with seasonal sound-speed profiles, live ships around Gibraltar and a 48-hour weather forecast. Premium removes the range and frequency limits and adds the GPU solvers, the 10-day GIOPS ocean forecast, custom sound-speed profiles, live ships worldwide, a 15-day weather forecast and reverberation. See Plans & Pricing.
Clairwave uses established, peer-reviewed propagation models with real bathymetry, ocean sound speed and seabed data for the place you pick. Its ship-noise estimates are compared with hydrophone recordings; you can explore the results in Live Validation. Treat the output as a model prediction for planning and screening, not as a measurement.
Clairwave combines global GEBCO bathymetry, seasonal ocean climatology, the GIOPS 10-day ocean forecast, seabed sediment maps, weather forecasts and live AIS ship positions. You never have to find or load these yourself: pick a place and Clairwave fetches them.
Yes. Every plan can export an HD picture of the vertical slice, a ray-sweep video and the run settings. Premium adds cross-device profile sync, so your working area follows you between devices. Runs started from an AI assistant come with a link that opens the result in Clairwave.
Yes. In ChatGPT, open Plugins, search for Clairwave and click +. In Claude or Grok, add https://www.clairwave.com/mcp as a custom connector. Then ask in plain language. No Clairwave account or API key is needed. See AI Assistants for the steps.
Sign in, open your Account window and click Upgrade to Premium; payment runs through secure Stripe checkout. Manage subscription in the same window lets you change your card, download invoices or cancel.
Yes. The Enterprise plan installs Clairwave on your own or national hardware, including air-gapped networks, with training and in-service support. Contact us to book a demo.
Start with the Getting Started Guide (PDF), which ends with a troubleshooting chapter. For anything else, write to [email protected].
Live vessel tracking, physics-based propagation, signal-excess listening-range maps, and hydrophone-validated prediction. No install, no wait.
Launch ClairwaveFive methods, one answer. A 75 m shelf over a soft clay seabed at 100 Hz, where the seabed controls everything, solved by an exact reference solution and by each Clairwave propagation solver exactly as it runs on the platform.
Median difference to the exact solution over the whole water column, 1–10 km.
The GPU solvers, CWTDE, CWTD and CWNI, are our own CUDA code running on NVIDIA GPUs. Clairwave is a member of the NVIDIA Inception program.
CWTDE does not solve for one frequency at a time. It steps the full acoustic wave field forward in time on the GPU, the way sound actually travels, and still lands within 0.1 dB of the exact solution across the whole water column. Because it works in the time domain, the same run handles what frequency-domain models struggle with: elastic and layered seabeds, ice cover, and pulsed or broadband signals such as sonar pings and impulse responses.
A seabed slower than the water leaks low-frequency sound out of the water column, so the loss climbs steeply with range, here past 120 dB by 10 km. Small modelling errors grow quickly in that regime, which makes it a demanding test for any solver. The exact reference integrates the wavenumber spectrum with a closed-form depth solution; the four platform solvers reach it by four different routes and agree within about 1 dB.
Recorded predicted-vs-measured validation: Clairwave's physics-based acoustic model benchmarked against an independent hydrophone (Lime Kiln, Haro Strait) — an archived dataset you can explore below.
Loading validation data…