THALWAG — OCEAN SCIENCE
The unfinished map.
In 1972, we took the first photograph of the entire Earth from space. In 2012, we mapped Mars at a resolution of 20 centimetres per pixel. In 2026, more than 80 per cent of Earth's ocean remains unmapped at any useful resolution, and less than 0.001 per cent of the deep seafloor has ever been directly observed by any instrument or human eye.
This is not a funding failure or a technology failure. It is a failure of institutional imagination. Space exploration had a NASA. Weather had a WMO. The ocean — which regulates the climate, supplies half the world's oxygen, and feeds more than a billion people — has never had an organisation whose sole purpose is to understand it in real time, at depth, at scale.
The Indian Ocean is among the most climatically significant bodies of water on Earth. It drives the Asian monsoon, modulates sea-surface temperatures across the tropics, and absorbs a disproportionate share of anthropogenic heat. It is also among the least observed. The World Ocean Database shows that the Indian Ocean has received roughly one-fifth of the oceanographic survey attention of the North Atlantic, despite covering a comparable area.
World Ocean Database, NOAA National Centers for Environmental Information
What data does exist is geographically biased in a way that reflects wealth more than science. The majority of historical deep-ocean temperature profiles come from a small number of countries: the United States, the United Kingdom, Germany, France, and Japan. The waters above which India's 200,000 fishing vessels work every day are among the least characterised on Earth.
01 — THE APPROACH
You cannot measure the ocean.
So we read it instead.
Direct measurement of the entire ocean is physically and economically impossible. THALWAG does not attempt it. Instead, the system reads the ocean: it collects sparse observations from distributed vessels, then uses ocean physics and data assimilation to reconstruct the complete state — temperature, salinity, oxygen, current — from those sparse signals.
Reading is not guessing. A marine physicist reading a temperature gradient knows that it implies a density difference, which implies a pressure gradient, which implies a current. The ocean is a physical system that obeys equations we understand well. Those equations are strong enough constraints that, given a sufficient distribution of observations, the complete state can be inferred — not perfectly, but usefully, and with known uncertainty.
The relevant precedent is weather forecasting. Meteorologists do not measure the entire atmosphere. They maintain a sparse network of radiosonde stations, surface observations, and satellites, and they use those to initialise and continuously correct numerical weather models. Global weather forecasts have improved by roughly one day of skill per decade for fifty years — not because the atmosphere became easier to observe, but because the models became better at reading it.— ECMWF, 2023 annual report
THALWAG applies that logic to the ocean, starting in the waters where India's fishing fleet operates and where the observation gap is largest.
02 — THE PRINCIPLES
Eight laws of total
observation.
These are the principles that govern how THALWAG approaches the problem of observing an ocean that cannot be fully measured. They are not engineering specifications. They are commitments about what the system must do and what it must never do.
01
Observe where no one else is.
The observation gap in the ocean is not random. It maps precisely onto geography and wealth. A system that observes where research vessels already go adds little. THALWAG must reach the waters that are blank on existing charts.
02
Read signals, not just measurements.
A single temperature reading is a fact. A transect of readings is a gradient. A time series of gradients is a physical story the ocean is telling. The system must be designed to read the story, not just record the facts.
03
Let physics constrain the model.
The ocean obeys equations. When observations are sparse, the model must honour those equations rather than interpolate freely. Physically impossible states are not candidates, no matter how well they fit sparse data.
04
Require the model to be wrong.
A model that cannot be contradicted by new data is not science. Every observation must be capable of disproving the current model state. Self-correction is the mechanism, not the failure mode.
05
Anchor every datum in time and place.
An observation without a precise coordinate and timestamp is not an observation. Provenance is not metadata appended to a measurement. It is inseparable from the measurement itself.
06
The observer is a partner, not a source.
The fisher who carries a sensor is not a data source to be extracted. They are a co-investigator whose local knowledge — of currents, seasons, and fish behaviour — informs what the sensors measure. Partnership is not courtesy; it is design.
07
Open by default, closed by exception.
Ocean science requires the full record, including the observations that say nothing unusual happened today. Data held privately fragments the record. Closure must be justified, not assumed. All THALWAG observation data is published under CC BY 4.0.
08
All ocean is connected.
A salinity anomaly in the Bay of Bengal affects monsoon rainfall over the Deccan Plateau. Heat absorbed in the Arabian Sea moves into the global thermohaline circulation. There are no local ocean problems. Observations made anywhere feed a model that is relevant everywhere.
03 — QUESTIONS
Three questions
people actually ask.
Ocean observation involves unfamiliar instrumentation, statistical methods, and physical assumptions. These are the questions that come up most often from fishers, journalists, and scientists encountering THALWAG for the first time. The answers are direct and do not pretend to more certainty than we have.
- Is this the same as satellite sea-surface temperature data?
No. Satellites measure the uppermost microns of the ocean surface — the radiometric skin layer — not the water column below. THALWAG sensors are deployed in the water, measuring temperature, salinity, and dissolved oxygen at depth. Sea-surface temperature from satellites tells you what the skin of the ocean looks like. It does not tell you what the ocean is doing underneath, which is where the climate-relevant processes — heat storage, stratification, deepwater formation — occur.
Satellite and in-situ data are complementary. The satellite sees everywhere; the in-situ sensor sees deep. Both are necessary. THALWAG data is designed to be assimilated alongside existing satellite products, not to replace them.
- If most of the ocean is unmeasured, how can a model be accurate?
Ocean physics. The Navier-Stokes equations for fluid motion, combined with the equation of state for seawater, constrain what physically possible ocean states look like. When observations are sparse, the model does not guess — it solves those equations within the bounds set by available data. This is called data assimilation, and it has been the basis of operational ocean forecasting for decades.
Accuracy improves with observation density. The THALWAG model is more accurate in the Arabian Sea than in the eastern Pacific simply because there are more observations in the Arabian Sea. The model reports its own uncertainty; in poorly-observed regions, the uncertainty is larger. That is the honest answer.
- What parameters does a participating vessel actually measure?
A standard THALWAG sensor package measures: water temperature (accuracy ±0.01 °C), practical salinity (±0.01 PSU), dissolved oxygen (±0.5%), and pressure and depth. GPS coordinates and timestamps are logged automatically with each reading. Readings are taken at a configurable interval, typically every five minutes while the vessel is underway.
Vessel identity and precise route are anonymised at the vessel operator's discretion before transmission. The transmitted record contains a vessel-class identifier, not an individual vessel ID, unless the operator explicitly opts into named attribution for citation purposes.