Water intelligence for organisations
The public site and the professional layer are one data model — the same entities, observations, sources and rights, with the depth and the access a professional user needs. This page states plainly what you can have today, what is close, and what does not exist yet.
Available now
Not a roadmap. Every item here is something you can use or ask for today, and the counts are computed from the registries rather than written down.
Governed datasets Free
44 datasets, each carrying a semantic version, a stated methodology, a changelog, coverage computed from the data itself, a licence position resolved across every contributing publisher, and a citation. Download where the rights permit it.
Machine-readable access Free
Flat JSON for metrics, sources, geographies, entities and observations, per metric and per subject. Generated from the same registries as the pages, so the API cannot disagree with the site.
Source and licensing intelligence Free
210 sources with commercial reuse, redistribution, derivative use and attribution recorded separately per source, and the date each was last checked. For some organisations this register is more valuable than the observations: it is a map of what water data lawfully exists.
Bespoke extracts and analysis
A geography, a period, a resolution or a format that is not published; a set of sites rather than a country; a written analysis rather than a file. Answered against the actual coverage and the actual rights, with a straight answer where the answer is no.
Analytical outputs Free
4 data-led findings and 13 longer analyses, each with the method published beside the result and the limits stated in the same breath. Derived layers — climatology, flow regime, runoff depth, flood peaks — carry their method version and every input identifier.
Rights confirmation in writing
For a specific set of sources and a specific intended use, including client-facing deliverables. This is the question that most often stops a report going out, and it is answered from the recorded register rather than in general terms.
Close, and worth talking to us about early
Built or half-built, and shaped by what the first few users actually need. If one of these is the thing you want, saying so now genuinely affects the order they arrive in.
Multi-location comparison
Comparisons are currently enumerated rather than open, because an engine accepting any two subjects produces comparisons that are not legitimate. The eligibility rules that make an open comparison safe now exist — geography, resolution, units, overlap, completeness — and are already enforced on the joined-signal layer.
Hydrology explorers
Rainfall against groundwater, river flow, snowpack, reservoir storage and drought indices, with event overlays and lagged association. The engine is built and one pairing can currently be drawn from the data held; the rest are blocked by named collection gaps rather than by the software.
Larger and bespoke exports
The bulk archive is split into million-row parts because that is what opens in ordinary tools. Larger single extracts, filtered cuts and database-ready loads are a delivery question rather than a data one.
Richer API querying
Today’s API is flat files: complete, free at any volume, and unable to answer a query it was not pre-generated for. A query layer over the same model would add filtering and selection without changing what the data is.
Not built, and not pretending otherwise
Each of these is blocked by something specific rather than by a lack of intent, and in every case the blocker is named.
Alerts
Blocked by refresh cadence, not by data. See below — this is the one worth reading in full, because the reason is not the one most people assume.
Portfolio and facility intelligence
Water context across a set of sites a client holds. The entity, facility and geography model supports it; what does not exist is any way for an organisation to give Hydrionis its own site list, which is an accounts problem rather than a hydrology one.
Private organisation datasets
Data belonging to one organisation cannot be served from a public file, and Hydrionis is public files. This needs authentication that exists for a reason, and it does not exist yet.
Enterprise feeds and service levels
A service level is a promise about refresh and availability. Hydrionis will not offer one before it can keep it, which means scheduled collection has to come first.
What actually blocks alerting
Most of what Hydrionis holds cannot support an alert and never will. A statutory annual return published a year in arrears is a historical record, and there is no threshold worth notifying anyone about a year after it was crossed. Two sources changed that: the German federal gauges and the Dutch national network both publish continuously, at fifteen- and ten-minute resolution, and Hydrionis stores daily means from both. That is genuinely timely enough for a threshold alert on river level or discharge.
An alert is also the first feature that genuinely needs an account, because a notification has to reach a person who chose to receive it. That is a real user boundary rather than a commercial one, and it is the point at which accounts stop being decoration. Hydrionis has none, and adding them to look commercial would be the wrong reason: the data is static files, so there is nothing to throttle; a saved comparison can live in the browser; and the licences are open, so a commercial user needs permission from nobody.
What Hydrionis will not do
It does not and will not replicate proprietary market intelligence products. Where a commercial source is the recognised reference for an area — desalination market data being the clearest example — Hydrionis records that it exists, does not purchase or collect it, and builds equivalent factual coverage independently from lawful open sources. That distinction is recorded on the relevant source pages.
A paid Hydrionis tier will never widen what a publisher withheld. Access classes can only restrict further, and no invoice changes a licence — which is why the rights position is resolved per source and stated per dataset rather than summarised in a sales page.
Pricing
There is no subscription and no price list. The public layer is free and stays free; bespoke work is priced against the actual work when there is a specific requirement to price. Inventing a tariff for products that do not exist would be the easiest thing on this page to do and the fastest way to stop being worth trusting about anything else.