EEA Data and Maps - Water

It is the environmental evidence base Europe argues over. The European Environment Agency's Analysis and data portal is the entry point to the agency's water datasets, indicators and map products, and the water content concentrates in the Waterbase series - the EEA's generic name for its databases on the status and quality of Europe's rivers, lakes, groundwater bodies and transitional, coastal and marine waters, on the quantity of Europe's water resources, and on emissions to surface waters from point and diffuse sources of pollution.

The themed releases give the collection its span. Waterbase - Emissions 2026 carries 1977 through 2025 of loads to surface waters keyed by country and spatial unit, published July 2026. Waterbase - Water Quality ICM 2024 spans 1900 through 2024 - a century of chemistry - shipped as CSV plus SQLite. Further releases cover water quantity, biology and quality, each linking a metadata record, a factsheet and dataset-definition CSVs that document every column before you ever load a file. Alongside the tables sits the WISE freshwater information system, the interactive dashboard layer over the same evidence base.

Datadory turns that catalog into delivered data: typed rows, definitions attached, vintages mapped between database versions. Get a sample of this dataset cut to your countries, determinands and years, and judge the rows before anything ships. It sits inside our water utilities data hub beside the rest of the industry's catalog.

One aggregated observation from the Waterbase state-of-environment tables, exactly as it lands in a delivered extract:

collection     : Waterbase WISE State of Environment (SoE)
table          : Waterbase_T_WISE6_AggregatedData
countryCode    : FR                        # reporting country
site           : FRGR04067200              # monitoringSiteIdentifier
determinand    : Potassium                 # CAS_7440-09-7
year           : 2011                      # phenomenonTimeReferenceYear
mean           : 3.15      uom : mg/L      # resultMeanValue / resultUom
sampling_period: 2011-01--2011-12          # parameterSamplingPeriod (ISO 8601)

Read it as a measurement chain rather than a number. Every value arrives with its whole identity attached - which country reported it, which physical site it came from, which substance was measured and under which registry code, which year the statistic describes, what unit it speaks, and precisely which months the sampling covered. The minimum and maximum ride alongside the mean in the full extract, so dispersion comes built into the row instead of being reconstructed from somewhere else.

That completeness is the practical difference between a dataset and a spreadsheet. A panel built from these rows assembles with a group-by, not a parsing project - and every row you request arrives in exactly this shape, scoped to whichever sites, substances and years your work names.

Seven fields form the spine of every aggregated observation, verified against the published dataset-definition documentation: where the record was reported (countryCode), what it was measured at (the WISE monitoring-site or spatial-unit identifier), what was measured (the observed-property determinand code and label from the Eionet codelist), when it applies (reference year and ISO 8601 sampling period), how much (the aggregated result statistics) and in what unit (resultUom).

The dictionary is what makes the collection joinable across decades and borders. Determinand codes resolve against a controlled codelist, so potassium measured in 2011 in France shares a vocabulary with anything measured anywhere else in the network; site identifiers follow WISE conventions, so site rows roll up cleanly to river-basin-district and national aggregates where tables publish them. Additional fields on request: the wider release envelope extends past the core grid - matrix breakdowns, result qualifiers, spatial-unit hierarchies and companion definition tables among them - and we pin those columns against your sample rather than promise them blind. Name the analysis and the extra columns arrive mapped onto it.

• Geographic: EEA member and cooperating countries across Europe - 32-plus countries - reporting under shared vocabularies. Site-level records roll up to river-basin-district and national aggregates in several tables, so a single schema serves both local and policy-scale questions.
• Temporal: releases span 1900 through 2025 depending on theme - Water Quality ICM running 1900-2024, the emissions release carrying 1977-2025 - refreshed annually or biennially per theme. Few water records anywhere offer a century-scale baseline for trend work.
• Granularity: monitoring-site level aggregated statistics by determinand, matrix and reference year, the finest consistent grain published; river-basin-district and national aggregates appear alongside in some tables. Tens of millions of rows across the themes in total.

Set against the wider Datadory catalog - where the average quality score across all 1,744 datasets is 7.81 - this record scores 8/10, carried by verified column documentation and the sheer depth of the reporting network behind it. For vocabulary: Waterbase is the table family, and the site-level grain is what monitoring-site level statistics describes.

API, files, or your warehouse. Daily, weekly, or hourly.

You pick the channel and the cadence; extraction, typing and vintage-mapping stop being your problem. Because the source publishes in themed releases on an annual-to-biennial rhythm rather than a daily drip, most teams take the full history once as a backfill - every theme back toward its first year in one pass - and keep new releases rotating in on whatever rhythm their models expect. Whichever way you take it, the seven-field spine travels unchanged, and version-to-version shifts between database vintages are mapped upstream instead of letting your joins discover them mid-analysis. A sample ships first either way, sized to test inside your pipelines the same day.

• Water-quality modelers build long-run trend panels from monitoring-site statistics - aggregate by determinand, matrix and year across rivers, lakes and groundwater without touching a parsing job; see data scientists use cases.
• ESG and compliance teams screen emission pressure around assets and holdings using the 1977-2025 emissions layer, the accounting side of the collection that sits upstream of any outcome claim; see ESG emissions analysis.
• Environmental journalists and academics cite comparable multi-country monitoring statistics and century-scale baselines in chapters, papers and investigations - numbers published by the institution itself, attributable and defensible; see journalists academics use cases and citation-grade research.
• Product builders wire European water layers into applications on a stable, documented field spine; see developers builders use cases.
• Consultants and market researchers anchor European water chapters in harmonised multi-country statistics whose cross-country comparability was enforced upstream; see market researchers use cases.

Data scientists and ML engineers (relevance 2 of 3) get a documented, decade-spanning panel with typed columns and controlled codelists - feature engineering starts on real rows, not format archaeology. Market researchers and consultants (2 of 3) get comparable multi-country statistics that make European water chapters quotable across borders. Journalists, academics and students (2 of 3) get the institutionally published version of Europe's water story, citable down to the site and the year. Developers and data-product builders (2 of 3) wire water layers into applications against a stable dictionary. Investors and quant researchers (1 of 3) use it as a diligence layer - water-quality and emission pressure around assets - and competitive intelligence and product teams (1 of 3) track the regulatory-environment backdrop shaping water-tech rivals, country by country.

Three things worth settling upfront. First, scope: this collection rewards precision - name the countries, determinands, sites or basin districts and the year range, because tens of millions of rows are only useful once cut to a question. Second, grain: observations are aggregated statistics per site, determinand and year, not raw grab samples; if a workflow needs sub-annual measurement events, say so at scoping and we will enumerate what the collection can and cannot support before delivery. Third, vintages: themed releases are reissued on annual or biennial cycles, and values can shift between versions - Datadory maps those shifts so a corrected figure never masquerades as a new one downstream. Tell us the decision the data feeds and the sample arrives shaped around it.

Cards worth reading next: the country-year counterpart, the England and Wales benchmark, the facility-level emissions ledger from the same agency, and the primers that decode the vocabulary.

World Bank Water & Sanitation Topic Data answers questions that name countries rather than places smaller than them - withdrawals, stress, access coverage on a common national grid. Our head-to-head comparison scores both line by line. Discover Water - England & Wales Water Company Performance Dashboard benchmarks all 19 English and Welsh companies on leakage, interruptions and pollution incidents - operator performance where this record is environmental state. EEA European Industrial Emissions Portal (E-PRTR) comes from the same agency but flips the lens from the receiving waters to the industrial sites doing the releasing, and EEA Plastics Topic Data follows one pollutant family end to end. Where this record lands overall: the best water utilities datasets ranking, the European Environment Agency source profile, and the water utilities data guide for the full industry stack.