Dataset Deep-Dive

A century of river chemistry, in one archive

GRQA gathers 17 million water-quality observations going back to 1898. Paired with a global map of treatment plants, it becomes a training set for pollution models.

By Water Data Group Desk · Edited by Jagadeesh
5 October 2026 · 1 min read

In brief

The Global River Water Quality Archive holds 17 million observations with records going back to 1898.
HydroWASTE maps 58,502 wastewater treatment plants worldwide, with discharge data.
Together they let a model link what is measured in a river to what is released upstream.

Water-quality data is scattered across thousands of agencies, each with its own parameter names and units. The Global River Water Quality Archive (GRQA) harmonises them: 17 million observations, with the oldest records dating to 1898.

Measurements need causes

A concentration measured at a gauge tells you what is in the river, not where it came from. HydroWASTE supplies one of the main answers: 58,502 wastewater treatment plants worldwide, with discharge data. Joining the two lets a model learn how upstream discharges show up in downstream chemistry.

For salinity, a third source helps: a global compilation covering over 16,000 stations and 7.1 million measurements from 1980 to 2019, for both surface water and groundwater.

This article was drafted with AI from Water Data Group's dataset directory and checked and edited by Jagadeesh before publication.

A century of river chemistry, in one archive

Measurements need causes

Why one dataset of 6,830 catchments matters more to flood AI than any new model

India’s 22,965 groundwater wells: what machine learning can and cannot learn from them

Field note: asking an AI for a live river level, and where the answer comes from