Explore global water quality and its evolution over 20 years. 

Contributing to Sustainable Development Goals really gets us out of bed in the Derilinx Team! 

Clean Water and Sanitation is critical for Sustainable Development. Being able to manage risks linked to Water Quality is paramount. 

The new Open Data powered application “Quality Unknown” enables users to track overtime major risk categories that impact water quality across the globe at a 50-kilometre spatial scale. 

This app is the fruit of a close partnership between the World Bank Water Practice team and the Derilinx. It represented a significant challenge (and opportunity) to deliver a powerful story powered by Open Data. It really crystallised the need to progress from the “data published” stage to “data linked and visualised” stage to create impact!

Quality – Not so unknown any more!

About Quality Unknow – The App

This application visualises global water quality risk at a disaggregated level for the entire world, from 1992-2010 (Quality Unknown: The Invisible Water Crisis). Use this application to create maps or download data for analysis on water quality at the country or local level.

SDG 6.3.2 identifies 3 major categories of water quality parameters that pose global water quality risks: nutrients like nitrogen (nitrate-nitrite) (N) and total phosphorous (P); measures of salinity like electrical conductivity (EC); and widely used umbrella proxies for water quality like biological oxygen demand (BOD), and dissolved oxygen (DO). Even though in-situ water quality data are sparse in time and space, data sets of most of the drivers of water quality are available at global scale and over time. By using machine-learning algorithms that combine data on these drivers, these water quality parameters are predicted at a 50-kilometre spatial scale across the world.

This work primarily relies on machine-learning algorithms known as Random Forests, a method that seeks to find the combination of factors that explains observed water quality by estimating thousands of decision trees. As opposed to many other machine learning algorithms, Random Forests is a transparent method. It relies on little parametrisation and can be consequently applied to a broad range of pollutants in a harmonised methodology.