Cocktail of 27 different pesticides found in drinking water of 1 in every 4 municipalities

2020 Shortlist

Cocktail of 27 different pesticides found in drinking water of 1 in every 4 municipalities

Category: Open data

Country/area: Brazil

Organisation: Repórter Brazil, Public Eye, Agência Pública

Organisation size: Big

Publication date: 15/04/2019

Credit: Ana Aranha, Laurent Gaberell, Sylvain Lesage, Carla Hoinkes, Luana Rocha, Babak Fakhamzadeh

Project description:

Our story unveiled, for the first time, nationwide detection of pesticides in the drinking water. Using the Brazilian freedom of information act, we had access to a public database and discovered that the Brazilian water is broadly tested positive for 27 pesticides (11 of which banned in Brazil and 21 in the European Union). It was the first time this information was made public. Our feature-story reports on nationwide contamination and our interactive map allows city level searches pointing out concentrations above the Brazilian safety limit (where standards are considered flexible) and concentrations that breached the European Union standard (world’s

Impact reached:

This publication had unprecedented repercussions and the impact is still being measured:

– Water companies were called to explain the data from our map by city councils across the country;

– Public audiences specifically about our findings were held in the state of Minas Gerais;

– Our team received an invitation to be part of a public audience held in the Brazilian Congress in order to present the information investigated and debate measures of transparency;

– Two public defenders opened investigations about the presence of pesticide in the water of their cities based on information unveiled by our publication;

– Large cities, such as Brasília, the country capital, had never sent the results of their tests to Ministry of Health (database manager). After being held accountable, water companies promised to be more transparent about the result of their tests. We are still following this story.

– The impacts were, in many cases, on a local level, but it was a direct result of the map repercussions: we tracked over 400 local news-stories (Internet, radio and TV) about pesticides in the water quoting our investigation or map as source. For over a month after the publication, our team answered calls, emails and helped local reporters who were following our investigation to write about their region.

The impact was a result of the repercussions in national and international media. Our original news-feature became the headline of the country’s second largest news webportal UOL (https://noticias.uol.com.br/reportagens-especiais/coquetel-com-agrotoxicos-esta-presente-na-agua-de-1-a-cada-4-municipios/#tematico-3) and was republished by Exame magazine (https://exame.abril.com.br/brasil/1-em-4-municipios-tem-coquetel-com-agrotoxicos-na-agua-consulte-o-seu/).

At least 3 international media outlets published stories about our findings quoting our investigation as source:

– Telesur (South America): https://www.youtube.com/watch?v=tfE0dJjb5IE&feature=youtu.be

– The Guardian (UK): https://www.theguardian.com/world/2019/apr/26/brazil-finds-worrying-levels-of-pesticides-in-water-of-1400-towns

– CGTN (China): https://america.cgtn.com/2019/05/12/water-contaminated-by-pesticides-causes-fear-in-brazil

The findings were also published as part of a broader report by Public Eye on hazardous pesticides: https://www.publiceye.ch/fileadmin/doc/Pestizide/2019_PublicEye_Highly-hazardous-profits_Report.pdf

Techniques/technologies used:

We used R with tidyverse packages (dplyr, tidyr) to clean, explore and statistically process the data. The first prototypes have been designed as Observable notebooks.

Special care has been required to prepare lightweight spatial data for the online map, while maintaining precision when zooming on even the smallest municipalities of Brazil. The spatial data has been formatted as topojson, and projected with the Brazil Polyconic geographic projection (EPSG:5530), using packages such as topojson, shapefile and d3-geo by Mike Bostocks and Philippe Rivière.

The web application has a simple architecture and does not rely on any specific JavaScript framework. The state and the views, as well as the map, the graphics and the legend are managed with D3.js (d3-dispatch, d3-selection, d3-geo, plus some additional plugins such as d3-annotation by Susie Lu). The map is rendered in a canvas due to the large number of polygons, but all the remaining interactions use SVG elements. The CSS is based on the Bulma framework. All the code is versioned with git and published on GitHub.

The webapp has been integrated into the WordPress website (http://portrasdoalimento.info/agrotoxico-na-agua) as an iframe.

What was the hardest part of this project?

We struggled to find a way to present the data to the overall audience in a way that all Brazilians could understand. This was our biggest challenge: to simplify a convoluted database. The information was hidden under layers of technical codes and jargon that were hard to crack – and that not even all technicians understand. The large repercussions proved that our efforts met this goal, unveiling information of high public interest and prompting a call for transparency across the country.

The large repercussion also led us to another hardship and, later on, another relevant impact:

After being held accountable by local media, some of the companies responsible for testing the water started to question our interactive map. Under severe scrutiny and legal threats, our team had to re-check a few specific cases. We proved there was no mistake in our methodology and, by doing so, we realised that the ones making the mistakes were a few of the water companies (the ones responsible for inputting the data in the database). The questions raised about the map unveiled a technical misunderstanding, which needed to be clarified by the Health Ministry. At least two water companies were making mistakes when inputting their data and these mistakes might have gone undetected for years in case the map had not been published.

In conclusion, our efforts to read the data in a way the overall public could understand helped to correct a communication problem between the Health Ministry (database manager) and some of the water companies (responsible for data input).

What can others learn from this project?

Some very relevant stories are still untouched by journalism. Our team had been covering the issues of pesticides for years, but never thought that our drinking water could offer such risk. Neither that such a rich database existed in Brazil.

International collaboration is hard to operate, but sometimes it is fundamental to find a story. Had we been alone in this front, we might have never looked for this data, as this has never been an issue discussed in our country. Also, the parameters to read the data were largely improved by the comparison to the European Union regulation. Had we not used this parameter, our conclusion might have been that there was no story, since the water contamination was mostly within the Brazilian standards (regulation which is questioned in our report). It was the comparison with the European Union standards that made us “see” the problem.