2021
Amazônia Sufocada (Suffocated Amazon)
Country/area: Brazil
Organisation: InfoAmazonia
Organisation size: Small
Publication date: 18 Aug 2020
Credit: InfoAmazonia project team: Juliana Mori, Thadeu Melo, Rodrigo Menegat, Guilherme Guerreiro Neto, Izabel Santos, Juliana Arini, Leandro Chaves
Project description:
Suffocated Amazon is a I special coverage of Brazilian Amazon fires and deforestation over 2020 dry season (specially August – October). The project has three main fronts that, together, inform the public about the serious crisis taking place in the Amazon. An interactive map with real-time visualization of Amazon fires and automated analysis revealing what are the most critical areas burning. Over the map, a layer of geolocated stories, produced by local journalists, reporting and contextualizing the degradation observed via satellite from the ground. And, a Twitter robot which sends fire alerts on the most critical Conservation Units and Indigenous
Impact reached:
As the project was an extensive coverage along the fire season, which lasted almost 5 months in total, there were several impacts on our audience, and the results show it helped to improve understanding of the fires related to deforestation by our readers and the new readers that came with the project. And also helped them to understand on a regular basis since they could see every day where the fires were, how many they were, and then see the context of what was happening.
Besides the platform originally designed for the project on our website, several stories were published by DW Brazil (which became a re-publisher partner of the project), broadening the audience reach, and also by several national and local outlets in an organic way.
We also had a lot of feedback from environmental and local community organizations, especially indigenous organizations which shared a lot of our reports and posts on social media.
About change of laws or policies, I highlight one of our stories, in Acre state, that reported how the buffer zone of National Park Serra do Divisor was burning more than usual, driven by a bill that intends to weaken the protection of the conservation unity so to make possible the construction of a road that, guided by the intention of draining soybeans to the Pacific, intends to divide the park in half, disrespecting the traditional communities that live there and have not been heard in the process. After our report, the Public Ministry of Acre filed an appeal, citing our report as a source, and requesting the suspension of the construction process till at least the affected communities were duly consulted. The process is still going on.
Techniques/technologies used:
We used several tools and techniques for grabbing the Nasa satellite data, building our own fire database from it, cleaning it, and cross-referencing the fires data with geospatial information (indigenous lands, conservations units, municipalities, estates, and biomes) in order to be able to spatially analyze it and create geo-data visualizations from it and, later, to build the Twitter robot with these same data (and some other specific analyzes). The programming was done mostly in Python, the pre-processing of the geospatial data in QGis/Google Sheet,s and the automated maps were created in Mapbox, and updated daily through the Mapbox API upload.
The workflow happens as follow:
1. We pre-process data from NASA/FIRMS fires and the territorial limits of the municipalities, states and protected areas in the Legal Amazon.
2. Every day, we update this data with new information that we capture from NASA/FIRMS databases.
3. With the updated data we create GeoJSON and MbTiles files that are then sent to Mapbox.
4. From the data and maps styled in Mapbox, we generate the layer maps shown on the special project page and the threads that are published daily on Twitter.
All the code is open-source and we have it documented here: https://github.com/RodrigoMenegat/amazonia-sufocada/blob/master/README.md
What was the hardest part of this project?
The hardest technical part of the project had to do with the satellite data definition. We were almost done publishing the automated fire map in August when the reference satellite used by the Brazilian National Space Agency as a reference (the Aqua satellite, from Nasa) presented technical problems and the publication of data was suspended by NASA.
Because of this, we had to not only to develop our database from scratch but also deepen our understanding of the different satellites and the differences in capturing fire alerts among them, starting a fruitful content partnership with the Brazilian National Space Agency (Inpe).
As we decided to continue the project with another satellite that had more resolution and a longer life span (the Suomi-NPP/VIIRS instrument), the amount of active fire captured was much larger than the official reference satellite. With this, we had two additional challenges: explain in the reports and in the Twitter robot which was the data used, why they were so much more fire alerts than those usually published, and also deal with the large amount of data generated. The fire season ended up with more than 1 million hot spots recorded by the S-NPP satellite in the Amazon, all these geolocated points, updated daily on our map server, were not always usually handled by Mapbox and we had to develop codes to handle this automatically and show all the points in the map.
On the other hand, reporting locally during the pandemic found all the restrictions journalism was submitted this past year, and most of the reports, which covered traditional and isolated communities, had to be done remotely. Thus, it was harder than the pre-pandemic scenarios to get the communities’ reports in order to understand on the ground basis what we’re monitoring by satellite imagery.
What can others learn from this project?
One thing other journalists could learn from this project is the impact on environmental narratives of what we call geo-journalism, the connection between Earth Sciences/geo-spatial data journalistic stories. I believe by being a broad coverage, which was mainly looking at satellite data and their spatial distribution to frame the reports, brought all together this strong element of our production: data connected to the stories, also made from them.
