You may have heard about the tailings dam collapse at Feijão mine in Brumadinho, Brazil. The disaster occurred two days ago on Jan 25 and at least 58 people were found dead, while 300 are still missing. A Sentinel-2 acquisition was planned for today, therefore tonight I checked the EO Browser to see if the mud flow was visible.

Before/after images of Brumadinho mudflow from Sentinel2 imagery (false color composite using the near-infrared band)
Bingo! I just needed to tinker a bit with QGIS to get an independent estimate of the mudflow area... From the EO Browser I downloaded the near-infrared band of the two above images, because the near-infrared band makes it easy to distinguish the vegetation from other surfaces and I computed the normalized difference: (Before-After)/(Before+After).
Normalized difference of the NIR bands of Jan 07 and Jan 27 images (band 8)
I converted this grayscale image to a binary image using a threshold of 0.2, vectorized the result and manually edited the polygon especially in the lower part of the deposit where a thin cloud cover dampens the contrast between both images. Then I displayed the polygon over the Openstreetmap layer... This is when I noticed that the mudflow deposit was already present in Openstreetmap!
Openstreetmap snapshot near Brumadinho on 27 Jan 2019
Using the QuickOSM plugin for QGIS, I downloaded the OSM vector data and extracted the "mudflow" polygon. Its attribute table indicates: "source:image: Sentinel Satellite S2-L1C 2019-01-27 at". Its area is 2.9778 sq. km. My polygon has an area of 2.9585 sq. km.
The extent of the Brumadinho mudlfow by myself and from OSM
It means that someone, somewhere, has updated OSM using today's Sentinel-2 image... That's an impressive example of how open data like Sentinel-2/Copernicus data can serve collaborative projects and emergency mapping.
Oh by the way! This is how it looks in Google Maps:
Google Maps snapshot near Brumadinho on 27 Jan 2019

3 thoughts on “Sentinel-2 + OpenStreetMap = ♡

  1. This kind of speed is indeed a strength of OpenStreetMap. In August 2018, I observed the same after the collapse of the Ponte Morandi viaduct in Genoa, Italy: this case, the first update with the extent of the mudflow came within 36 hours. The first changes were based on photos published on news sites: is first in Version #9 that the Sentinel-2 image is given as the source. This changeset is dated 27 Jan 2019, 20:15 UTC. So you might have been there earlier!

Laisser un commentaire

Votre adresse de messagerie ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.