RACE Upscaling Competition Results

The Winning Ideas to be Featured on the European ‘Rapid Action Coronavirus Earth Observation’ Dashboard

Sabina Dolenc
Oct 8, 2020 · 6 min read

In April, the European Space Agency (ESA), in coordination with the European Commission, launched a special edition of the Custom Script Contest, focused on the support of space assets during the COVID-19 crisis, managed by Euro Data Cube group. Following a format similar to the Sentinel Hub Custom Script Contest, but further to looking into new algorithms, we were in the quest for ideas on how satellite data could help monitor and mitigate the situation after the pandemic. After seeing the quality of the best contributions, ESA decided to offer them an additional opportunity— to upscale their ideas and integrate them into a common place: the ‘Rapid Action Coronavirus Earth observation’ dashboard. The portal, also known as RACE, provides access to key environmental, economic and social indicators to measure the impact of coronavirus lockdown and monitor post-lockdown recovery.

The authors behind the ideas included in the RACE upscaling competition have “simply” used open and free Copernicus resources on the Euro Data Cube to process them to full European scale. From these submissions two overall winners were selected and are now being integrated into the RACE dashboard. Besides their ideas being showcased on the international dashboard, the winners have also received additional prizes for their effort. A super-prize of 10.000 Euro, meant for the best one, was actually awarded to two teams, in recognition of the quality of their work. The third and the fourth best contributions have received 5.000 Euro prizes.

The ESA’s jury awarded the following ideas.

Truck Detection — Sensing Trade from Space

Author of the idea: Henrik Fisser from Julius-Maximilians-University Würzburg, Germany

Henrik’s proposal uses a novel approach for tracking the motion of trucks. The individual trucks are too small to be detected as objects in Sentinel-2 satellite images but their motion leaves behind a telltale ‘rainbow reflectance track’ that marks their presence.

This new innovative method for detecting trucks was already integrated into RACE dashboard as an economic indicator — the amount of roaming trucks may indicate how the economy of a country is performing. The results show that truck traffic in EU has decreased by an average of 11% compared to median truck counts for the years 2017–2019, and by a significant 42% on the French and Belgian parts of the European route ‘E 40’. For more information see ESA’s ‘Monitoring trucks and trade from space’ post, materials submitted to the Contest, and the author’s GitHub repository.

This new truck detection method was released to the public via the RACE dashboard, and it’s the first citizen-contributed indicator.

Measuring Airport Activity from Sentinel-2 Imagery to Support Decision-Making during the Pandemic

Authors of the idea: Mauricio Pamplona Segundo, Rodrigo Minetto, Cole Hill, Allan Pinto, Ricardo Da Silva Torres and Sudeep Sarkar

Authors have focused on training the model for detecting airport activity during the Covid-19 pandemic. The number of commercial flights was considerably reduced which influenced the global economy. The differences in standard behavior can indicate economic recovery in the months after the pandemic, and unexpected changes can be reported to the authorities.

The solution used the Euro Data Cube Sentinel Hub and eo-learn resources. It demonstrated how to automatically detect parked and flying airplanes over time, analyze the temporal signal, identify anomalies, and correlate changes in the number of airplanes with the Covid-19 outbreak in April 2020.

Measuring airport activity during the pandemic will be soon included into the RACE dashboard as an economy indicator. For more information see the submitted materials available here, and in the authors’ GitHub repository.

Foreseeing the Transportation Modal Shift

Authors of the idea: Michel Deudon and Zhichao Lin, France

Monitoring boat traffic before, during, and after pandemic could help understand the transportation modal shift and help businesses reorganize, as well as provide transparency to citizens and governments. The authors of this submission, awarded as one of the best two thematic ones, offered a principled and unsupervised machine learning approach to detect boat traffic from Sentinel-2 imagery. This approach was guided by the following reasons:

  • To measure and quantify traffic transparently, understand modal shift,
    optimize supply chains.
  • To correlate actions, decisions and observations, support sustainable
    agendas, enforce regulations.
  • To reduce pollution, protect biodiversity, defend local communities,
    raise public awareness.

The proposed approach could be extended to monitor fisheries in designed MPA, vulnerable UNESCO World Heritage Sites, Natura 2000 areas, at scale from space. This would empower European regulatory bodies and encourage further regulations aiming to protect society and the environment.

During the creation of this proposal two years of data in 21 areas in Europe, each covering 200 square km, were analyzed. For more details and a background story we recommend reading the Monitoring boat traffic with public satellites blog post by Michel Deudon. See also the submitted material which is available here.

Water Quality Monitoring for Main European Rivers

Authors of the idea: Giulio Meucci and Francesco Mancuso, Italy

Giulio and Francesco presented the idea to monitor relevant indexes like turbidity of the main rivers in Europe using Sentinel-2 imagery. The turbidity is directly related to the exploitation of the river caused by goods transportation, tourism and agriculture. They evaluated MNDWI (Modified Normalized Difference Water) and NDVI (Normalized Difference Vegetation Index) in order to guarantee an appropriate selection of the water bodies, calculated turbidity and mapped the values on a properly defined color scale.

The processing was performed on Sentinel-2 L2A data, the colormap was taken from MATLAB and resampled on 64 values.

The authors of the idea compared data collected during two different periods; pre-lockdown (January and February 2020) and during the lockdown (April 2020), and obtained the following results for a few different locations.

As seen on the compared images above there is an important turbidity reduction in the rivers they have been inspected. The authors have concluded that these results could be strongly related to the human activities that involve rivers usage. For more details see also their material submitted to the Covid-19 Custom Script Contest available here.

We congratulate all participants and the winners. This kind of public engagement shows that the resources provided by the Euro Data Cube can be used in combination with Copernicus satellite data to monitor impacts of the coronavirus pandemic. We are looking forward to seeing them being used for a good cause.

There is an ongoing Sentinel Hub Custom Script happening right now, with the deadline for submissions on October 31st 2020. Make sure to use the opportunity and submit your script, helping the community to grow and getting a chance to win a monetary prize. More information on the Custom Script Contest web-page.

Relevant links:

Sentinel Hub Blog

Stories from the next generation satellite imagery platform