A RACE to understand COVID-19 lockdown impacts
...using very high resolution satellite data
A guest post by Sara Aparício
As the world faces an unprecedented global pandemic, countries around the world have implemented measures to slow down the spread of the new coronavirus — putting one third of the global population on lockdown. Nationwide quarantines are affecting nonessential businesses in many countries in hopes of containing the outbreak, changing the normal course of human activities — impacting a wide-range of economic sectors.
The European Space Agency has put in action RACE: Rapid Action COVID-19 Earth Observation (…) to address the impacts of pandemic (…)
The European Space Agency (ESA) has put in action RACE: Rapid Action COVID-19 Earth Observation (EO) to support addressing the understanding of some of these impacts. Here is a brief summary of some of the tasks that the Engineering Support Team of ESA Data Application and phi-lab divisions (composed by RHEA and Solenix personnel) are addressing. More specifically, the impacts on supermarkets in main EU cities affected by lockdown, commercial ports and industrial areas. We are analysing over the time the progression and behaviour change in these activities which might support a EU-level scope comprehension — utilizing data available through Euro Data Cube.
Sentinel Hub use on COVID-19
Many users are reaching out to Twitter to share their own visualization of Copernicus Sentinel-2 data taking the advantage of quick and easy access enabled by Sentinel Hub. For example, the impressive resolution of 10 m enabled to depict the decreasing trails left by ships on the cleaerer waters of Venice, or even messages of gratitute and support to the National Health System (NHS) written by a farmer on his lands. Nevertheless this resolution is not ideal to depict smaller features like cars or trucks.
However Sentinel Hub allows you to ingest very high resolution (VHR) data —and it was exactly with Pleiades (2 m resolution pansharpened at 0.5 m resolution) and Planetscope (3.7 m resolution resampled at 3 m resolution) that we started to draft some conclusions.
(…) VHR data suggests that (…) supermarkets are witnessing a considerable decrease of physical customers.
VHR enabled to analyse parking lots serving supermarkets, ports and industrial areas as well cargo activity in the water from over a time frame comprising April 19 and February to March of this year 2020. The first main results from this VHR data shows a considerable difference in supermarkets’ affluence to parking lots. Some medium-sized commercial surface had a considerable decrease (in some cases over 50%) in physical customers whereas large-sized although not as acute also suggest a considerable decrease in the number of cars.
On the other hand — even though it might be too early draw conclusions, imagery acquired over main ports suggest that ship traffic has not been significantly impacted. PlanetScope and Pleiades data over these and other locations were ingested into the Euro Data Cube to allow quick and easy analysis. Keep scrolling down to understand how to do it.
Building our own COVID-19 EO data cube: step-by-step.
It allows you to access very high resolution data, importing your own datasets through bulking ingestion of data. In this case, the goal was to focus on specific Areas-of-Interest (AOIs) (in these case priority industrial locations over Europe). The query of the data using the API was tailored allowing to minimize the square meters we wanted to import into the EO data cube. Below you can find a script to import just the data you actually need. And here were the main steps:
- Define your AOI (define their coordinates) & dates of interest.
- Create GeoJSON (apply a buffer or draw your polygon on QGIS, for example).
- And use Jupyter Notebook below to ingest the data cube:
Future steps
The future steps will mainly consist of continuing weekly analysis with additional historical data and new acquisitions which will be available on Euro Data Cube, which will be supported by an AI-based object detection prototype.
