fbpx Skip to content

COVID-19 : Context

COVID-19 : With currently almost 1 200 000 confirmed cases worldwide (1), the coronavirus pandemic is no longer in doubt. It is now a phenomenon that affects everyone in the world in one way or another. Half of humanity is confined and those who are not are inevitably seeing changes in their daily lives.

It is in this very particular context that we have decided to observe the climate changes that are the result of this international phenomenon. At Murmuration-SAS, thanks to satellite images, we were able to correlate the advance of the coronavirus with a decrease in the level of NO2 in tropospheric air and a drop in turbidity in the waters of major ports. In other words, this pandemic is having a positive effect on the planet in terms of air and water pollution.

In this new article we will discuss another aspect of ecology: the greener aspect of satellite images, that of vegetation. Confined populations around the world are going out much less in parks, forests, construction work and deforestation are suspended, does all this have an influence on the general state of vegetation in certain areas? This is what we will see in this article.

vegetal cover
Fig.1 – Silence coming back in our forests ?

Land Plant Cover or NDVI

To report on the evolution of vegetation we will use a measure well known to researchers in this field: the NDVI or Normalized Difference Vegetation Index. In general, the NDVI is a simple graphical indicator that can be used to analyze remote sensing measurements, often from a space platform as in our case, to assess whether or not the observed target contains living green vegetation. This indicator will here be a proxy in the sense that it does not directly describe changes in cover but will allow us to create the graphs and curves necessary for our understanding of the phenomenon.

Concerning the data we are working on, they are accessible from several sources but we opted for Google Earth Engine (2) where there are dataset from many satellites. After having presented and used the Sentinel-2 and Sentinel-5P satellites in previous articles, this time it’s the eighth engine of the American Landsat program to be in the spotlight. Landsat 8, or Landsat Data Continuity Mission (LDCM), has been in the air since 2013. It is the result of a collaboration between NASA and the United States Geological Survey (USGS). It is the latter’s images that we are going to process.

But how do you get NDVI values from a satellite image? 

This will involve selecting the appropriate spectral bands (what the satellite sees) and from these, creating a new one. Below, for your understanding, are some of the landsat 8 bands.

Spectral bandWavelengthResolution
Bande 3 – Green0,525 – 0,600 µm30 metres
Bande 4 – Red0,630 – 0,680 µm30 metres
Bande 5 – Near infrared0,845 – 0,885 µm30 metres
Bande 8 – Panchromatic0,500 – 0,680 µm15 metres
Fig.2 – Examples of the “spectral bands” of Landsat 8

This new band will contain the NDVI values of each pixel of the image, and is calculated according to the following equation :

NDVI = NIR – RedNIR + Red   with NIR being the Near InfraRed value for the corresponding pixel, and the same goes for Red.

Mathematical note: With this equation we understand where the “normalized difference” of the NDVI comes from.

Finally, if we filter our data temporally and spatially, apply this new treatment and display it on a map, this is what we get.

covid-19 Render of the processed satellite images.
Fig.2 – Render of the processed satellite images.
Covid-19 Render of the processed satellite images.

Here above are satellite images transformed by our process. It is easy to see fields and their boundaries as well as plantations in a circle.

The area around it may look desert-like due to the colour, but it is not. Indeed, it is the choice of the bands used that gives this sand colour where there is no vegetation.

Comparison criteria and filters

Let us recall our objective: to correlate the advance of the coronavirus and containment with a visible impact on the world’s vegetation. To do so, we have chosen very specific temporal and geographical criteria, as well as exploitable means of representation. 

  • Temporal criteria

It is not possible in this study to make pre-confinement/during containment comparisons. Seasonal variations around the world greatly affect vegetation cover. Therefore, we were forced to make no year-to-year comparisons for the same months of the year. For example, the date range of direct interest to us is from 2020-01-01 to 2020-04-25 and we will compare it with the same dates but in 2019.

Geographic criteria

The geographical starting point for our study was placed in China, near Wuhan, the cradle of the contamination. In order to have varied comparisons, we selected different places: the protected national park of Hubei Shennongjia, a forest in the suburbs of Wuhan and the vegetation within Wuhan itself.

Covid-19  Delimitation of the Hubei Shennongjia Park area in a "geometry" object for Google Earth Engine
Fig.3 – Delimitation of the Hubei Shennongjia Park area in a “geometry” object for Google Earth Engine

After generating graphs on these areas, we had to have control tests. To do this, we used French sites of the same category (the state-owned forest of Arc Chateauvillain as a protected area) and other sites close to the contamination sites, particularly in Italy.

  • Criteria for representation

In these locations and for the selected time ranges we were able to represent the results in two forms. Curves of the evolution of the vegetation surface over time and a graph of vegetation distribution. This second solution proved to be the best because the curves were too imprecise and finally quite invariant from year to year.

Covid-19 NDVI
Fig.4 – Distribution of NDVI by year over the period January/April

With all these parameters in hand, the next step is to analyze these graphs to find a pattern that would provide an answer to our study.

It takes more to tip the balance

Unfortunately, contrary to the correlations highlighted in the two previous papers related to COVID-19, it is not possible to reproduce such relationships in this paper, and after comparing vegetation distribution patterns in similar locations around the world, the results are inconclusive. The analysis we were able to make is that if the coronavirus has indeed had an effect, it would be necessary to compare our current results with a climato-meteorological forecast of what the year 2020 should have been.

 At the microclimatic scale it is the same conclusion, it would seem that the inertia linked to a change in global vegetation is far too great for the very ephemeral “shock” of the coronavirus. Moreover, the precision of the images and filters not being perfect, making any correlation between these two phenomena is for the moment impossible. 

The duration of this containment does not allow for any significant change to be observed and its impact to be studied. This is not a bad conclusion in itself, since it means that, in order to have a real environmental impact, long-term actions must be advocated, and that short-term actions, as we already know, are not enough.

So what are the actions that have a significant impact on our indicator and the world’s vegetation?

Opening : the great reforestation and the case of Pakistan

Around the world, there are many areas of forest poverty. UNESCO recommends an afforestation rate of 12% or more. That is to say that 12% of the surface of the territory must correspond to plant cover. As an example, in 2015 France had a rate of 31% (3) whereas in Pakistan a few years ago it was 2%. These critical situations call for major manoeuvres in the countries concerned. Indeed, in Pakistan between 2015 and 2017, one billion trees were planted and other projects were launched. Today Pakistan has an afforestation rate of 5.37%.

Ethiopia, India, Haïti, Senegal with the Great Green Wall … many countries are embarking on such actions. The goal: fighting against deforestation, drought and climate change. 

Despite all these initiatives, global deforestation continues, and containment may not have helped. Between January and April 2020, in the Amazon, deforestation was 55% higher than in the same period of 2019 and the highest deforestation in the January-April period since monthly statistics began in 2015.

covid-19 the fields are replacing the forest in Brazil
Fig.5 – Easy to see on our maps, the fields are replacing the forest in Brazil.

Finally, let us talk about the reforestation projects that exist on every continent. On multiple scales, things are moving to counter the annual loss of plant surface area. On our site we have highlighted two trustworthy projects:

Both are working nationally to fight deforestation and offer you the opportunity to participate. A last project, international this time, involves precarious populations to reduce extreme poverty by providing work and raises awareness of this crisis among the inhabitants. Check out Eden Project (https://edenprojects.org/) for more information. 

It’s that kind of action that makes things happen. This study, like so many others, shows that staying home is not enough to save the lungs of the Earth. The current crisis therefore still gives us faith to quantify the directions that need to be taken to sustainably reduce the negative impact of human activity on the environment, and it is up to us to take these steps to move intelligently into the future.


  1. : https://coronavirus.jhu.edu/map.html
  2. : Google Earth Engine
  3. : French forests (french link)
  4. : https://www.la-croix.com/Sciences-et-ethique/ (french link)

Do not miss any opportunity

Subscribe to the Newsletter