A golden age for Earth Observation
By Sara Aparício, Young Graduate Trainee at the European Space Agency
We are in the edge of an exciting new era for Earth observation (a.k.a. monitoring Earth from space), and anyone can be part of it. Last April, I had the pleasure of holding a series of demonstrations on exploiting large scale satellite data from Copernicus, the largest single Earth observation programme in the world — and the most ambitious to date. These demonstrations, made on behalf of European Space Agency (ESA) and Earth Engine, took place during the European Geosciences Union (EGU) General Assembly 2017 in Vienna, Austria.
Copernicus data is produced by satellite missions called the ‘Sentinels’, aiming to autonomously and accurately provide global coverage of the Earth. Each mission (composed at least by 2 satellites) has different sensors on-board and capabilities. Sentinel-1 (all weather radar mission) and Sentinel-2 (high-resolution optical mission) have six and five day repeat cycles at the Equator, respectively. The revisit frequency increases greatly at higher latitudes since orbit track spacing vary with latitude, increasing overlap near the poles.
Copernicus, the largest single Earth Observation programme in the world — and the most ambitious to date.
These satellites (along with Sentinel-3 and the upcoming missions 4, 5, 5P and 6) provide together a comprehensive picture of Earth’s health with unprecedented wealth of measurements — both in quality and quantity, opening the floor to more extensive research and paving the way to new fields of research. And all data is freely available and free of charge.
However challenges are also emerging
The ongoing and upcoming amount of data being produced will allow for things which would be impossible to be done in the past — but also poses new challenges. This ever increasing rate at which satellite data volume is growing, is in fact overwhelming. And much, so much more, is yet to come. By the end of 2017, the operational Sentinel-1, -2 and -3 satellites alone will continuously collect a volume of 20 Terabytes per day.
Yep. 20 Terabytes. That’s ‘just’ 20,000 GB every single day. The challenges come with storage and processing capacity needs to handling Earth observation data — not to mention time-consumption in running tasks. Want another example? Wagner (2015) gives a nice one: it could take around 2.5 years to download 1 Petabyte of Sentinel-1 data and a staggering 63 years to pre-process it on your own computer. Ouch.
By the end of 2017, the operational Sentinel-1, -2 and -3 satellites alone will continuously collect a volume of 20 Terabytes per day.
There is the urge of maximizing the value of Earth observation data. Science moves fast, and opening data access — and means to use it — allows progress in areas previously inaccessible or time consuming boosting cross disciplinary boundaries. (Isn’t that what ‘open science’ is also about?).
(…) urge of maximizing the value of Earth observation data.
There is no recipe for the perfect way or workflow for handling satellite data. What makes Earth Engine a very interesting tool is the fact it can enable non-traditional users and non-experts to access and make use of satellite data in a cloud-based environment using very intuitive scripts and visualizing the results in the same place. All is done within one single platform — erasing the challenges coming from download data and processing needs.
Now, lets go for the examples shown during EGU.
At EGU, we showed how one can create an app on Earth Engine simply to explore the vast amount of Sentinel-1 and Sentinel-2 data archives for fast data visualization with a user interface that enables to filter data and to choose the best visualization parameters in a matter of seconds. Here are some cases that composed interesting examples:
Ship and Maritime activities monitoring: Strait of Dover
Synthetic Aperture Radar (SAR) data from Sentinel-1, offers a great way to monitor ship routes. In this case, we can see the result from increasing time range and hence increasing as well the number of stacked SAR images. When selecting the maximum backscatter pixel values it enables us to very quickly and visually define the shipping routes in the Strait of Dover over 5 months. On the other hand it is also quite interesting to see the wind-farms recognizable by their stationary and organized structure.
Ice monitoring: Petermann Glacier
SAR data is also highly valuable for monitoring seasonally or permanently ice-covered ocean and land in Polar Regions. The newly famous crack in Petermann Glacier (Greenland) was actually found almost ‘accidentally’ by @StefLhermitte while using Earth Engine.
Land /ocean change and pollution: Ship-breaking activities in Alang
The combination of visible bands from Sentinel-2 in a RGB composite creates an image similar to what our eyes see. In this case we can see one of the greatest ship-breaking areas in the world, in Alang (India), and the changes over a year in ship dismantling activities, land-use change as well as tidal differences.
And finally… a look over Africa.
One last example was another script which sorts and selects Sentinel-2 data to create a cloud-free mosaic over Africa — another great way to demonstrate how few lines of code save probably months of work!
In case you missed these presentations at EGU you can access them here.