Mapping and tracking social & environmental impact with Earth Observation and Big Data Analysis
With Elke Sauter, Liliand Gonzalez, Andrei Bocin-Dumitriu and Evertone Valiati Hemerly.
This talk was given by Elke Sauter and Liliand Gonzalez at the Hague Humanity Hub, on what is earth observation (EO) and how is satellite imagery used to develop new strategies to improve the health and wealth of our planet. Its a fascinating talk for me as this is a completely new subject that I want to learn more about. The first concept I do understand which is high/low resolution images. This is the same principle as looking at ordinary photographs which can be sharp contrast (high) or slightly fuzzy (low) resolution. Secondly, you can have temporal resolution which means how much time does it take to revisit the same site. You may want to have a look at the satellite images that are taken over a period of time or have images that are up-to-date to show the most recent snapshot of the area. Thirdly, depending on the type of remote sensors available you may be seeing reflected light from a surface (passive sensor) or light from the sensor transmitted to the surface and reflected back to the sensor (active sensor) which will also give the distance from the surface at the same time (LiDAR).
With these large datasets from passive sensors (satellites), you can collect colour information that is not available to the naked eye. Along this larger light wave bandwidth than what we could normally see, you can capture more useful information. With the help of these sensors, you can then view infrared and ultraviolet spectrums. These can help with identifying a spectoral pattern by categorising the bandwidths. Each element has it’s own spectoral pattern which can be useful in identifying types of vegetation. Then this is the genius bit that I hadn’t entirely realised, each of the coloured pixels can be split into these bands and then allocated a value, like a chessboard covered in real (decimal) numbers. Satellite images are not just pretty pictures, but actually have a grid system of cells running behind them enabling computation of the images in and of itself. Mind blown. When the colour itself doesn’t matter but the output numbers in the cell, then the results could look dramatically different to what we normally imagine to be the true colour. More about colour theory here.
Red can denote vegetation for instance. By using various band combinations and training algorithms to spot trends, you can answer questions such as ‘how healthy is your crop?’ or ‘how green is your city?’. You can also spot where good healthy vegetation is and where poor vegetation may require your attention. You can spot where there is water and where the threat of wildfires may spread to. The EO Browser already have the indices done for you and help you run simple analysis online. The Dutch have their own freely available 50cm coverage derived dataset, which you can find out about here.
With active sensors (LiDAR & RADAR), you can search beneath the clouds and through vegetation to penetrate to the soils below using radio detection and ranging. However, can use when raining as this distorts the feedback loop of the sensors. You can also find that you end up with too much data to handle over wide areas, but this is very useful in highly local areas and will help you find out ‘how are my trees growing?’.
There are many applications for making use of this freely available satellite imagery, which is where Space4Good comes in. Space4Good aims to help non-governmental organisations and charities understand natural and human phenomena. The view from above enables you to arrive at data-driven decisions based on satellite technologies and other geodata sources. By observing and analyzing these events from space, you can measure and evaluate your impact on society or the environment – for good.
The only reservation I have with the freely available satellite imagery is that the resolution can be pretty poor, which means that outside of huge agricultural areas, the actual usefulness can be limited, unless joined-up with other potentially propriety data sources (expensive) or sensors on the ground (difficult to maintain). The heavy users of satellite imagery will always be governmental departments that can afford to buy commercial satellite data too and have a ready supply of cadastral data etc. NGOs and charitable organisations will find it difficult to use such broad-brush data on its own without access to commerical satelltie imagery on a regular basis and then there is not only the time spent on getting hold of the data, but store and use for the intended purpose can be quite unwieldy.
Let’s hope we can progress with making these data sources more accessible for all, as this can only assist us all achieve our Sustainable Development Goals (SDGs).
Want to keep in touch with what we do? sign-up to our e-newsletter here.
Have a new project? Then call us on 01326 337072 or email us at firstname.lastname@example.org.