The countless opportunities unlocked by satellite images
Understanding the diversity of available satellite images and their applications
The democratization of optical satellites, whether private or public, has led to a dramatic increase in the number of available images, thus multiplying the number of use cases potentially addressed thanks to this data. From the support of rescue operations to the assessment of deforestation, including military usage, potential uses are so numerous that it can quickly become overwhelming.
Heka, especially through its Computer Vision Lab, has developed an expertise in the exploitation of such images by carrying out substantial projects such as helping to protect sensitive environmental areas, detecting illegal activities or estimating solar potential for buildings.
Through a series of three articles, our goal is to provide a clear overview of the satellite image ecosystem. This first article aims to introduce various use cases currently feasible thanks to available data, and how they can be gathered by domain, actors or types of data. The following two articles, more technical, will focus on :
- The actors along the data chain and the issues they raise;
- The different steps required to obtain a good looking image from the data captured by the satellite.
I — Overview of satellite image use cases
Ask a random person what would be the first application using satellite images that pops to their mind. The answer would probably be military and surveillance usage. The war opposing Russia and Ukraine is actually a very good example of such use. Photos taken by Maxar, a major private company in satellite imagery, helps to assess the movement of the Russian army and their positions, as stated in an article published by Le Monde (a French newspaper) on the 1st of March 2022.
However, there are many more ways to use satellite images besides military applications, that are less controversial. As an example, climate and environment can also benefit from satellite observations, with the MODIS satellites (two satellites launched by NASA). With their low resolution and high number of spectral bands, they were specifically designed for earth and climate measurements, enabling the evaluation of water level changes, the global snow cover trends or the monitoring of vegetation health.
With the advent and democratization of deep learning computer vision algorithms, the number of use cases skyrocketed. Let’s suppose that someone is looking to develop solar energy in an area and wants to know which buildings should be equipped first. For this purpose, they could use satellite images to analyze roof surfaces, obstacles and orientations to target the most promising ones. Let’s say you want to map roads of a certain area, but the fast evolution or the complexity of its access makes the task complicated. Why not use satellite images to automatically chart this territory ? Detecting the amount of available oil in tanks, assessing deforestation, predicting the production of a specific field, avoiding a new Titanic sinking by detecting icebergs… are just a few examples of use cases unlocked thanks to satellite images.
As stated previously, the number of use cases are endless. We will show you how to successfully navigate this world by classifying them, whether by domains, actors, or types of images used.
II — Major axes of analysis : Actors, types of images and field of applications
1. Actors
One can classify the end-users of satellite images into three categories : military, other public organizations and private actors. From the assessment of situations to the support of operations, one can easily imagine the potential military applications. However, the ability of each country to spy on its neighbors using satellite images is a major issue that military forces have to apprehend.
Other public organizations can benefit from these images. When it comes to the supervision of protected areas, satellite images could be a real asset to detect illegal constructions or wild dumps contaminating the soils. Finally, private actors can also benefit from satellite images adapted to their business needs, such as farmers evaluating the health of their fields or gas pipeline owners detecting leaks.
2. Types of images
This wide variety of topics is ensured by the range of satellites and sensors orbiting the Earth. Three key notions are important to understand:
- Temporal resolution: how often can I access a new image of my area
- Spatial resolution: how many square meters are represented in one pixel of my image
- Sensors: which information does my sensor capture
Optical images, for instance those required to map an area or establish the solar potential of a building, are very similar to the ones captured by your smartphone. To acquire such images, sensors need to capture data in the optical range, leading to black and white photos (panchromatic) or color images (made of green, blue and red). Other spectral bands may be useful, as they provide other information as near-infrared band which is often used in agriculture. Associated with the red band, it leads to the creation of vegetation indexes and simplifies vegetation detection and discrimination. This explains why MODIS satellites have 36 spectral bands, as it widens the scope of analysis. Finally, other technologies based on active sensors (which send and receive information instead of simply receiving it) helps avoid certain constraints that come with passive sensors. SAR (synthetic-aperture radar) is a good example as it can be used to detect vessels or icebergs and avoid collisions without being impacted by the presence of clouds.
The temporal and spatial resolutions are also use case specific. Analyzing a glacier melt requires years of data on a large scale, while observing the Russian forces in Ukraine requires high precision images that are frequently updated. Hence the difference in satellites used for these different tasks.
3. Fields of application
Ultimately, usages can be categorized into different fields of application. Environment, oil and gas, agriculture and logistics are some of the major topics we identified.
Analyzing climate is a vast subject that uses many types of data provided by weather satellites. With the launch of the first weather satellite in 1959, researchers have built a consequent historical database. Besides, the precision and versatility of online data enable deep predictions of weather. For example EUMETSAT (the European Organization for the Exploitation of Meteorological Satellites) recently developed RGB products for multispectral imagery, used later by the Alaska region to predict weather despite its complex region. On the other hand, historical data can be used at different timescale : online to manage marine traffic in protected areas, dailies for starting forest fires, monthly for forest health monitoring or deforestation assessment, and yearly for climate change with multiple applications such as dwindling ice covers, desertification, evolution of coral reefs or sea shores…
Environment is not the only field that can benefit from satellite images; a large number of applications is also available for the fossil energy sector. For example, automatic studies of floating roof tanks can help assess the current oil stock, which is generally difficult to estimate. Thanks to the shadows made by the tanks (see the picture below), it is possible to evaluate the quantity of oil within the tank. By combining this information with the height and radius of the tank, the total quantity of oil is computable. Some companies also use satellite images to secure pipelines by detecting gas and oil leaks and spills. Others companies can benefit from these images by scanning large and remote areas in order to find exploration areas and discover new oil deposits.
Besides optical images, other types of satellite images can offer many opportunities, especially agriculture. Indeed, SAR images enable farmers to get insights on soil composition and moisture, regardless of the climatic conditions. They can then adapt their day to day operations and increase the precision in the quantity and spread of fertilizers, pesticides or water. As explained previously, near-infrared bands help discriminate the different types of crops. The data collected then helps to estimate the current health of a field and its future production.
Logistics can also largely benefit from SAR images. When natural disasters occur and damage facilities and communication channels, they are a real asset to assess the situation (defining which areas have been touched, where the help should be deployed first, etc.) and define a realistic plan of action. They can also be helpful to secure transport, especially sea traffic. Even if the Automatic Identification System (AIS) is now supposed to be installed on boats, this system can be disabled and collisions could occur. Such incidents could be prevented with the use of SAR images, as vessels can easily be tracked thanks to this technology. Other maritime dangers such as icebergs can also be identified. Another potential use case is the automatic detections of roads in order to quickly update maps in fast-evolving environments or remote locations.
Conclusion
The large variety and diversity of satellite images available opens the door to countless opportunities in various fields of application. A project or business idea might even have sprouted in your mind while reading this article, but let’s not get carried away too quickly, we have only skimmed the surface of the issue.
Indeed, you will first need to identify what type of satellite images are suited for your use case and how to acquire it (free of access or purchase). Once collected, the data will then have to be preprocessed and properly structured in order to fully exploit it. These topics will be the subject of our next two articles coming up soon.
References
- Le Monde, ‘Guerre en Ukraine : des images satellites montrent un immense convoi militaire russe près de Kiev’, March 2 2022,<https://www.lemonde.fr/international/article/2022/03/01/guerre-en-ukraine-des-images-satellitaires-montrent-un-immense-convoi-militaire-russe-pres-de-kiev_6115636_3210.html>
- Sozzi, Marco & Marinello, Francesco & Pezzuolo, Andrea & Sartori, Luigi. (2018). Benchmark of Satellites Image Services for Precision Agricultural use.
- Nikiat Marwaha Kraetzig, UP42, January 15 2021, accessed February 2022,<https://up42.com/blog/tech/a-definitive-guide-to-buying-and-using-satellite-imagery>
- Md. Mubasir, ‘Oil Stroage Tank’s Volume Occupancy On Satellite Imagery Using YoloV3’, September 2 2020, accessed February 2022, <https://towardsdatascience.com/oil-storage-tanks-volume-occupancy-on-satellite-imagery-using-yolov3-3cf251362d9d>
- Wikipedia, Moderate Resolution Imaging Spectroradiometer, accessed December 3 2021, <https://en.wikipedia.org/wiki/Moderate_Resolution_Imaging_Spectroradiometer>
- Drones Imaging, <https://www.dronesimaging.com/wp-content/uploads/2013/05/documentation/indice%20de%20v%C3%A9g%C3%A9tation%20NDVI.pdf>
- [1] Potential directions for applications of satellite earth observations data in Egypt — Scientific Figure on ResearchGate. Available from: https://www.researchgate.net/figure/Optical-properties-that-lead-to-retrieval-of-a-vegetation-index_fig3_241111282 [accessed 2 Mar, 2022]
- [2] Monitoring of Oil Tank Filling With Spaceborne SAR Using Coherent Scatterers — Scientific Figure on ResearchGate. Available from: https://www.researchgate.net/figure/Appearance-of-an-oil-storage-tank-with-a-floating-roof-in-a-nadir-optical-image-a-and-a_fig1_351757672 [accessed 2 Mar, 2022]
- [3] Gao, L.; Song, W.; Dai, J.; Chen, Y. Road Extraction from High-Resolution Remote Sensing Imagery Using Refined Deep Residual Convolutional Neural Network. Remote Sens. 2019, 11, 552. https://doi.org/10.3390/rs11050552
- [4] https://pixnio.com/free-images/2017/02/13/2017-02-13-11-15-22.jpg
