Much continues to be written about Earth Observation (EO) and AI. The focus has remained on GEOINT, or the defense and intelligence communities. Deep pockets, technical expertise, clear and tangible location-based problems characterize that world.
Outside of GEOINT. Earth Observation and AI have long remained an unknown.
But that is beginning to change. Earth Observation and AI are driving major changes to the GEOINT paradigm, making it more democratized and exposing the previously hidden value of EO and AI to a wider audience. In this article we will discuss 5 ways the paradigm is being changed:
Shining a light into the unknown ..
Stepping back for a moment. My world has long been centred on the unknown. Let me change that .. to the little known. In academic terms that is the study of location. Call it geospatial, location intelligence, area of interest. It is the combination of location data, analytics (see artificial intelligence and machine learning for two examples) and an output which provides insight or an answer to a question.
The location industry has a long history, focused on experts in GIS and remote sensing. Struggling for a clear description of their work, some GIS experts have resorted to: ‘Google Maps on Speed.’ Outside of expert circles the world of location technology is poorly explained and, thus, poorly understood.
Earth Observation and AI sit squarely within the location technology sector.
What is Earth Observation downstream ..
From Wikipedia:
Earth observation (EO) is the gathering of information about the physical, chemical, and biological systems of the planet via remote-sensing technologies, supplemented by [Earth]-surveying techniques, which encompasses the collection, analysis, and presentation of data.
Advances in sensor technology has led to lower prices and miniaturization of hardware. The result has been a tsunami of new location-based data being collected. In the field of EO that has meant imagery, radar (SAR, inSAR), hyperspectral, thermal infrared and more. The data collected is large in quantity and both expensive, and difficult to process and consume. But new algorithms in the form of artificial intelligence (AI) and machine learning (ML) are transforming data processing and analysis.
We are living through a time where location data and analytics are being democratized (translated: no longer an expert only domain).
The EO downstream sector concerns the conversion of data into value- added products
Fear and opportunity ..
Over the last few years much attention has turned to EO. Climate change, sustainability, and ESG are just some of the areas of focus. Investment money has poured into both the upstream (hardware) and downstream sides of EO.
I wanted to get a better perspective of the downstream EO market, so I turned to Liz Derr. Liz is the co-Founder and CEO of Simularity, a US based geospatial AI technology company.
Simularity’s software automatically analyzes geospatial imagery and data to find and classify unusual changes across vast areas
As an expert in the field of EO downstream, I asked Liz to share her thoughts. Change detection using satellite imagery, is a particular area of my interest, and a key area of focus of Simularity. Liz walked me through the 3 main areas of change detection:
- Persistent (eg. new military base, city expansion)
- Transient (eg. typhoon, flood)
- Anomaly or unusual changes (different to normal changes such as tides)
I asked Liz what were the main challenges for remote sensing experts when it came to change detection analysis. She highlighted a number of areas:
- The painstaking nature of the work; days looking pixel to pixel seeing what has changed.
- Errors can easily be made, and changes missed.
- High-res imagery can be hugely expensive.
The advent of AI/ML offers a dramatic improvement in analysis, Liz suggested 4 key areas:
- Helps analysts prioritize where they look.
- Streamlines analysts job.
- Provides an unbiased view of what is going on.
- Allows the observation and tracking of incremental changes (anomaly)
I asked Liz what have been the biggest challenges to the adoption of EO downstream solutions. She mentioned fear (of the unknown), and skepticism around the capabilities of AI. But recognised that the industry “acknowledges that automation is necessary to process all that imagery efficiently. It is a significant change in process and culture to have images processed automatically before they get to the analyst. As with any change, it needs planning, training, and new internal processes to be successful. But the benefits of undertaking this evolution in process are significant for the analysts and their clients.”
Democratizing GEOINT: 5 Ways
Though EO and AI remain centred on GEOINT, long term, the real prize is widening its use to the commercial sector. Fear of the unknown, and technical challenges, have limited non-GEOINT uses. But we are beginning to see light at the end of the tunnel. Five key areas stand out:
- Easier access to pre-processed EO data
- Tools which analyse that data without the need for experts
- Improved Visualization
- Improved Connectivity
- More Value for Lower Cost
I asked Liz for her thoughts in this area. She shared with me some remarkable work done by Simularity which demonstrates the incredible value of EO downstream and its democratization.
Incredible insight ..
One of Simulary’s products is called Automated Image Anomaly Detection System (AIADS). It uses artificial intelligence to analyze new satellite-based imagery in a historical context, identifying changes that diverge significantly from historical norms. That is incremental change or anomalies. Using AIADS, Simularity is analyzing new, publicly available Sentinel 2 imagery as it arrives, validating findings with remote sensing specialists.
The company has set up a service, using AIADS, which focuses on the South China Sea; a region of territorial disputes between Brunei, China, Taiwan, Malaysia, the Philippines, and Vietnam. In early 2021 AIADS picked up unusual activity on Mischief Reef, an area claimed by China, Philippines and Vietnam. Simularity produced a report which indicated “that China has been clearing some areas, constructing facilities and installing radar equipment in the artificial island” (source: The Philippine Star). AIADS had discovered activity which was unknown to both the Philippine and Vietnamese governments. This led to a flurry of activity in the Phillipine DND (Department of National Defense), to verify the activity.
Unlike many others in the EO market who are focused on map making, Simularity is looking to efficiently derive insights about change using satellite and aerial imagery. For AIADS, Liz shared:
A path to the future ..
This is a very powerful example, which demonstrates the capabilities of new EO downstream tools. In this case the focus is on disputed territories, but the uses of anomaly detection by tools like AIADS are many. Specialists are critical to AIADS. But unlike in the pre-AI days, their focus is now only on those areas identified by the system as needing additional scrutiny.
EO downstream offers incredible future possibilities. For solution providers, the challenge now is to focus their sales and marketing efforts on education and sharing customer value.
Recent Comments