The future of EO lies in both large satellites and small satellites. The current scenario shows that big satellites are still needed for environmental monitoring. According to Derr, “NASA’s most recent National Academies’ Decadal Survey includes many important research questions that seek to identify how natural and anthropogenic changes affect ecosystems. Having multiple sensors gathering data at the same time may be the only way to get clear answers.” This requirement cannot be met by small satellite constellations as the problems of spatial and temporal data registration from different satellites in the same constellation and from different constellations will pose severe challenges. On the other hand, frequent revisit and fast data turnaround are invaluable features of small satellites.
In terms of sensors, Synthetic Aperture Radar (SAR) has long promised to be a game changer, but massive satellite platforms were needed, which restricted their use. Today, new technologies have shrunk SAR sensors to enable them to be put on small satellites and the explosion of data from such satellites is clearly visible. The evolution of new sensors like RF detectors and Greenhouse Gas (GHG) detectors are adding a new dimension to EO.
Governments have been major consumers of EO data. Apart from their own satellites, they have awarded massive contracts to private players. However, analyzing data from the beginning requires large investments and will be possible only in government laboratories and in well-funded academia. As Derr says, “Defense departments will be the main drivers of geospatial analytics for the foreseeable future. Large countries with well-funded militaries can afford the best imagery, Artificial Intelligence, and the most skilled GEOINT analysts.” However, she goes on to add that “in general, non-governmental customers aren’t trained analysts, don’t want to hire trained analysts, don’t want to run analysis software, and don’t want to learn how to acquire satellite images. They just want answers.” This is where the future of EO analytics will lie.
According to her, “Freely available satellite imagery and open-source tools can be used by researchers, scientists, local governments, nonprofits, and smaller organizations to gain a level of situational awareness that they have not had previously. Developing this market with affordable analysis is important if EO companies want to diversify their customer base, so that their revenues and success aren’t completely controlled by politicians and military budgets.”
Another interesting aspect is that of citizen scientists. “The digital accessibility of so much information is leading to a rise in the number of citizen scientists who can collect data needed for research, and online open-source investigators that can use things like social media posts to identify what is happening, where, and who is doing it,” Derr says.
According to Antonides, “The key driver for the commercial sector is always return on investment. The boom in Software as a Service (SaaS) offerings that provide analytics, dashboards, and insights into targeted types of data, that is, HubSpot, Google Analytics, Loggly, etc., will continue to grow and diversify. I believe that companies like Databricks and other SaaS data platforms will continue to lower the barriers of entry to enable more companies to expand their use of analytics. In order to better utilize these SaaS data platforms, I think there is a growing niche for smaller Analytics as a Service startups that are able to help bootstrap other startups that are in the seed to B round stage of growth.”
Finally, where will the industry be in terms of Machine Learning (ML) and Artificial Intelligence (AI)? According to Antonides, “It probably goes without saying that ML/AI are going to continue their explosive growth. Combine attractive jobs with new techniques, libraries, and platforms that continue to enable new ideas and you have a recipe for growth. An area I believe you are going to see a large amount of growth in the next few years is edge-deployed ML/AI models.”
All said and done, the promise of EO is yet to be achieved in full measure. We are getting there but impediments remain. Antonides says that the limiting factors are cost, transparency, and trust. Many 4IR technologies are simply still too expensive to integrate into many consumer-level manufacturing and industry applications. Transparency and trust are related limitations. Many of these newer technologies rely on breakthroughs and science that are difficult to understand for many audiences. AI, in particular, invokes a broad set of connotations from the fantastical to the dystopian. “The rise of transparent AI, the ability to explain and understand the results from AI, highlights both the promise and challenges of this technology. Until these technologies are commonplace, or at least better understood, there will be hesitancy by many people to trust and openly rely on these technologies,” Antonides concludes.