An Unexpected EO Journey

A thought….curiosity, a playful mind, a good portion of optimism mixed with a pinch of naivety and you have it — a recipe for a truly unexpected journey.

This is exactly what happened when developing a forest monitoring and timber tracing system at Beetle ForTech to curb illegal logging.

But first things first: It goes without saying that forests are vital — let me emphasize this — VITAL for the planet, the climate, our habitat, for YOU. Yet, we keep seeing massive forest areas disappear. One major driver is illegal logging: 1/3 of trees are harvested illegally. Think about it -1/3. That is the size of Iceland year after year. That’s why we developed a system to trace each single harvested tree so that illegal timber does not enter supply chains.

The entire system is essentially based on applying trancing tags on individual logs. But since criminal energy in forestry is enormous, activities on the ground need to be verified. And the only scalable solution to do that is space-borne remote sensing. So we developed a system, connecting our encoding devices (in-situ measurements) with earth observation (EO) technologies. This powerful connection allowed us to verify activities on the ground, at specific locations of harvest. What is even better about this interconnection is, that each harvested and labeled tree represents a ground truth label that can be used for machine learning.

Extremely motivated by the potential of this system, I started to develop an EO-based forest monitoring and verification system. I conducted many expert interviews, did a lot of research, and read loads of papers. I was ready, I thought. On paper all made sense. I simply had to connect all the dots. We hired our first employee, a geospatial engineer with 7 years of practical experience to quickly ramp up the project. We were ready…..

Valley of Tears

I could not have been more wrong.

The episode that followed I now call the “valley of tears”.

We started to develop the first features and very soon we realized that things were very different from what had previously been said in conversations or was written on paper. For instance, in central Europe forestry is characterized by small-scaled and scattered properties, as opposed to eastern Europe, where vast contiguous forest areas exist. So where do we get the data we are interested in? Where and how to store data? Do we need a Geoserver for our application? Yet, many Copernicus-based services, especially preprocessing steps, such as the calculation of forest masks rely on the entire 100x100 km Sentinel tile for the algorithm to work. How can we adapt these algorithms to fit our use case?

We also found that industry experts could not give reliable accuracy measures of their applications. They sort of dogged these questions, were vague and imprecise. 100% of them used my favorite phase “it depends…”. We further learned that the grade of automatization is far behind what we would have expected. This forced us to build our own primitive prototype in order to validate what “experts” were trying to sell to us. From a business point of view, it was impossible to calculate resource requirements or do any cost calculations whatsoever.

Oxymorons Everywhere

So I did a bit of research and I found three pretty interesting sources: ESA Strategy 2025, ESA Technology Strategy, and Eurospace RDT priorities.

What these papers basically tell us is that ESA wants to become a leading global player in the space sector, with a strong focus on the processes surrounding satellite development and tech. This should be achieved by reaching a critical investment mass to kickstart the space sector, next to foresting innovation on all ends, startups, and end-to-end processes (whatever they mean by that).

Yet, we at Beetle ForTech had such a painful, tedious, nerve-racking, Sisyphus-like experience entering the space market. This episode appears to be at odds with Eurospace RDT priorities, which, among others, state that

• bottom-up process should industry-owned and fully transparent
• all technology suppliers can join in, from small to large companies and including research entities, it is a voluntary process
• there are no entry barriers to the process and there is no heavy cost associated to the participation: all is done electronically
• the consolidation process ensures that all technology requirements support the development of a capability not yet available in Europe

So the question is, was it just us and our inability to simply connect the dots, or is there a discrepancy between paper and reality?

I believe that there is no binary answer to that. Truth is, getting more and more involved with the space community, the more I realize, that it is an expert-driven industry focused on upstream developments. In a recent visit to the EU Space Week 2022 in Prague, it became very clear to me that the space industry is indeed moving forward in a bubble with experts talking to experts. At the end of the day, this creates huge barriers to entry making it virtually impossible for outsiders to join the community (my view could also be driven by selective perception, but let’s keep that topic for another discussion).

Space Market — Not Ready for User Uptake

All this points to the conclusion, that the space sector is not ready for user (or StartUp) uptake, opposing claims made on paper. ESA does not lay out concrete steps on how to achieve these goals and their silver bullet seems to be throwing money at Startups. How does this fit into open data and open algorithm policies enacted by ESA?

What is missing in all this is a strategy on how to engage a critical mass of people, not money, to kick-start the space sector.

How is this done you ask?

Well, I could tell you right away.

But given that long blog posts perform worse than shorter ones plus I really want YOU to understand what’s going on, I have to tell you…..you need to wait a bit.

In the meantime, frequently check this blog, as I will continue to add content reflecting on our journey into the space sector. I am sharing my experiences and learnings with you, hoping that you can avoid some of the pitfalls along the way. The journey covers the following topics:

• Space market and structure
• Funding for space applications
• Data — vendors
• Data — retrieval, structure, formats
• Data — quality, artifacts
• Data — Masks
• Data handling in python
• Pros and cons of time series analysis
• Solutions ready to use
• Cloud providers / Cloud computing
• Serverless functions
• AutoML tools
• Labels — access and preparation
• Sentinel data mapping to Keras / Pytorch
• How to integrate EO results
• Building a prototype

The space sector is rich, and there is much to be learned.

I would like to hear your feedback and your thoughts.

• Did you experience something similar?
• Or is my point of view driven by selective perception?
• What are your market entry experiences?
• What are the topics that interest you the most?
• Any other thoughts?

You can either drop me a message in the comment section below, or via LinkedIn.

Stay tuned,

this is Matthias