Last Week in Earth Observation: April 18, 2023

+ Reviewing Announcements from Planet Explore

Aravind

Apr 18, 2023

Hey! Welcome to a new edition of ‘Last Week in Earth Observation’, in which I attempt to curate the major developments in EO from the week that just passed and provide some thoughts & analysis on some of them.

Along with the usual summary of developments in EO, you will be reading some of my thoughts on the announcements from Planet at the Explore conference.

PS. Yes, this is getting to you a day late. While I should have written this over the weekend, as I usually do in order to be delivered on Monday, I turned 32 this past Saturday - so, I decided to take a couple of days off to explore DC. Worth it!

Four Curated Things

Major developments in EO from the past week

1. Financial Stuff: Funding, Contracts and More 💰

Climate risk platform ClimateAi has raised $22M in Series B funding;
Hohonu, a US-based climate tech startup, raised a $1.8M pre-seed funding round to build coastal monitoring products;
Supply chain monitoring firm Everstream closed $50M in Series B funding;
NASA awarded Capella with a contract to provide high-resolution SAR data for evaluation within the Commercial SmallSat Data Acquisition program (last week, we mentioned Iceye had been awarded a similar contract);
NV5, a major geospatial consulting firm has completed the acquisition of the subscription-based geospatial software solution from L3Harris;
IBM is exploring a sale of its weather business, The Weather Company, which it had acquired from The Weather Company for $2B;

2. Strategic Stuff: Announcements and Partnerships 📈

Hyperspectral imagery analysis firm Metaspectral is partnering with SkyFi to have satellite imagery directly available on its platform;
Aerial imagery provider Bluesky, which is also a partner of SkyFi, is also joining hands with EO platform UP42.
French EO analytics firm Kayrros announced the launch of its Forest Carbon Monitor solution based on remote sensing data;
Brazil and China finalised their agreement to build and launch the Cbers-6 satellite with a radar instrument, specifically aimed at Amazon rainforest monitoring among others (in Portuguese - h/t Rafaela Tiengo);
Picterra announced the integration of Meta’s Segment Anything Model (SAM) onto its platform, while Esri followed soon after with the integration of SAM onto its ArcGIS system.

3. Interesting Stuff: More News 🗞️

The UK’s future with the EU’s Horizon and Copernicus programmes is in doubt casting doubts over the future of EO research and commercialisation;
Microsoft Azure demonstrated how a ChatGPT-like demonstrative AI could be used to request relevant satellite data from a number of EO data providers;
SpaceNews published this piece that discusses the age-old problem of the lack of transparent pricing of satellite imagery and this one that talks about how the US military still lacks timely access to satellite imagery (yes, really!);
The Overture Maps Foundation, created by AWS, Microsoft, Meta and TomTom, released the latest version of their interoperable and open map data, including 3D shapes;
An overview of how the NASA-ISRO SAR satellite mission NISAR will be used to map the seismic zones of the Himalayas.

4. Click-Worthy Stuff: Check These Out 🔗

This webinar (on Jan 25) from Airbus Intelligence with speakers from SkyWatch, Intelinair and 3VGeomatics that I will be moderating;
This film from Deloitte (in which I am one of the many speakers) on the future of space tech and how it creates value for non-space organisations;
This webinar series from the Food and Agriculture Organization of the UN on EO for agriculture;
The Satellite Applications Plan 2023-2027 (Le Plan d’Applications Satellitaires) from the French Ministry of Ecological Transition;
This paper proposing a roadmap for how EO can be integrated into the global stocktake process;
This cool video from Planet presenting Queryable Earth, a ChatGPT-like generative AI tool to make EO ‘searchable, conversational and context-aware.’

One Discussion Point

Analysis, thoughts, and insights on developments in EO

5. Quick Thoughts on Announcements from the Planet Explore Conference

I attended and spoke at the Planet’s Explore Conference in Washington DC last week. It was a well-put-together event, even bigger than I thought it was going to be. As expected, Planet made some announcements, which Kevin Weil, the company’s President of Product & Business summarised neatly on Twitter.

While there were some important updates on what the acquisition of Sinergise means, the upcoming Pelican and Tanager constellations, the launch of the Planet Startup Program, and the important announcement of Analysis-Ready PlanetScope Data, making it more readily usable for the customers, the big one, according to me was regarding the evolution of the Planetary Variables - belonging to what I call, the category of ‘usable, useful EO data products.’

A ‘usable, useful EO data product’ is anything that any interested user can start using off-the-shelf without really worrying about what the underlying technology is and how it works1 - one that any software engineer or data scientist can start working with, without knowing or understanding EO!

Planet announced four of them: Roads and Building Detection, Soil Water Content, Biomass Proxy, and Land Surface Temperature, available via APIs with three more Forest Structure, Forest Carbon, and Vegetation Encroachment to come soon. The former group (of 4) leverages technology developed by VanderSat, Planet’s acquisition in 2021 while the latter group (of 3) will leverage technology from Salo Sciences, Planet’s acquisition in 2022.

Why I’m Excited

I will be honest - I am excited about this. I have been looking forward to having more data products in EO, like in the world of weather. Take any weather product - temperature, wind speed, humidity etc., they have two characteristics:

It is a generic data product that is intuitive and ready to use via an API. One just needs to know how to use an API and why they are using it. The rest is taken care of while being scalable;
It makes complex look simple - these weather data products are a result of the most advanced data assimilation techniques in the world from several sensors, from multiple sources. But it is made to look easy!

The EO industry was badly in need of such an intermediary layer between raw EO data (pixels) and functional EO-derived applications (insights). I think Planetary Variables (PVs) fill that gap very well, especially how easy it is for literally anyone to start deriving value from EO, specifically for commercial use cases. These PVs are, in my vision, the perfect example of products that will be built by tens of thousands of EO experts to be used by hundreds of millions of end-users. In fact, PVs have already started to create considerable value.

Further, the introduction of PVs as APIs means the end-user is now abstracted away from the complexity of EO pricing. A data product delivered via an API makes the business model simple and extremely scalable2. Those who are happy with the results can continue to just use what Planet provides, while those who would like to improve them can do so by building on top of what is already been done - potentially ensuring more data sales for Planet and perhaps, other EO companies that Planet might partner with.

A win-win-win situation for Planet, the users and the EO industry.

What I Hope For

While I am obviously looking forward to how these PVs facilitate a more straightforward adoption of EO, especially for end-users who do not have remote sensing expertise, I have some questions and concerns. Instead of being pessimistic and providing criticisms, I want to express them as hopes:

Transparency & Openness

Anything that is a black box will be and will need to be questioned in EO, as the results provided by PVs will be used to make decisions with socioeconomic and environmental impacts.

How does the Soil Water Content API work? What datasets were used to calculate Land Surface Temperature? What methodologies were used to estimate Biomass Proxy?

Therefore, I hope the degree of openness always remains high, from today until the end, whether in the form of peer-reviewed science (which VanderSat has done before) or otherwise through partnerships with research institutions.

Collaboration & Partnerships

Some potential customers might be alarmed at Planet’s move down the value chain. To be fair, they have been testing the waters on analytics for a while, but this might make some question whether Planet is actually competing with them. Customers wouldn’t want that and certainly, Planet wouldn’t want that. This is why it might be the perfect opportunity to strengthen partnerships and leverage the domain expertise of those who actually operate in those market verticals to make the PVs more effective, making it a good outing for both parties. So, I hope strong collaboration ensues (we have seen some proof of that).

Accessibility & Standardisation

How easy will it be to access and try one of the PVs? Will they be available only on a Planet-owned platform (hint: that might be from Sinergise)? In reality, the more the number of relevant users that can access and use them, the better feedback that Planet gets to improve them and perhaps, develop further ones. But, given the complexity of these data products3, there might be reason for caution and Planet might be right in pursuing an incremental release of these products. I hope that Planet will work with enabling organisations like the Radiant Earth Foundation to create inclusive and diverse access points while making sure that organisations like the Open Geospatial Consortium are involved in making the data products standardised and certified.4

TL;DR: Planetary Variables are a step in the right direction both for Planet and for the EO industry, provided that the implementation is transparent, the execution is collaborative and the approach is standardised, making the products accessible for all. The jury is still out!

PS. No podcast episode this week - I took a break. Back next week!

Until next time,

Aravind

While some users might not care about the underlying technology, datasets or methodologies, some will certainly care a lot. But, the core idea is that adoption is made easier by simplifying the complexity while retaining the option to look under the hood if needed.

https://hbr.org/2021/04/apis-arent-just-for-tech-companies

Without going into too many details, it is hard to assume one API with one standard methodology will work exactly the same way everywhere in the world - models might need to be continuously fine-tuned, ground truth data globally might need to be collected for continuous validation and all of this based on fundamental EO data that continues to evolve, as Planet and space agencies around the world continue to launch more satellites. Simply put it is a tad more complex than Tesla pushing software updates to their vehicles over the air - not impossible, but needs diligence.

For more on this thinking, I encourage you to read these two pieces from Arjen Vrielink: Audit-Ready Data and Propagating the data graveyard.