Newsletter · · 4 min read

Earth Observation Essentials: August 11, 2025

Transparency and Trust in the EO-AI Era

Welcome to the new format of the TerraWatch newsletter. A few quick housekeeping notes before we get into the real content.


📢 Reminder: The TerraWatch Newsletter Is Evolving

As announced, the TerraWatch newsletter is moving to a new model starting September 1.

To help you get familiar with what’s coming, this month you will be receiving previews of the new formats:


📈 EO Market Highlights

Major developments in EO

🤝 ISS STOXX, a financial service firm that provides market indices and ESG data for institutional investors, has acquired Sust Global, a geospatial risk analytics firm, to expand its climate-risk modelling capabilities.

💡
Why this matters: Strategic, value-adding use cases of EO will continue to see higher adoption in commercial sectors such as insurance and financial services.

This is another example of backward vertical integration, where a key market player acquires an EO-based service provider (typically its supplier) to bring geospatial capabilities in-house.

🛰 Satellogic has expanded its partnership with HEO, a non-Earth imaging data provider, highlighting the growing trend of EO companies entering the space object monitoring market (Maxar and BlackSky being the other ones).

🏭 The US Administration is planning to shut down two NASA missions part of the - Orbiting Carbon Observatory series (OCO-2 and OCO-3) - aimed at measuring carbon dioxide and plant health.

💡
A bit more context: Ending these missions would mean that the US would lose access to CO₂ monitoring data, which is critical for climate models and emissions accountability. Alternative data streams from ESA’s upcoming CO₂M and Japan’s GOSAT missions would partially fill the gaps, albeit with reduced temporal coverage.

💡 Insight Bytes

A quick dose of analysis from TerraWatch

Transparency and Trust in the EO-AI Era

We first created this graphic to capture something I felt was often missing in EO conversations. What we call “an insight” in the EO community is rarely just a single algorithm or a dataset. It is the result of multiple steps, each with its own assumptions, uncertainties, and human choices.

At the time, my aim was to show how many layers sit between raw satellite data and the final decision a user makes - from acquisition and processing to interpretation and delivery, with each stage have an impact on the final result.

Now, in the era of embeddings and foundation models, this separation feels even more important. The more these steps are wrapped inside large, powerful models, the harder it becomes to understand what’s happening under the hood.

The risk is that speed and automation start to replace transparency and explainability. And in EO, where insights often drive real-world decisions in areas like disaster response, environmental monitoring, or infrastructure management, that’s not a trade-off we can afford.

For me, this graphic is a reminder that even with AI, every step in the EO chain matters. As we embrace new tools, we need to ensure we don’t lose the ability to interrogate how an answer was reached. The value of EO doesn’t just lie in the final insight, but in the trust we can place in every step that got us there.


🔍 Recommended Reads

Interesting links to check out


🛰️ Scene from Space

One visual leveraging EO

Wildfire Smoke Blankets Canada and the US

Canada is experiencing an intense wildfire season, with smoke plumes severely impacting air quality both domestically and across the United States. This false-colour image from Copernicus Sentinel-3, captured on 3 August 2025, shows a massive smoke cloud drifting over Canada and the US.

Credit: European Union, Copernicus Sentinel-3 imagery

Until next time,

Aravind.

Read next