In the first blog in this series, we defined what “science-grade” Earth observation data actually means. It is not a marketing label, but a set of conditions: calibrated measurements, validated accuracy, consistency over time, and the ability to compare data across geographies and systems with known uncertainty.
The next question is more practical. If this is what science-grade requires, how do you build a system that can actually deliver it?
Science-grade data isn’t something you fix in processing. Differences in viewing geometry, atmosphere, and sensor behavior show up directly in the data. If they’re not controlled at capture, they carry through and compound over time. This is why science-grade data has to be built at the system level.
The EarthDaily constellation is designed around these principles.
EarthDaily’s constellation is designed as a measurement system rather than an imaging system. At a high level, it combines:
This combination of spectral depth, spatial resolution, and consistent acquisition allows the system to support measurement over time, rather than isolated observations.
Science-grade data depends on consistency. Measurements change with viewing angle, illumination, and acquisition timing. If those aren’t controlled, the variation carries into the data.
EarthDaily handles this at the point of capture, using stable, repeatable observation conditions:
In practical terms, this means the same location is observed under the same conditions – same angle, same geometry, and consistent time of day – on every revisit. Changes in the data therefore reflect changes on the ground, not differences in observation.
Science-grade systems measure physical properties of the Earth. Pixel values must correspond to quantities such as reflectance or temperature, not just visual brightness. Without that, data remains an observation rather than a usable signal.
EarthDaily’s system is structured around calibrated signals across approximately 22 spectral bands, spanning visible, near-infrared, shortwave infrared, and thermal regions. This allows vegetation condition, moisture, and temperature to be measured consistently over time.
Spatial detail is also treated as a measurement problem. The system is designed as a true 5 m measurement, rather than relying on sharpening to enhance apparent resolution. This ensures that the detail present in the data reflects what the sensor actually measures.
Combined with a swath of approximately 240 km, the system maintains consistency across both fine detail and large geographic areas.
Measurements drift over time. Without continuous calibration, that drift carries through the data, especially across sensors and time-series.
EarthDaily treats calibration as an ongoing operational process. Measurement stability is maintained through:
These calibrations are traceable to physical reference standards, maintaining comparability across instruments and over time.
Stability is maintained through continuous calibration and validation across global reference sites. The system is cross-calibrated daily against eight established science missions, including Landsat and GOES, with continuous monitoring and correction of drift. Measurements remain comparable across time and systems.
Science-grade data depends on correcting for atmospheric and environmental effects. That is not possible without the right spectral information.
The system uses 22 spectral bands across visible, near-infrared, shortwave infrared, and thermal regions. This is not about adding more bands. It is about being able to isolate the signal from the conditions it is captured in.
Visible imagery alone is limited. It reflects how the surface appears under specific conditions. Shortwave infrared and thermal bands carry different information -- moisture, material properties, temperature – that is less dependent on those conditions and more directly tied to physical change.
This is important because most of the variation in Earth observation data comes from atmosphere, illumination, and surface complexity. Without sufficient spectral depth, those effects get mixed into the signal.
With the right spectral coverage, they can be separated and corrected. Changes in vegetation condition, surface moisture, and temperature can be measured consistently over time.
A single high-resolution image can show detail. It cannot reliably show change. Science-grade data depends on the ability to compare measurements over time. That requires consistency across:
If measurements vary due to observation conditions rather than real-world change, analysis breaks down.
Long-running missions like Landsat and Sentinel-2 matter because their measurements hold up over time. Applying that same level of consistency to daily, global coverage is the real challenge.
In traditional Earth observation, the "heavy lifting" happens after the data is downloaded. Users often spend more time correcting for atmospheric haze, sensor drift, and shifting viewing angles than they do performing actual analysis.
By controlling these variables at the point of capture, the EarthDaily system delivers data that is effectively Analysis Ready. This shifts the focus from data preparation to operational results:
This engineering approach ensures that the constellation meets the CEOS Analysis Ready Data (ARD) framework. By adhering to these global standards, the system delivers data that is radiometrically consistent and geometrically aligned, making it immediately usable for large-scale monitoring and AI workflows without the need for additional normalization.
Science-grade Earth observation is defined by the ability to measure change reliably over time.
As Earth observation moves into operational use, the question is simple: does the signal hold up over time? That depends on how the system is built – how observations are made, calibrated, and processed.
And that is what turns Earth observation from imagery into measurement.