Docv1.0 / 2026
Rangeλ 400 - 2500 nm
StatusRefCal · testing
●   Calibration infrastructure / Earth observation

Making Earth observation data comparable.

SpectraWorks builds the calibration layer that makes spectral data from different sensors comparable, traceable, and defensible.

See what we do -> Contact us
400
450
500
550
600
650
700
Visible spectrum, nm - extended into SWIR for hyperspectral instruments
§ 01
The problem

The problem is not a lack of Earth observation data.
It is confidence in the measurement.

Spectral data is increasingly used to support decisions in climate, compliance, agriculture, mining, and environmental monitoring. When measurements come from different sensors, platforms, and processing chains, comparability becomes the hard part.

Without a stable calibration reference, different datasets can appear precise while remaining difficult to compare. That uncertainty does not stay in the dataset - it lands on the person who has to defend the result.

§ 02
The technology
RefCal

A calibration reference for cross-sensor spectral data.

RefCal is SpectraWorks' calibration transfer technology. It is the reference point that makes cross-sensor spectral measurements comparable.

With RefCal, spectral data from different sources can be compared, traced, and defended.

Try a visualisation of RefCal ->
RefCal runs as an automated system inside analysis workflows. The demonstration is an interactive view of what it computes for one example area.

§ 03
Your archive

One deposit, many readings.

Your archive holds the same ground seen by different satellites on different dates, and the readings do not agree. Even flagship EO programmes like Sentinel require months of cross-calibration in-orbit to align their measurements, despite sensors built to the same specification. RefCal applies the same principle to every measurement - without the dedicated tandem campaign.

Fig. 1 / One site, several readings Illustrative · 3 sensors · 2 dates
  • Sentinel-2A
  • Sentinel-2C
  • Landsat 9
First date
First date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.262, Sentinel-2C centred at 0.271, Landsat 9 centred at 0.249; each curve has its own spread and they do not line up.0.200.240.280.32Share of sunlight reflected in band B12
First date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.262, Sentinel-2C centred at 0.271, Landsat 9 centred at 0.249; each curve has its own spread and they do not line up.0.200.220.240.260.280.300.320.34Share of sunlight reflected in band B12
First date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.262, Sentinel-2C centred at 0.271, Landsat 9 centred at 0.249; each curve has its own spread and they do not line up.0.200.220.240.260.280.300.320.34Share of sunlight reflected in band B12
First date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.262, Sentinel-2C centred at 0.271, Landsat 9 centred at 0.249; each curve has its own spread and they do not line up.0.200.220.240.260.280.300.320.34Share of sunlight reflected in band B12
Second date
Second date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.281, Sentinel-2C centred at 0.273, Landsat 9 centred at 0.298; each curve has its own spread and they do not line up.0.200.240.280.32Share of sunlight reflected in band B12
Second date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.281, Sentinel-2C centred at 0.273, Landsat 9 centred at 0.298; each curve has its own spread and they do not line up.0.200.220.240.260.280.300.320.34Share of sunlight reflected in band B12
Second date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.281, Sentinel-2C centred at 0.273, Landsat 9 centred at 0.298; each curve has its own spread and they do not line up.0.200.220.240.260.280.300.320.34Share of sunlight reflected in band B12
Second date: likely B12 reflectance at one location from three sensors (illustrative)Illustrative. Sentinel-2A centred at 0.281, Sentinel-2C centred at 0.273, Landsat 9 centred at 0.298; each curve has its own spread and they do not line up.0.200.220.240.260.280.300.320.34Share of sunlight reflected in band B12

The same location recorded by Sentinel-2A, Sentinel-2C and Landsat 9 on two dates. Each reading has its own spread, and the spreads do not line up.

§ 04
Where it matters

Reliable spectral data changes the quality of decisions.

SpectraWorks works on the calibration layer for decisions where measurement confidence matters more than visual impression. Our initial applications focus on geological analysis for mining, resource extraction and rare earth elements.

A · 01

Mining & resource extraction

Spectral signals can guide prioritisation across vast tenements, from base metals to rare earth elements - but only when the data is calibrated well enough to support confident interpretation.
"Is this anomaly real, or instrument drift?"
A · 02

Carbon & climate

When emissions, land-use change, or vegetation baselines are inferred from remote sensing, the credibility of the result depends on the reliability of the measurement underneath it.
"Has the signal changed, or has the sensor changed?"
A · 03

Compliance & supply chains

Deforestation, agriculture, and sourcing claims depend on data that can be compared across time, regions, and providers - and stand up under audit.
"Can this be defended to a regulator or auditor?"
§ 05
Notes & commentary
From the blog

Calibration, uncertainty, and Earth observation.

Original commentary when a calibration story breaks, curated reading from the literature, and technical notes on the measurement layer underneath Earth observation.

All posts ->
● Technical 06 Jul 2026 · 7 min

Sentinel-3 recalibrated a decade of ocean colour. That is the system working.

On 30 June 2026 EUMETSAT began reprocessing a decade of Sentinel-3 ocean-colour data onto updated calibration. Every prior chlorophyll number moves. Almost nobody noticed. The mechanism behind the quiet correction, and what it says about the sensors that never get one.
Read the post ->
§ Contact

Can your EO data stand up to scrutiny?

If your work depends on comparable spectral measurements, cross-sensor reliability, or defensible outputs, get in touch.

Your message is emailed to info@spectraworks.nl. See the privacy policy.