Interpretation limit
A search threshold does not guarantee every studied pixel is usable. Colour changes can reflect processing rather than ground changes.
Public sources / EU
View and compare satellite imagery and metadata over time. Retain collection, scene identifier and display settings to distinguish the acquisition from the exported rendering.

AI-generated illustration
Read this source in practice
Copernicus Browser combines collection, period, cloud coverage and visualisation choices. Filters select products; then check the useful area's visibility and displayed processing.
The reference documents the reading method. The example is fictional; this note does not attest a portal test today or a mission outcome.
On a small screen, scroll the table horizontally to read all columns.
| Scope | Observation | Interpretation |
|---|---|---|
| Whole product | 15 % | Below the 20% threshold |
| Area of interest | Cloud-covered | Local observation obstructed |
A search threshold does not guarantee every studied pixel is usable. Colour changes can reflect processing rather than ground changes.
Product, area, period, threshold, bands or visualisation and obstruction observed in the target area.
Consulted reference : Copernicus Data Space — About the Browser · Consulted on
Updated on :
Record sensor, acquisition date, product and processing. Two views of one satellite product are not independent acquisitions.
View and compare satellite imagery and metadata over time. Retain collection, scene identifier and display settings to distinguish the acquisition from the exported rendering.
Cloud cover, resolution, processing and acquisition date constrain interpretation.
Product identifier, sensor, acquisition date, area, resolution, cloud cover and processing applied.
Publication and acquisition dates differ; inspect clouds over the precise area of interest.
Access, download and API conditions depend on the service. Consult the official documentation and the resource terms.
Compare map landmarks, satellite acquisitions and observation layers with their dates.
Views with different resolutions and processing cannot be treated as identical photographs.
Compare sourcesFix area and dates, open USGS and Copernicus scenes, then use NASA Worldview to place the change in context.
Sensors, resolution, cloud cover and processing differ; similar colors do not prove identical ground conditions.
Compare sourcesIdentify the feature and coordinates, compare public layers and select a satellite acquisition for the relevant date.
Coordinates, map update and acquisition are different references; record their respective precision.
Compare sourcesInspect the exact resource and its terms. Being viewable does not authorise redistribution; retain the resource licence and restrictions instead of inferring them from the portal.
Record consultation date, covered period, last update and version separately. For revised data, retain the file or exact extraction reference.
The web page, export and API may differ in coverage or terms. Check pagination, filters, authentication, quotas and licence for the access method used.
Check each layer’s terms and required attribution. Inclusion in a viewer does not automatically give identical rights for every basemap and product.
Align area, acquisition, sensor, resolution and rendering. Check clouds and masks locally; different palettes or composites can create visual differences.
Verify the context of a public imageDOCUMENTED REFERENCE
MSI acquires four bands at 10 m, six at 20 m and three at 60 m. This table describes native acquisition, not the pixel size of any export. Spectral descriptions are indicative; exact responses depend on the instrument. The diagram compares the same 60 m square extent: 36 cells at 10 m, 9 at 20 m or 1 at 60 m.
| Reference | Value or form | What it identifies |
|---|---|---|
| B01 | 60 m | Coastal and aerosol band. |
| B02 | 10 m | Blue band. |
| B03 | 10 m | Green band. |
| B04 | 10 m | Red band. |
| B05 | 20 m | Red-edge band. |
| B06 | 20 m | Another red-edge band. |
| B07 | 20 m | Another red-edge band. |
| B08 | 10 m | Broad near-infrared band. |
| B8A | 20 m | Narrower near-infrared band: distinguish it from B08. |
| B09 | 60 m | Water-vapour absorption band. |
| B10 | 60 m | Cirrus band; not provided as surface reflectance in the L2A product described. |
| B11 | 20 m | Short-wave infrared band. |
| B12 | 20 m | Another short-wave infrared band. |
B01: native spacing 60 m; nominal square cell area 3600 m². Area ratio to a 10 m cell: 36.
Geometric calculation from documented spacing; neither a ground measurement nor a count of independent observations.
L1C provides top-of-atmosphere reflectance; L2A provides surface reflectance after atmospheric correction. Retain level, band, tile, date and processing version. Resampling a 20 m band to 10 m does not make it a 10 m acquisition.
No. Grid spacing guarantees neither object separation nor positional accuracy. Contrast, optical response, mixed pixels and processing also matter. The integrated comparison calculates only nominal square-grid cell areas.
Documentation consulted on .
DOCUMENTED REFERENCE
The Sentinel-2 L2A scene-classification mask contains codes 0 to 11. The technical names below follow the consulted Copernicus table. SCL is produced at 20 m and may be resampled; it is not a continuous land-cover measurement.
| Reference | Value or form | What it identifies |
|---|---|---|
| 0 | NO_DATA | No data: do not replace with zero reflectance. |
| 1 | SATURATED_OR_DEFECTIVE | Saturated or defective pixel. |
| 2 | CAST_SHADOWS | Cast shadow, distinct from cloud shadow; check processing version for an archive. |
| 3 | CLOUD_SHADOWS | Cloud shadow. |
| 4 | VEGETATION | Vegetation class. |
| 5 | NOT_VEGETATED | Non-vegetated class. |
| 6 | WATER | Water class. |
| 7 | UNCLASSIFIED | Unclassified pixel: distinct from class 0 with no data. |
| 8 | CLOUD_MEDIUM_PROBABILITY | Medium-probability cloud class. Code 8 does not mean 8%. |
| 9 | CLOUD_HIGH_PROBABILITY | High-probability cloud class. Code 9 does not mean 9%. |
| 10 | THIN_CIRRUS | Thin-cirrus class. |
| 11 | SNOW / ICE | Snow or ice class. |
Classification is algorithmic and can contain errors. Define excluded classes, retain the mask and baseline, and visually inspect the area. Averaging SCL codes does not yield an average probability.
Bilinear interpolation may create values that identify no class. Use a categorical method and document the rule; Copernicus describes mode resampling for the 60 m SCL output.
Documentation consulted on .
DEEPER READING / PRACTICE
A list indicates where to look for and what limits to keep in mind. It helps to prepare a consultation but does not replace the reading of the obtained document.
Updated on .
Select a precise question, open the source and raise the request, result and date. Compare the possibilities announced with the information actually available at the time of the consultation.
The availability of a service and its content may vary. The directory does not turn an absent, partial or ambiguous result into a definitive response.
The consultation report documentes the search path, document found, useful elements and reserves. The result is included in the file with its origin.