The Global Centre for Risk and Innovation (GCRI)
Satellite images become evidence only after calibration, atmospheric correction, georeferencing, cloud masking and validation. Long records need consistency across sensors and decades. This page sets out how raw observations become analysis-ready data and validated products, how accuracy is assessed and reported, and what evidence serves the users who rely on maps and time series built from space.
The interactive map needs JavaScript; the layer list and sources are below.
Layers 3
Earth observation, for this page
Keyboard, with the map focused
Hold Ctrl or ⌘ and scroll to zoom
Latest available
NASA GIBS / EOSDIS · ASTER GDEM (METI, NASA) · Blue Marble (NASA) · © OpenStreetMap contributors (ODbL) · Natural Earth
Loading the latest NASA imagery.
Drag to move. Ctrl or ⌘ and scroll, pinch or the buttons to zoom; hover for the values under the pointer; reset returns to the region.
For spatial temporal evidence, in GCRI's spatial-intelligence reading, the map opens on true-colour imagery, land cover and night lights with true colour from NASA GIBS, framed on the Arctic; every other NASA layer, live feed and forecast waits in the layer list.Satellite and model products as NASA publishes them, each with its own date; indicators, not assessments of any place; claim ceiling INTEGRATED.
Layers and dates
Satellite data accuracy is how closely products derived from Earth observation match reality, established through calibration, correction, geolocation and validation against independent reference data. Analysis-ready data and harmonised multi-sensor records make long time series reliable, with accuracy reported per class and with intervals.
The working screen behind this subject: its structure, terms and states. Names are placeholders and values illustrative.
Spatial temporal evidence · Flood basin AOI · Activation ACT-531
next 6 passes
| P-1 | Radar | n/a | Processed |
| P-2 | Optical | 79 | Cloudy |
| P-3 | Radar | n/a | Downlinking |
| P-4 | Optical | 22 | Scheduled |
Ground truth, resolution and uncertainty. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
Radiometric and atmospheric correction convert recorded signals into surface reflectance or backscatter. Errors in correction propagate into every derived product.
Images must align precisely with each other and with maps. Misregistration of even a pixel creates false change along edges.
Long time series combine sensors with different characteristics. Harmonisation is needed so that change reflects the ground, not the instrument.
Classified maps and estimates must be checked against independent reference data, and accuracy reported with confidence intervals.
Each term as this page uses it. Where a standard-setting body defines the term, the wording follows that body.
| Term | Definition |
|---|---|
| Atmospheric correction | Removing the effects of the atmosphere to retrieve surface reflectance from satellite signals. |
| User's accuracy | The probability that a mapped class actually matches the class on the ground. |
| Harmonisation | Adjusting data from different sensors so they can be combined consistently. |
Each indicator carries its unit, its cadence, the kind of source behind it and the evidence label a value would carry. No values are shown: a programme supplies them under its own authority.
| Indicator | Unit | Cadence | Source type | Label |
|---|---|---|---|---|
| Overall map accuracy | percent with confidence interval | per product | Accuracy assessment against reference data | ESTIMATED |
| Class-specific user's and producer's accuracy | percent | per class | Accuracy assessment | ESTIMATED |
| Geolocation error | metres | per product | Geometric validation | OBSERVED |
| Products meeting analysis-ready data specification | count | per assessment | CEOS ARD assessments | OFFICIAL |
| Reference samples per class | count | per assessment | Validation design | DECLARED |
| Time series gaps | percent of expected observations missing | per location | Archive analysis | OBSERVED |
The topic as a computational graph: named sources feed the methods, an evidence review gates the outputs, and the decisions stay with the authorities that hold them.
Fig. 02Nodes carry their type, two metric rows and a state; connectors show what feeds what; dashed frames name the owners; the timeline is the cadence. Design of the pipeline, not a record of a run.
Named public datasets and institutional records, each with its licence class. Sources by agreement are used only under their terms and never republished.
| Source | What it gives | Licence class |
|---|---|---|
| CEOS Analysis Ready Data specifications and assessments | Processing requirements and product status | Public, CEOS terms |
| Harmonized Landsat and Sentinel-2 (NASA) | Consistent surface reflectance from both missions | Open, NASA data policy |
| Good practice guidance on accuracy assessment, such as Olofsson and others | Methods for estimating area and accuracy | Licence per publication |
| Digital Earth Africa and Digital Earth Australia | Analysis-ready data cubes and validated products | Open, CC BY 4.0 |
| CEOS Working Group on Calibration and Validation resources | Calibration and validation protocols and test sites | Public, CEOS terms |
Space and Earth Observation prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| Who decides | The question | Evidence this page serves |
|---|---|---|
| Mapping and statistics agencies | Whether EO-derived statistics meet quality needs | Accuracy assessments; quality statements |
| Data providers | How to process and document products | ARD pipelines; harmonisation methods |
| Users of EO maps | How much confidence to place in a product | Quality statements with intervals |
Evidence specialists ask what a product can support and with what confidence.
Error enters at every stage: sensor noise and calibration, atmospheric correction, geolocation, the algorithm that converts measurements into a variable, and the classification rules that turn continuous signals into categories. A pixel often mixes several land covers. Temporal sampling can miss short events, such as a flood peak that passes between overpasses. Validation data may come from places and seasons that differ from where the product is used.
A well-established finding in remote sensing is that simply counting classified pixels can give biased area estimates. Statistically rigorous approaches combine a map with a probability sample of reference observations to estimate area and its confidence interval. The same principle applies widely: evidence from space is strongest when paired with designed ground sampling.
Comparing products with independent reference data using sampling designs from spatial statistics. Reference data has its own errors.
Carrying error from inputs to outputs. Correlated errors are often ignored, understating uncertainty.
Field surveys, in-situ sensors and citizen observations. They are sparse and unevenly distributed.
Testing how results change with resolution using geospatial information systems. Aggregation can hide local extremes.
The current service level is R0: information and enquiries. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.