What can and cannot satellite imagery show in a bank review?

An observation date is not today’s review date

A source records what was observed at acquisition. Its processing or publication date can be later, and the review may happen later still. Keep those dates separate. Two usable observations can show a difference between them without establishing exactly when the change happened.

For example, the Fischamend public record compares 20 August 2023 with 19 August 2025. It does not provide continuous monitoring across that interval or establish the present condition of the site.

Optical, radar and aerial evidence answer different questions

Optical satellite observations

Sentinel-2 measures reflected sunlight across 13 spectral bands at spatial resolutions of 10, 20 or 60 metres. The usable detail depends on the band and product. A large surface change may be distinguishable while a small feature or concealed work remains unresolved. Enlarging the displayed image does not create finer source detail. Source: Copernicus Sentinel-2 mission and instrument.

Cloud, shadow and missing or degraded pixels can limit a comparison. Sentinel-2 products include quality information such as scene classification and cloud masks; those indicators also have documented limitations. They require interpretation for the selected area and question. Source: Copernicus Sentinel-2 products and quality limitations.

Radar satellite observations (SAR)

Sentinel-1 uses synthetic aperture radar, an active microwave sensor with day-and-night, all-weather imaging capability. The returned signal depends on factors including surface geometry, roughness and moisture. It requires a different interpretation from an optical photograph. Source: Copernicus Sentinel-1 mission and SAR principles.

For a bank review, radar suitability must be established for the specific question. A radar change alone does not establish asset identity, engineering completion or contractual acceptance. Sensor capability is not a claim that CRI already operates a validated radar service for every site.

Aerial orthophotos

An aerial orthophoto is an aerial photograph corrected for geometric distortion from camera tilt and terrain. It can serve as map-like visual context. It is a different source from a satellite image, with its own acquisition date and resolution. Source: USGS explanation of orthophotos.

The historical aerial view on the CRI page is labelled for orientation. Its product-year reference is not a verified acquisition day, and it cannot confirm a fictional construction milestone.

Confidence belongs to a defined statement

In the Fischamend example, the stated HIGH confidence applies only to the large roofed-footprint change. It does not extend to completion, technical quality or payment eligibility. The record can therefore carry that confidence while still requiring documentary review.

Data Quality Uncertainty (DQU) keeps limitations of the source and its fitness for the review question visible: for example coverage, timing, resolution or an unverified boundary. It is distinct from a credit-risk score. These public explanations do not introduce a percentage-accuracy claim or a new scoring threshold.

Read the Evidence Pack status meanings. Evidence quality, an unresolved question and a human’s scoped acceptance are separate parts of a review.

When to stop interpreting and request other evidence

• The relevant area cannot be observed: record the gap. Do not infer that nothing changed.
• The boundary, date or comparison is unsuitable: clarify the scope or obtain a suitable source before making a site-specific claim.
• The question concerns hidden work, legal status or contractual meaning: request the relevant document or professional assessment.
• A reply answers only part of the question: retain its scope and exclusions; leave the rest open for the responsible reviewer.

For the intended CRI workflow, the stopping point follows the evidence needed for the question. More images cannot, by themselves, resolve a question that requires an acceptance record, a valuation or an expert inspection.

Discuss a specific review question

Bring the asset context, the available source types and the question a responsible reviewer needs to answer. An Evidence Readiness Check can examine whether the evidence and workflow are suitable for a bounded evaluation.