Network Rail Grades Satellite Data Asset by Asset

Network Rail’s earthworks portfolio runs to 190,000 assets: roughly 73,000 soil cuttings, 14,000 rock cuttings and 102,000 embankments, many of them Victorian, built before modern soil mechanics and geotechnical design standards existed. Ground Engineering set out on 18 September 2026 how satellite radar is now applied across that portfolio. Under the ASCEND programme, InSAR has been adopted at full British network scale, measuring ground movement to within a few millimetres and refreshing after every satellite pass, a weekly cadence no examination programme can match. The decision worth studying sits one step later: Network Rail and SatSense score how far each measurement deserves to be believed, cutting by cutting, before a geotechnical engineer reads it.

Satellite data without that score leaves the judgement with whoever opens the file, unrecorded. A scored feed makes reliability an attribute of the asset record, where it can be filtered, reported and defended in a renewals paper.

The Suitability Index Turns Four Physical Constraints Into One Number

The Earthwork InSAR Suitability Index (EISI), developed jointly by Network Rail and SatSense, combines four factors, each scored between 0 and 1 and averaged into a single figure held against the asset.

Factor	What constrains it	What a low score changes
Coverage	Vegetation scatters the radar between acquisitions; steep terrain causes layover and shadow; construction and agriculture alter the surface	Too few usable points, so the examination interval stands and evidence comes from LiDAR or track geometry
Data quality	Noise and consistency of each point against its own history and its neighbours	A first flagged movement raises a review and waits for repeat passes to agree
Sensitivity to movement direction	Satellites detect motion along the line of sight only, so orientation and slope angle relative to the viewing angle decide what is visible	Perpendicular movement may go unrecorded, so a quiet signal is weak evidence of stability
Sensitivity to movement rate	Brittle failures show little deformation before collapse, and rainfall washouts can progress faster than the revisit interval	The exposed failure mechanism sits outside what the satellite shows, and the control stays with drainage and examination

The fourth row decides how the programme is described internally. Network Rail’s framing is that InSAR is one line of evidence among several rather than an early warning system in its own right, and the index ships with training and user guides written to head off a false sense of security. Publishing the limits alongside the score is a governance step, and it is the part most condition-data programmes leave out.

The Score Reaches the Engineer Inside the Application They Already Open

SatSense delivers processed InSAR through an API into Network Rail’s internal GIS and into GyRO, the earthwork asset management application it is developing with JBA. Flagged movement, trends against a historical baseline and the suitability context arrive together, in the tool the engineer opens anyway. ASCEND places that feed alongside three others: biennial national aerial LiDAR surveys producing digital terrain models, train-borne track geometry read for embankment instability, and on-site examination.

Network-wide monitoring currently runs on C-band data from the Sentinel-1 constellation, to be supplemented by L-band data from the NASA-ISRO NISAR mission, whose longer wavelength penetrates vegetation canopy more effectively, with higher-resolution X-band used on a targeted basis. Coverage on a wooded cutting is therefore a moving figure, which is a further argument for scoring suitability per asset instead of quoting one accuracy number in the contract.

A feed handed over as a supplier portal login is read by the specialist who owns the login. A feed written into the system of record is read by whoever opens the asset, including the planner scheduling next month’s work.

Where a Suitability Score Sits in Maximo Manage

Operators taking a scored external feed into IBM Maximo Manage have four places to put it, and the choice decides whether the score survives contact with the work.

Hold the suitability value as a specification attribute on the asset classification, one attribute per data source, refreshed by the supplier through an Integration Framework enterprise service. Any report or query can then filter the estate by how much the feed is worth on each asset.
Land the movement measurement itself as a meter reading against the asset or location, with the condition monitoring point carrying the action limits and the job plan. The mechanics of that path, including reading type, rollover and the cron task that evaluates the limits, are set out in meter readings that keep condition-based PMs honest.
Route on two thresholds rather than one. Movement above the action limit on an asset with a strong suitability score generates a work order for engineering assessment. The same movement on a weakly scored asset generates a request for examination, because the evidence justifies a look.
Name the role that owns both thresholds, normally the geotechnical or asset engineer for the route, and put changes to them through the same change control as any other maintenance rule. Thresholds nobody owns drift until the generated work is ignored.

Reduced Examination Effort Has to Leave a Written Trail

Engineers still visit assets where direct investigation is required; the satellite decides which visits happen first. Reducing reliance on repeat on-site examinations is the stated aim of the SatSense contract, announced in September 2025, and it runs into a regulatory question. The Office of Rail and Road records in its 2025 to 2026 health and safety report that financial pressures affected the delivery of renewals and other planned work during the year, and that where activities are deferred or reduced, Network Rail must demonstrate that risks have been assessed and suitable mitigations are in place. The same report covers nearly 400 RM3 assessments, with Asset Management and Change Management among the areas that included some ad hoc ratings.

That sets the evidence standard for any regulated operator. A decision to extend an examination interval on satellite evidence holds up when the reading, the suitability score, the decision, the date and the named decision-maker sit on the asset record, in the same way that statutory inspection findings belong on the asset record rather than in a report library. A decision recorded in a programme slide is not evidence a regulator can audit.

Drainage Failures Sit Outside What Radar Can See

In November 2025 a passenger train derailed near Shap after striking material from a landslip caused by blocked drainage and heavy rainfall. The ORR notes that monitoring equipment had been installed at that high-risk location and did not prevent the incident, and its ongoing investigation is examining infrastructure management, maintenance arrangements and the effectiveness of monitoring systems and alarm responses. Drainage capacity, and whether a channel was built as designed, are not observable from orbit at any wavelength.

A Freedom of Information response obtained by New Civil Engineer in May 2024 records 147 land movements for the 2023 calendar year, of which 13 were identified by remote monitoring, against 31 found by track maintenance staff and 31 reported by train drivers. Those 13 came from deployed tiltmeters, four years into the £33M planned in Control Period 6 for remote monitoring, sensing and weather diagnostics on earthworks and drainage. Confidence in a trend builds over repeated passes, so the first year of a satellite feed largely builds the baseline later years read against.

The staffing question stays hard. A flagged movement is a control only where someone is contracted to review it within a defined window and holds the authority to impose a speed restriction or raise an examination. That review capacity competes with the examination backlog it was bought to relieve, and buying the data does not create it.

Position

The question to put to a satellite data supplier before signing is the suitability distribution across the operator’s own asset population. The proportion of assets that reach a usable score is the real coverage number, and on a vegetated or steeply cut estate it falls well short of the map. Contract on that proportion, on the API that writes the score into the system of record beside the reading, and on a named response window for a flagged movement. Operators who buy on those terms gain a defensible basis for where examination effort goes. Operators who buy coverage area gain a second dataset, held by a specialist, that the asset register never sees.