Remote Sensing
Asset Monitoring
Linear and dispersed infrastructure is expensive to inspect precisely because it is linear and dispersed. A transmission corridor, a pipeline right of way, a rail alignment, a portfolio of remote solar sites. The asset is thin and the area is enormous, so physical inspection is either infrequent, expensive, or both.
The failures that matter on these assets are usually visible from above and usually gradual: vegetation encroaching on a conductor, a structure appearing inside a right of way, erosion around a foundation, a site that has stopped progressing. Satellite and drone observation does not replace inspection. It tells the inspection where to go, which is worth considerably more than another patrol.
How the work runs
Four phases, and what comes out of each.
2 weeks
Establish the asset base and the failure modes
We start from the asset geometry you hold (often less accurate than assumed) and the specific failure modes worth detecting. Encroachment, vegetation intrusion, ground movement and progress verification are different detection problems with different sensors and cadences, and a monitoring programme that tries to cover all of them uniformly does none of them well.
Outputs
- Validated asset geometry with positional accuracy assessed
- Prioritised failure modes with their detection requirements
- Buffer and threshold definitions per asset class
2–4 weeks
Build the baseline
Change monitoring requires a defensible starting state. We build a baseline across the asset base (surface condition, vegetation height and proximity, structures present within buffers) so that subsequent detections are genuinely change rather than something that was always there. This is also where a surprising amount of pre-existing encroachment usually surfaces.
Outputs
- Baseline condition across the asset portfolio
- Pre-existing encroachment and intrusion register
- Risk ranking by segment against the defined thresholds
3–5 weeks
Set the detection running
Detection is automated on the revisit cycle with thresholds tuned to what actually warrants a visit. The default failure of these programmes is alert volume, a system that flags everything unusual gets ignored within a month. We tune to actionable exceptions and report the false-positive rate honestly, because the operations team will measure it whether we do or not.
Outputs
- Automated detection on the revisit cycle with tuned thresholds
- Alert routing into the existing work management system
- Measured false-positive rate with a tuning protocol
2–3 weeks
Close the loop with the field
Every alert that produces a site visit produces a ground observation, and those observations retrain the detection. Without this loop the system's accuracy is frozen at commissioning. We build the feedback path and the review cadence, then hand over.
Outputs
- Field verification workflow feeding observations back to the model
- Review cadence and retraining protocol
- Handover with documentation and operator training
What good looks like
The ranges we design toward.
5–12 days
Detection latency
Time from a change occurring to it being flagged, against a quarterly or annual patrol cycle.
40–70%
Inspection effort
Reduction in routine patrol once field visits are directed by detection rather than by schedule.
70–90%
Alert volume
Reduction from a first-pass detector to a tuned one, without losing genuine detections.
Deliverables
What you are left holding.
- 01Validated asset base with a documented baseline condition
- 02Pre-existing encroachment and intrusion register, risk ranked
- 03Running detection pipeline with alerts routed into your work management system
- 04Measured false-positive rate and a protocol for tuning it
- 05Field feedback loop so accuracy improves rather than decays
What an engagement looks like
A representative shape.
Illustrative. This is the shape of work in remote sensing, not an account of a specific client. We do not publish named results, and we would be wary of a firm that publishes them without the client in the room.
16 weeks
- 01Situation
An agri-processor was committing to procurement volumes months before it had any reliable view of planted acreage, relying on dealer and field-officer reporting whose historical error nobody had measured.
- 02What we did
Built a Sentinel-1 and Sentinel-2 time series across the sourcing districts, trained crop classification on collected ground truth, and ran the satellite estimate in parallel with the incumbent reporting for a full season.
- 03Where it landed
The satellite estimate was available several weeks earlier and diverged from field reporting by a consistent margin in two districts, which ground checks confirmed. The parallel season is what earned the commercial team's trust; the technology alone would not have.
Where this connects
Rarely a self-contained problem.
Decision Maps
Detections are only useful when the operations team can see them on a map alongside work orders and crew locations.
Supply Chain & AnalyticsSupply Chain Resilience
Monitoring critical supplier sites and logistics corridors from orbit is a cheap early-warning layer on a mapped dependency.
Questions
The ones we get asked.
What size of change can you actually detect?
It depends entirely on the imagery you are willing to pay for. Free Sentinel-2 at ten metres reliably detects new structures above roughly one hundred square metres and significant vegetation change; it will not see a single pole or a narrow track. Sub-metre commercial imagery sees both and costs orders of magnitude more per square kilometre. We size this against your actual failure modes, most encroachment that matters is large enough for free data, and paying for daily sub-metre across an entire network is rarely justified.
Can you measure stockpile or excavation volumes?
Yes, from stereo satellite imagery or drone photogrammetry, typically to within two to five per cent depending on surface and method. Drone is more accurate and requires site access and flight permissions; satellite stereo works anywhere but needs a suitable acquisition. For recurring volumetrics on an accessible site, drone is usually the better economics.
How do you avoid drowning the operations team in alerts?
By treating alert volume as the primary design constraint rather than an afterthought. Thresholds are set against what genuinely warrants a visit, detections are confidence-tiered so the team can work the top tier first, and the false-positive rate is measured and reported every cycle. A monitoring system nobody acts on is worse than none, because it also carries a licence fee.
Does this replace physical inspection?
No, and we would be wary of anyone selling it that way. It redirects inspection. The regulatory and safety-critical inspections still happen; what changes is that routine patrol stops being uniform across a network where risk is anything but, and the inspector arrives already knowing what they are looking for and where.
Insights
Related reading.
What satellites can and cannot tell you about a crop
Sentinel data is free, covers the country and revisits every few days. So why do most agricultural earth observation projects still end in a folder of attractive maps?
What CBAM actually asks of an exporter
It is usually described as a carbon tax at the EU border. For a producer it behaves much more like an evidence problem, and the exporters paying most are frequently the ones emitting least.
Closely related
More on Remote Sensing →Change Detection
What changed, where, when and by how much, across areas too large and time spans too long to inspect any other way.
Remote SensingCrop Intelligence
Acreage, sowing progress, crop health and yield estimates across a district or a portfolio, weeks before anyone can walk the fields.
Start a conversation
Talk to us about asset monitoring.
Most useful first conversations are thirty minutes and involve no proposal. Describe the problem you are looking at and we will tell you honestly whether it is one we can help with.
Or write directly to hello@shyenova.com
