Prototype, sample imagery is illustrative
by IntelliMass

Territory Intelligence Platform · BackRoadIntel for First Nations

Your territory. Your record. Your evidence.

Your territory's ground truth, held by the community.

Every drive your own people already make returns three layers at once — what the driver saw, what the road did under the wheels, and the video of the whole way through. A permanent record of conditions across your territory — roads and trails, corridors, sites, land — as they change over time. The community holds the record. Not the proponent, not the ministry.

A forest-service road threading a forested river valley in coastal British Columbia, mountains and mist beyond
A permanent record of conditions across the whole territory, held by the community.
Why it matters

When the only record is someone else’s.

A proponent documents its own work. A ministry documents its own roads. When the condition of a route or a site is in question, whoever holds the dated record holds the argument. Every drive your own people already make can build one that’s yours.

What the record gives you.

Community egress

Whether the routes out are passable, documented before fire season, not during it.

Assets under agreement

Roads, corridors and infrastructure covered by impact-benefit or revenue-sharing agreements — verify condition without fielding a crew. When a road under agreement is damaged, the before-condition is on the record, not in someone's recollection.

The territory, on the record

Roads, trails, sites and land across the whole territory, captured on every drive — what it looked like on that date, not what somebody reported later.

Advocacy evidence

A dated visual record for funders, proponents and government tables, so the ask is documentation, not description.

Reclamation & remediation

Whether a site was restored, and when. A dated visual record of reclamation obligations as they come due, held by the Nation rather than filed by the party that owes them.

High-risk roads and trails

Condition records for the routes people depend on in high-risk terrain — main roads, back roads and the trails that only your people use — before the season, not during the event.

This is where BackRoadIntel started, active deployments with communities across British Columbia.

Whose data it is

The community holds the record.

The record — the video, the sensor readings, the reports — belongs to the community that produced it. Ownership from the first drive, not something you request back later.

You own the data

The community owns 100% of the data collected in its territory — raw sensor readings, video and reports. It's held in the community's name, and available from the first record.

OCAP®-aligned by design

We reference the First Nations principles of OCAP® — Ownership, Control, Access and Possession. Ownership, Control and Access hold from the first record: the community owns the data, decides who sees it, and can take a complete copy at any time. Possession — the record physically on infrastructure the Nation controls — is the sovereign tier.

Sovereign tier — the Possession in OCAP®

Host the record on infrastructure the Nation owns: a cloud account in the Nation's own name, region and keys, or an Indigenous-owned data centre. Sovereignty is something you buy, not a promise about it.

OCAP® is a registered trademark of the First Nations Information Governance Centre (fnigc.ca). There is no OCAP® certification, so nobody — us included — can call themselves OCAP-certified. We align with the principles; the sovereign-hosting upgrade is real work, quoted per deployment.

In the field.

The same record, across the roads, trails and roadside assets you're responsible for, not just the road surface.

A community egress route on a steep forested mountainside above a river
Community egressWhether the routes out are passable
A single-lane wooden bridge over a river in coastal rainforest
Assets under agreementInfrastructure across the territory
A remote road tracing a broad forested valley toward snow-touched mountains
The territory, on the recordWhat it looked like, on that date
One drive, three layers

One pass. Three layers of feedback.

A drive your people were making anyway comes back as three records of the same moment, locked to the same metre and the same date. Not three programs — one pass, read three ways.

1Field reports from your drivers

The person at the wheel taps once and says what they see — a washout, a widowmaker over the road, a gate left open. Captured as hazards and action items, located to the metre, in the words of someone who knows the territory.what the driver saw

2Road health, measured

The hitch-mounted sensor scores the ride continuously — no one has to notice anything or write anything down. Every pass lands on the same road axis, so the same stretch lines up across dates and you can see it move.what the road did under the wheels

3Video, the whole way through

Continuous, dated, geolocated video of the route — not the frames somebody chose to photograph. The picture that answers "what did it actually look like on that date", anywhere on the drive.what it looked like

Any one of the three can be argued with. Together they are hard to argue with: the driver's account, an instrument that was not forming an opinion, and the footage — all from the same pass, at the same place, on the same day. That is the difference between a photo album and an evidence file.

Watch a drive

The three layers, playing together.

One pass down a forest service road, as it comes back. The route scores itself on the map as the vehicle moves, the video runs beside it, and the driver's notes drop in where they were spoken — turn the sound on and you hear them in his own voice, which is the whole point of the first layer. A real drive, start to finish.

backroadintel · the record, playing
A drive playing back: the scored route on the map, the road video beside it, and the driver's field reports listed underneath1:06 · real drive · sound on
The map, the video and the spoken notes are one capture — this is what a pass looks like when it comes back.

See an example road report →

Territorial intelligence

Not one road. The whole territory, on one map.

Reserve lands, treaty territory, the industrial shared-use zones, the regional assets and the places that keep causing trouble — drawn as zones, with every road through them scored and every field report pinned where it was spoken. A road belongs to a zone; a zone belongs to the Nation. Open any one and you get its drives, its video and its reports.

The project explorer: every road in the territory ranked and scored, with the zones drawn on the map and a drive playing beside them
The project explorer — every road the Nation drives, ranked, scored and tied to the zone it runs through.

This is the part a proponent cannot hand you and a ministry does not hold: one picture of the whole territory, built from drives the Nation made, sitting under the Nation's own names for the places. It is the difference between answering a question about one road and being able to answer a question about anywhere.

A live project in the platform. Roads are ranked by length × roughness × repeat visits; the strip under each is that road end to end, 250 m a segment. Zones come from the federal survey, keyed by reserve number.

Industrial remediation timeline

Their trucks. Your road. Your numbers.

A hauling season on a community access road, measured before, during and after. Every reading lands on the same road axis, so the windows stack and the same stretch lines up down the page. The road was fine in March, the loaded trucks doubled its roughness by May, and the remediation took most of that back out — except in one place. None of it was a measurement campaign; it is drives the Nation was making anyway.

The road on the map

The same segments, where they actually are. The haul reddens the whole line; the remediation takes it back to green — except at the creek crossing, where it does not. Each chip is the average speed over that 8 km stretch and the minutes it adds against before the trucks — the 2 km detail runs under each strip below.

before the trucks48 km/h49 km/h59 km/h
after the haul26 km/h+8.7 min26 km/h+8.7 min36 km/h+1.9 min
after remediation40 km/h+2.1 min39 km/h+2.5 min51 km/h+0.5 min
Community access road 18.8 km · 94 segments · 200 m each
05 km11 km · creek15 km18.8 km
pre-industrial traffic01 – 14 Mar · 14 passes · roughness p90 5 · worst 5 · 0 % at 7+
54km/h—53km/h—45km/h—43km/h—60km/h—49km/h—43km/h—47km/h—58km/h—61km/h—
50 km/h · 23 min end to end
Hauling begins · 15 March
under haul17 – 29 Mar · 12 passes · roughness p90 9 · worst 11 · 37 % at 7+
38km/h+1.0min35km/h+1.2min31km/h+1.2min29km/h+1.4min47km/h+0.5min32km/h+1.3min30km/h+1.2min32km/h+1.2min45km/h+0.6min52km/h+0.1min
35 km/h · 32 min · +10 min · 30 % slower
haul finished26 Apr – 09 May · 13 passes · roughness p90 10 · worst 11 · 56 % at 7+
29km/h+1.9min27km/h+2.1min24km/h+2.2min23km/h+2.4min35km/h+1.4min24km/h+2.5min23km/h+2.4min25km/h+2.3min35km/h+1.4min39km/h+0.5min
27 km/h · 42 min · +19 min · 46 % slower
Remediation · 12 May
after remediation18 – 31 May · 15 passes · roughness p90 7 · worst 9 · 12 % at 7+
44km/h+0.5min44km/h+0.4min37km/h+0.6min36km/h+0.6min52km/h+0.3min35km/h+1.0min35km/h+0.6min39km/h+0.6min50km/h+0.4min53km/h+0.1min
41 km/h · 28 min · +5 min · 18 % slower
Change · before the trucks → after remediation0 improved · 87 about the same · 7 worsened
-10km/h+0.5min-9km/h+0.4min-8km/h+0.6min-7km/h+0.6min-8km/h+0.3min-14km/h+1.0min-8km/h+0.6min-8km/h+0.6min-8km/h+0.4min-8km/h+0.1min
50 → 41 km/h · +5 min end to end · 18 % slower
Roughness · level 1 (smooth) → 11 (severe) has video Changebetter  ·  about the same  ·  worse Speed · km/h per 2 km block, over the minutes it adds (on the change row, km/h lost)
Hover or click a 200 m segment — on a map or on any strip — to compare that same stretch across all three windows.

Basemap © OpenStreetMap contributors. The road line and its condition readings are constructed for this example; in the platform these are live maps that pan and zoom together.

Read across the windows: the road was in good shape before the trucks, roughness p90 5. Under haul it reached 9; by the time the trucks stopped it was 10 — double where it began. The remediation brought almost all of it back — except the creek crossing: one continuous 1.2 km run from km 10.6 to 11.8, and the only stretch anywhere on the road still reading worse than it did in March. That is the tell for a problem that is not a surface problem: water, or what is under the road. In minutes — the lane under each strip — a 23-minute drive became 42 under the worst of it, and still runs 28 after the repair: five minutes added to every trip anybody makes, each way. The rest was restoration the contractor owed; that one is a drainage repair, and now you can show why.

Illustrative — the shape of a real comparison, with constructed readings. The score is comparative: the same road against itself, window to window, so a stretch that degrades is visible. It is not calibrated, not a profilograph, and not a pavement condition index; an engineer's measurement is still an engineer's measurement.

Look at a record

The same road, three passes across the season.

Kilometre 4.5 of a community access road, from the same point on each pass, with the weather between them. When someone asks what the road was like before the trucks came through, this is the answer — dated, located, held by the community.

Illustration A community access road, dry and well graded, no defects.
Pass 1 · end of last season
L2Sound

Graded, drains clear, no defects.

Illustration The same road with potholes and runoff standing in the wheel tracks.
Pass 2 · through the season · +rain
L7Watch

Potholes and runoff in the tracks, shoulder starting to go. Grade it now, before it compounds.

Illustration The same road at fire-season peak: dry, loose and rock-strewn, with ruts and washboard across the running surface.
Pass 3 · fire-season peak · hot & dry
L9Severe

Degraded surface — loose rock, ruts and washboard, many hazards. Now it's a rebuild.

the baseline, dated — before the season, before the traffic the change, on the record the day it was driven what you bring to the table: a date and a view, not a description

Illustrative prototype. Production records open as hosted, view-only pages on a phone, no login.

Field annotations

Your record, spoken by the people who know the land.

The driver taps once and talks. It is the same device that is recording the video — the Dashcam unit in the kit — so the audio and the footage are one capture. The voice note is saved with the moment it was spoken — the point on the route, the time, the speed, the heading — then transcribed, titled and filed into the record at the place it happened. A form field gives you a category. A voice gives you what the person actually saw.

09:14·km 3.1·49.8204, -119.4918
0:06

“There’s a car in the ditch on this road.”

Field reportPhoto attached
09:22·km 5.8·49.8236, -119.4927
0:09

“It looks like smoke coming from the forest area on the right — but I’m not sure.”

Field reportUncertain — kept as spoken
09:31·km 6.2·49.8215, -119.4936
0:07

“Some rocks that just formed on the road here, we gotta take care of.”

RockModerate
09:47·km 9.4·49.8205, -119.4920
0:11

“There’s some equipment in the field on the left that looks out of place. Checking in on it.”

Field reportUncertain — kept as spoken

Two of those aren’t certain, and that is deliberate. “I’m not sure” is a legitimate entry. The record keeps the doubt exactly as it was spoken rather than rounding it into a category — and the next pass over that spot is what settles it.

1Spoken

Tap to talk on the Dashcam device — the same unit recording the video. Hands stay on the wheel, and the note is attached to the metre it was spoken at rather than typed up from memory hours later. A photo is tapped the same way.

2Transcribed

Every voice note becomes searchable text automatically, so a season of drives is something you can read through and search, not a pile of audio somebody has to sit and listen to.

3Cataloged

The transcript is read for what it describes — rock, water, debris, wildlife, a bridge, a washout — and dropped on the map as a titled field report you can filter and pull into a report.

In the vehicle, or out of it.

The Dashcam device records the same way on the windshield mount or in somebody's hand. A culvert you stop and walk down to, a slide you get out to look at, a site you cross on foot — spoken, located and filed into the record exactly like the kilometre you drove past.

Because the operator is a community member, what goes into the record is community knowledge — the person who knows that creek floods every spring, the one who remembers what the slope did the last bad year. That knowledge usually lives in somebody’s head and leaves when they do. Spoken into the record, it stays with the Nation.

Option Say it once, and it becomes a watch.

“Watch this area for rock slides going forward.” Sometimes the driver isn’t reporting today’s condition — they’re asking for the next drive to check something. With the watch option, a spoken note sets a watch from the driver’s seat, named and located as it’s said, then answered on every drive that covers it from then on. No form, no desk, nobody to ask. It is the fastest way a standing question should ever get created — out loud, at the spot, by the person who noticed.

Voice capture, automatic transcription and hazard tagging run in the Scout app and platform today. The four notes above are written as examples — the wording is ours, the format is what a real annotation looks like. Watches, including setting one by voice, are offered as an option on the program.

Watches

Set a watch. The record does the noticing.

A field report belongs to the drive it was tapped on. A watch is where a standing question goes instead — attached to a thing or a place, and answered on every drive that covers it. It doesn't have to be the road: a watch can be anything you drive past — a leaning tree, a culvert, a bridge, a gate, a slope, a site somebody has promised to restore.

One thing

A watched asset

A culvert, a bridge, a slide-prone slope, a gate. Named, located, and reported on every time a drive covers it.

A place

A watched zone

A stretch of route where a kind of problem keeps happening — a washout reach, a rockfall slope. Named for the problem, not the kilometre, because that's how the community already talks about it.

Watch — Leaning tree Watched asset Illustration Valley road · km 4.6 · set on drive 1
Driver's view of a conifer leaning slightly toward a line crossing above the road.
Drive 1
Logged

Slight lean toward the line. Nothing to do — but now it's a named thing with a question attached.

The same view a month later: the tree has leaned noticeably further toward the line.
Drive 2
Watch fired

The lean has increased. An email goes out the day the record is published.

The same view again: the crown now reaches across to the line.
Drive 3
Watch fired · act

The crown has reached the line. This is now a crew and a date, not a discussion.

The three frames above are an illustration — a worked example of how a watch behaves over successive drives, not a captured record from a community's corridor. The box doesn't move. A watch is set on a place, and what changes is what stands inside it. Nobody walked this line — a vehicle drove past three times on scheduled passes, and the record did the noticing.

WatchScopeAnswersLast event
Watch — Community egress routeZoneMain access road · km 0–14Any point dropping two levels or more.15 Jul · fired
Watch — River bridgeAssetValley road · km 6.2Deck and approach condition each drive.15 Jul · quiet
Watch — Common washout zoneZoneCanyon reach · km 8.5–9.7Standing water, a new washout flag, or an improvement of three levels or more.16 Jul · fired
Watch — Culvert at the creekAssetUpper road · km 11.3Blockage, scour, or a change in the flow through it.15 Jul · quiet
Watch — Rock slide zoneZoneBluff section · km 12.0–12.8New debris on the running surface after a storm or freeze-thaw.quiet since Apr
Watch — Reclamation site, north pitAssetHaul road · km 9.8Whether restoration has started, progressed or stalled since the last pass.quiet since May

Illustrative. In production, a fired watch emails the record's owner the day it's published.

What the Nation holds

A record of the territory nobody else has.

Pass after pass, season after season, every drive adds a dated, located layer: what the road was like, what the site looked like, what stood beside the route on that day. It builds into a time series of the territory, held in the Nation's name — one that no proponent, ministry or funder can produce for you, and none can dispute the date on.

Roads and trails are a good place to start — the drives are already happening. And it is not only the main routes: a trail your own people use, one that never sees a service truck or a ministry crew, is recorded the same way — dated, geolocated video of the route as it was on that day. Road-health scoring comes from the hitch-mounted sensor, so surface condition rides with the truck; the video record does not. But the record doesn’t have to start there, and it doesn’t stop there. Video alone, every drive, dated and located, is a record of the territory in its own right.

Your vehicles, your land, your record.

Your own people capture it

A capture kit in your own vehicles turns drives already happening into your records — trucks and trusted community members travelling your roads and trails, unlimited kilometres on one device. The skill stays in the community, and the record is built by the people who know the territory.

Unlimited kilometres, on your scheduleSign up for a year, or start smaller — a one-month or three-month program if the budget isn’t there yet. Contact us for rates.

Or, we capture it

For remote territory, or when nobody is making the drive: we bring the driver and the kit, capture on your interval — monthly, quarterly, or per project — and hand the record over. It's the Nation's from the first pass. No subscription.

Fixed fee per corridor

Either way it sits under one manager's signing authority, no procurement cycle. If your people are already on those routes, the kit is the way — the record and the capacity both stay home. If they aren't, we come to you.

If you need funding

Need funding? We work on it with you.

One kit and one driver covers a whole trail network — every road, route and access track your people already drive, on one device — and it is sized so a band office can start without a capital project. It also fits funding that already exists for this kind of work: non-structural mitigation, preparedness and community-held data. The Nation is the applicant. We work with your team and prepare most of the materials — the technical brief and the budget schedule — at no cost.

Federal · Indigenous Services Canada

Emergency Management Assistance Program (EMAP)

Non-structural mitigation and preparedness — documenting whether the routes out are passable is the kind of work this stream is meant for.

Federal · Crown-Indigenous Relations

First Nation Adapt Program

Assessing climate impacts on community infrastructure and emergency management.

Non-profit · CIRA

Net Good Grants

Community-controlled digital infrastructure — the record held in the Nation's own name.

Corporate · TELUS

Indigenous Communities Fund

Smaller grants for Indigenous-led projects — the size of a first kit and a season of drives.

Programs, intake windows and eligibility rules change, and what qualifies depends on the community and the project. Ask us what's open before you plan around any of them.

What we don't claim.

We don't do real-time alerting. This is a record of a drive, captured on your interval, not a live feed and not a service-level obligation.
We don't replace your inspections. Your inspectors and engineers still make the calls. We make sure there's a record of what the condition was.
We don't claim OCAP® certification. There is no such certification. We align with the OCAP® principles, and can host the record on infrastructure the Nation owns — that's a real deployment, not a compliance badge.

Evidence. Long term. Verifiable.

Three layers from every pass — what was measured, what was filmed, what was said — dated, located and held in the Nation's name. Years later it can still be checked against the drive it came from, by anyone you choose to show it to. The earlier the record starts, the more it is worth: tell us which route you would document first.

Sample analysis · constructed data

Industrial remediation timeline — sample record

Three months on 18.8 km of community access road — 1 March to 31 May, 54 passes — through a hauling season and the remediation that followed. Measured by the Nation's own Scouts, compared segment by segment along the road's own line.

RemediatedHigh confidence

The haul turned a 23-minute drive into a 42-minute one. The remediation gave back all but five minutes of it.

Before the trucks, not one reading on this road came back rough — zero out of 1,316. Under haul and after it, 649 of 1,222 did: more than half of every drive. After remediation it is 141 of 1,410, about one reading in ten. Average speed ran 50 km/h, then 27, and now 41.

Recommendation. The restoration obligation is substantially met on the running surface. It is not met between km 10.6 and 11.8 at the creek crossing — seven consecutive segments, and the only stretch on the road still reading worse than it did before the first loaded truck. At km 11.0 not one of the 14 drives before the trucks read rough; all 15 after the remediation do. That is a drainage repair, and it is now a dated, measured ask rather than an opinion.

Resolution

This comparison could have detected a change of 0.56 levels or more. The haul moved it 4.1.

Segment scatter

Standard deviation 1.48 levels across 94 paired segments — the spread when nothing has happened.

Coverage

94 of 94 segments measured in every window; 54 passes over the three months, about 14 a month.

Comparability

These windows were not driven at the same speed — 50 km/h before the trucks, 27 after the haul. That gap is the road forcing the change rather than a difference in how it was driven, but roughness is sensed through a moving vehicle, so the two are not independent: a slower pass over the same ground registers less. If anything that understates what the haul did.

Video

Every window carries linked video on at least 89 of 94 segments, so any reading can be looked at rather than argued about.

Method

200 m segment · statistic mode · trusted devices · Scout units only — driver-app phones are excluded by device class, so no phone reading reaches a road number.

The road, window by window

Each strip is the whole road, south to north, at roughly 200 m a segment, averaged across every pass in that window. On the change row, grey is no move at all and the palest tints are a one- or two-level drift — real, but inside what a single 200 m segment scatters by between passes, so not a finding on its own. A blue underline marks a segment with linked video, and under each strip is that window's average speed 2 km at a time, with the minutes it adds. Below, all 54 individual passes that sit behind these four windows are shown unaveraged.

05 km11 km · creek15 km18.8 km
pre-industrial traffic01 – 14 Mar · 14 passes · roughness p90 5 · worst 5 · 0 % at 7+
54km/h—53km/h—45km/h—43km/h—60km/h—49km/h—43km/h—47km/h—58km/h—61km/h—
50 km/h · 23 min end to end
Hauling begins · 15 March
under haul17 – 29 Mar · 12 passes · roughness p90 9 · worst 11 · 37 % at 7+
38km/h+1.0min35km/h+1.2min31km/h+1.2min29km/h+1.4min47km/h+0.5min32km/h+1.3min30km/h+1.2min32km/h+1.2min45km/h+0.6min52km/h+0.1min
35 km/h · 32 min · +10 min · 30 % slower
haul finished26 Apr – 09 May · 13 passes · roughness p90 10 · worst 11 · 56 % at 7+
29km/h+1.9min27km/h+2.1min24km/h+2.2min23km/h+2.4min35km/h+1.4min24km/h+2.5min23km/h+2.4min25km/h+2.3min35km/h+1.4min39km/h+0.5min
27 km/h · 42 min · +19 min · 46 % slower
Remediation · 12 May
after remediation18 – 31 May · 15 passes · roughness p90 7 · worst 9 · 12 % at 7+
44km/h+0.5min44km/h+0.4min37km/h+0.6min36km/h+0.6min52km/h+0.3min35km/h+1.0min35km/h+0.6min39km/h+0.6min50km/h+0.4min53km/h+0.1min
41 km/h · 28 min · +5 min · 18 % slower
Change · before the trucks → after remediation0 improved · 87 about the same · 7 worsened
-10km/h+0.5min-9km/h+0.4min-8km/h+0.6min-7km/h+0.6min-8km/h+0.3min-14km/h+1.0min-8km/h+0.6min-8km/h+0.6min-8km/h+0.4min-8km/h+0.1min
50 → 41 km/h · +5 min end to end · 18 % slower
Roughness · level 1 (smooth) → 11 (severe)has videoChangebetter  ·  about the same  ·  worseSpeed · km/h per 2 km block, over the minutes it adds (on the change row, km/h lost)
Every one of the 54 passesthe four windows above, unaveraged

The road runs left to right — the same 18.8 km as the strips above. Time runs down the page, oldest at the top, one row for every drive, and every drive covered the whole road. You can watch the haul arrive, spread and sit there; you can watch the remediation take it out again; and you can see the one column at the creek crossing that never comes back.

Fifty-four drives over three months, every one of them a trip somebody was making anyway. This is what a record looks like when nobody is running a measurement campaign — and it is a picture no single inspection could produce.

Weather between the windows

Thirteen weeks, 1 March to 31 May. The haul ran through the spring freshet — the worst time to put loaded trucks on an unpaved road, and the reason the creek crossing moved as far as it did.

Rain / melt
Freeze–thaw
MarAprMay

What more readings buy you

A single pass is not a guess — it is a true record of that road on that day, and it is evidence the moment it exists. What repetition buys is the ability to tell a change from the ordinary scatter between passes. Roughness is sensed through a moving vehicle, so no two passes land identically even on a road that did not move. The smallest change a comparison can resolve falls with the square root of the number of passes.

Smallest change this road can resolve
0.5123±3.371 pass±1.694±0.9712±0.5735levelsOne pass can only tell you something moved if it moved more than three levels. Twelve passes and a single level is readable. Thirty-five and you can see half a level — which is the difference between "something happened" and "the haul did this".
The same stretch, clustered by week
4567wk 1n=2wk 2n=3wk 3n=1wk 4n=4wk 5n=5wk 6n=3wk 7n=6wk 8n=5Each dot is one pass. The line is the week's mean, the band is its confidence. Early weeks sit on one or two readings and the band is wide — a dot outside it proves nothing. By week 7 the band is tight enough that the downward drift is the road, not the driving.

This is the argument for capture that rides along with drives people were already making: nobody schedules a measurement campaign, and the confidence accumulates for free. It is also why the verdict at the top of this report carries a resolution figure — a change smaller than the resolution is not a finding, however much it looks like one.

Where to look

The segments the haul moved most, read all the way through to the remediation. Four came back to exactly where they started — that is the restoration working, and it is why the verdict is what it is. Seven did not, and they are not scattered: they are one unbroken run from km 10.6 to 11.8, which is what a real local difference looks like as opposed to a measurement artefact. The video is how you tell those apart, so every row opens its clip.

Roughness level · 1 (smooth) → 11 (severe)Drives that read rough · level 7+Peak vertical acceleration · m/s²
AtBefore the trucksAfter the haulAfter remediationStill outstandingbefore → after remediationbefore → after remediation
2.40 km15100/14 → 0/150.8 → 0.8 0.0▶ clip
3.60 km48400/14 → 0/153.1 → 3.1 0.0▶ clip
4.80 km48400/14 → 0/153.1 → 3.1 0.0▶ clip
7.20 km510500/14 → 0/154.8 → 4.8 0.0▶ clip
10.60 km174+30/14 → 0/150.8 → 3.1 +2.3▶ clip
10.80 km3117+40/14 → 8/152.1 → 10.6 +8.5▶ clip
11.00 km5119+40/14 → 15/154.8 → 23.8 +19.0▶ clip
11.20 km5118+30/14 → 15/154.8 → 15.8 +11.1▶ clip
11.40 km4118+40/14 → 15/153.1 → 15.8 +12.7▶ clip
11.60 km4117+30/14 → 15/153.1 → 10.6 +7.4▶ clip
11.80 km3106+30/14 → 0/152.1 → 7.1 +5.0▶ clip

Drives that read rough counts the individual passes that came back at level 7 or worse on that segment — level 7 is the product's own threshold for rough. It is a count over the drives, not a restatement of the window average, which is why km 11.00 reading 15 of 15 after remediation says something the level alone does not. Peak vertical acceleration is what a level means in physical units: the midpoint of that level's band on the published log ramp, gravity removed. It is the scale rather than a second instrument — two segments on the same level necessarily show the same figure, which is why it sits beside the levels and not apart from them.

Illustrative. Readings are constructed to show the shape of a real comparison. Roughness is comparative — the same road against itself, window to window. It is not calibrated, not a profilograph, and not a pavement condition index; an engineer's measurement is still an engineer's measurement.