GeoGarmeux Spatial Industries — Insights

LiDAR vs. photogrammetry: when to fly which

2026-06-18T00:00:00Z

Ask which sensor is better and you will get a debate. Ask where each one fails and you get a decision. LiDAR pulses reach the ground through canopy gaps, which makes it the only serious option for bare-earth terrain under vegetation. Photogrammetry sees only what the camera sees — the top of the canopy, not the ground beneath it.

Photogrammetry earns its keep on open sites. Over bare ground, stockpiles, pavement, and structures, a well-flown photogrammetric survey delivers dense, colorized surface models at a fraction of LiDAR's sensor cost, and the orthophoto that comes with it is a deliverable in its own right. If the client needs to see the site, not just measure it, the camera is doing double duty.

Budget is the tiebreaker less often than people expect. The real costs are re-flights and rejected deliverables: photogrammetry flown over moving water, uniform snow, or dense vegetation produces holes and noise that no amount of processing rescues. Price the failure modes, not the flight.

Our rule of thumb after hundreds of missions: vegetation or engineering-grade ground surfaces mean LiDAR; open sites where imagery matters mean photogrammetry; corridors and complex jobs usually justify flying both, because the sensors cover each other's blind spots.

How utilities use EO data to get ahead of vegetation

2026-05-30T00:00:00Z

Fixed-cycle trimming is a bet that vegetation grows at the same rate everywhere. It does not. A five-year rotation over-trims slow corridors and under-trims fast ones, and the outage report always names a span that was two years from its scheduled visit.

Weekly satellite revisit changes the economics. Current constellations image most North American service territories every few days, at resolutions good enough to measure canopy proximity to conductors span by span. The workflow is a pipeline, not a project: imagery lands, a risk model scores every circuit against growth rate and outage history, and the trim plan re-ranks itself.

The organizational shift matters more than the sensors. Crews stop working a calendar and start working a ranked list, and the ranking is auditable — every score traces back to an image with a date on it. When a regulator asks why a corridor was trimmed in March instead of May, the answer is a map, not a memo.

Digital twins start with a survey, not a dashboard

2026-05-12T00:00:00Z

Most digital twin projects begin with a demo of the viewer and end with an argument about the data. The 3D model is the easy part; every vendor can spin a city on a screen. The hard part is that the model has to agree with the ground, and the ground was never captured to the accuracy the use case needs.

Start from the decision, not the visualization. If the twin exists to replace field inspections, it needs survey-grade capture of the assets being inspected — not a decorative mesh draped over old imagery. If it exists for planning, coarser capture is fine but the parcel and utility layers underneath must be authoritative. Accuracy requirements flow from the questions the twin must answer.

The second failure mode is staleness. A twin captured once is a souvenir; the value is in the update cycle. Before building anything, decide what gets re-captured, how often, and by which sensor — then size the budget for years two and three, not just the launch.

Teams that succeed treat the twin as a survey program with a good interface, not a visualization program with a data problem. That ordering is the whole trick.

What RTK networks actually buy you in the field

2026-04-21T00:00:00Z

Uncorrected GNSS puts you within a few meters of the truth, which is fine for finding a manhole and useless for setting one. Real-time kinematic corrections shrink that to centimeters by comparing the rover's signal against reference stations with precisely known positions, and a network of those stations means the crew never has to set up a local base.

The payoff shows up as absent rework. Utility locates that land on the correct side of the trench, as-builts that overlay design files without a manual shift, stakeout that machines can follow without an operator second-guessing it. One avoided re-mobilization typically covers a season of network subscription fees.

The honest caveat: RTK does not fix a bad datum, and centimeter hardware cannot rescue a workflow that mixes coordinate systems. Before buying accuracy, standardize the reference frame every crew and contractor delivers in. That decision is free and it is worth more than the receiver upgrade.

Open standards are the cheapest insurance in your GIS stack

2026-04-02T00:00:00Z

Nobody budgets for the migration out. Systems get selected on features, and five years later the real price appears: thousands of layers in a proprietary format, scripts that speak one vendor's API, and a quote for the export project that rivals the original license.

Open standards are how you cap that risk on day one. GeoJSON and GeoPackage for vector data, GeoTIFF for rasters, and OGC services — WMS, WFS, WMTS — for anything served over the wire. When every layer can leave the building in a format the next tool already reads, vendor negotiations change tone noticeably.

This is not an argument against commercial platforms; it is a procurement test for them. Ask two questions of any GIS you evaluate: what formats does it export without an add-on, and can another system consume its services without a license? Platforms confident in their value answer quickly.

Tracking construction progress from orbit, honestly

2026-03-14T00:00:00Z

A lender with forty active sites cannot walk them all, and monthly photo reports show whatever the contractor pointed the camera at. Satellite change detection offers the neutral version: the same sensor, the same angle, every site, every week. Earthwork phases are unmistakable from orbit — clearing, grading, and foundation work each leave signatures a model can flag reliably.

The limits deserve equal billing. Optical imagery cannot see through cloud, cannot measure vertical progress inside a structure, and resolves nothing about work quality. Treat orbital change detection as a screening layer that answers one question well: which sites changed, and which sites that should have changed did not.

The teams getting value pair the two scales deliberately. Satellites watch the whole portfolio and rank the anomalies; drones and field visits go where the ranking points. Nobody flies forty sites a week, and after the first quarter of doing this, nobody wants to.