LiDAR or photogrammetry: which one your site actually needs

Choose by site condition, not by price sheet. If the ground you need is under tree cover, fly LiDAR. If the answer depends on what a surface looks like rather than where it sits, fly photogrammetry. On the Gulf Coast most parcels ask both questions at once, which is why we collect both on one mobilization.

How each one gets its points

Photogrammetry works the way your eyes do. It finds the same feature in a stack of overlapping photographs, figures out where the camera was for each one, and solves for where that feature must be in space. Very fast, very patient stereo vision.

LiDAR measures. The sensor fires a pulse, times the return, writes down a point, and does it again several hundred thousand times a second while the aircraft flies a planned line. Nothing is inferred. The sensor does not have to recognize anything to record it.

Everything below follows from that difference.

Where photogrammetry stops working

Under canopy

To solve a point on the ground, photogrammetry has to see that point in several photographs. Under a live oak or a stand of planted pine it never sees the ground at all. You get an extremely detailed model of the top of the trees, and a surface built from it is wrong by the height of the canopy.

A LiDAR pulse can find a gap in the foliage, hit dirt, and come back. Not every pulse does. Ground return density under heavy cover is always lower than it is in the open, so canopy work gets flown slower with tighter line spacing, and the processing report should tell you what the ground density actually was. Ask for that number. A provider who will not give it to you has not looked.

On surfaces with no texture

Photogrammetry needs something to match on. Fresh white sand gives it almost nothing. So does a smooth concrete slab, a single color roof membrane, and calm water. What comes back is noise, holes, or a surface that looks smooth because the software invented it.

The laser does not care whether a surface has visual texture. It cares whether the surface sends the pulse back. Open water is the one place both technologies give up, because water either absorbs the return or scatters it somewhere the sensor is not.

In shadow

Under a building overhang, or on the north elevation of a coastal tower at the wrong hour, photography loses detail. LiDAR brings its own light. A shadowed surface returns as well as a lit one.

What photogrammetry does that LiDAR cannot

None of that makes photogrammetry the lesser tool. On open, hard, well textured ground it resolves detail finer than most aerial LiDAR sensors will, at a fraction of the sensor cost.

It also produces the one thing a point cloud cannot: readable surface condition. A crack pattern. A rust stain running down from a fastener. Where the coating failed and where it did not. A point cloud tells you where a surface is. An orthomosaic tells you what state it is in.

There is a second, less technical advantage. A board member can look at a colorized model and understand it in four seconds. Nobody outside this trade has ever looked at a classified point cloud and understood it.

Site by site

Why most of our sites get both

The question is usually wrong. Very few sites are all canopy or all pavement. A coastal parcel is beach, dune vegetation, parking, structure and scrub inside the same boundary, and a wooded parcel two miles inland has a cleared building pad in the middle of it.

So we fly both, tied to the same ground control. The LiDAR carries the geometry. The imagery carries the color and the condition. Because they came off one flight and one control frame, they agree with each other, and colorizing the cloud from the imagery gives you a single dataset that answers both questions.

Adding the second sensor to a flight you are already paying to mobilize costs very little. Finding out three weeks later that we collected the wrong one costs a mobilization and a schedule.

Five questions worth asking any provider

1. What will ground return density be under the canopy on this site, not on an open field?
2. How is the vertical being checked, and will I see the residuals?
3. Is classification being reviewed by a person, or accepted from an automatic filter?
4. What happens over the pond, and will the voids be marked rather than interpolated?
5. If this site turns out to need the other sensor, what does that cost me?

Answers to those five tell you more about a capture provider than any accuracy figure in a proposal.