LiDAR Photogrammetry Point Clouds DEM / DTM / DSM Orthomosaic

LiDAR vs. Photogrammetry: Which One Your Project Actually Needs

April 1, 2026

Side-by-side comparison of a LiDAR point cloud and photogrammetry orthomosaic of the same terrain

If you've already read the broader three-way comparison of LiDAR, photogrammetry, and traditional survey, this post zooms in on the specific decision most clients are actually making: drone LiDAR or drone photogrammetry, for a given project. Instead of a parallel spec table, here are the project scenarios that drive the choice — with the underlying technical reason and what the deliverables look like in each case.

The single most important difference

Photogrammetry reconstructs a 3D surface from overlapping photos using Structure-from-Motion algorithms. It maps the top of whatever the camera can see. LiDAR fires laser pulses that can pass through gaps in vegetation and record multiple returns — canopy top, mid-canopy, ground — from a single pulse.

Every other technical difference (accuracy, point density, cost, processing time) ultimately routes back to that core distinction. If the bare-earth surface is what you need and there's anything between the camera and the ground — vegetation, brush, low crops, even tall grass — photogrammetry will give you the wrong answer with high precision.

Scenario-by-scenario picks

Open-terrain construction pad

A 30-acre bare-earth pad in Riverside County, freshly graded, no vegetation. Photogrammetry. With RTK GNSS and a handful of ground control points, photogrammetry achieves 3–5 cm RMSE on hard surfaces and the orthomosaic that comes out of the same flight is genuinely useful for documentation. LiDAR works too, but photogrammetry is usually the more economical choice here, and the visual product adds value.

Stockpile volumes at an aggregate yard

Photogrammetry. Stockpiles are bare, the surface texture is excellent for SfM feature matching, and the visual record helps with auditing. Photogrammetry is industry-standard practice for stockpile-volume work on open, unvegetated piles.

Hillside drainage study with chaparral cover

LiDAR. This is the canonical wrong-tool-for-the-job scenario for photogrammetry. The brush is uniform, the model has no idea where the actual ground is, and the resulting bare-earth DEM will be biased high by 30–80 cm depending on shrub density. Drainage design built on that DEM will undersize or misroute conveyance structures.

LiDAR penetrates the canopy. Even on dense coastal sage scrub, a sensor like the DJI Zenmuse L2 flown low with good sidelap can record enough ground returns through gaps in the brush to build a usable bare-earth DEM. Accuracy under vegetation is looser than on open ground, so it should be checked and reported separately, as Vegetated Vertical Accuracy.

Construction progress monitoring (active site)

Photogrammetry, usually. Active sites are typically large, mostly bare, and benefit enormously from the visual orthomosaic at 2–5 cm/pixel. Photogrammetry-derived terrain is plenty accurate for cut/fill verification on cleared ground. Switch to LiDAR if the site has retained vegetation, water on the ground, or recently disturbed material with poor visual texture.

Watershed planning with riparian corridors

LiDAR. Stream corridors are vegetated by definition, and the question — where does water actually go — depends on the bare-earth surface beneath the vegetation. Photogrammetry produces a confident-looking but wrong canopy model. LiDAR delivers the ground-classified point cloud that hydrologic and hydraulic models actually need. See our watershed development guide for the full methodology.

Powerline / utility corridor inspection

LiDAR. Corridor work needs both bare-earth terrain (clearance analysis) and accurate wire / structure geometry (sag, attachment points). Photogrammetry struggles with thin features like conductors. LiDAR captures wire positions directly as point returns and the bare ground beneath them — both in a single flight.

Solar farm or large agricultural site selection

Depends on the vegetation. Cleared land or post-harvest bare fields: photogrammetry is fine. Land with standing crops, perennial pasture, or scrub: LiDAR. The decision criterion is the same: does anything on the surface block the camera's view of the actual ground?

Post-wildfire erosion and runoff assessment

Mixed — both have a role. The burned landscape itself often becomes photogrammetry-friendly (vegetation removed by fire). But if the question is about pre-fire vs post-fire change, the pre-fire terrain may need to come from existing LiDAR or 3DEP coverage, and the post-fire follow-up flights can usually be photogrammetric if access is unrestricted.

Aerial view of farm fields, vineyards and a farmhouse below rugged mountains

Boundary or legal survey work

Neither. Boundary work requires a licensed surveyor with a total station or RTK rover producing a sealed record. See the three-way comparison post for the role of traditional ground survey.

What you actually get from each (and why it matters)

From photogrammetry

A 3D mesh and dense point cloud derived from overlapping images, typically 200+ points per square meter on surfaces with good texture.
A high-resolution orthomosaic at 2–10 cm per pixel (this is the visual product — a flat, georeferenced aerial photo of the site).
A digital surface model (DSM) — the top-of-canopy / top-of-roof / top-of-stockpile surface.
A digital terrain model (DTM / bare earth) only if there's enough bare ground to filter — on a vegetated site, this is unreliable.

From LiDAR

A classified point cloud in ASPRS LAS/LAZ format with returns sorted into ground, vegetation tiers, buildings, water, and noise. 100–400+ pts/m² typical.
A bare-earth DEM at 10–50 cm resolution, derived from ground-classified returns. This is the workhorse deliverable for any hydrologic or grading workflow.
Optionally: digital surface model, canopy height model, intensity raster, contour lines, slope/aspect rasters.
A formal accuracy report against ASPRS Edition 2 (2024) with NVA and VVA values.

Note that photogrammetry doesn't naturally produce a usable bare-earth model in vegetated terrain. It's not a deliverable-quality issue — it's a fundamental sensor limitation.

Quick numbers

Typical figures for each method. The accuracy for your project should come from its own RMSE report against independent checkpoints:

Photogrammetry (drone, RTK + GCP)

Vertical RMSE: 3–10 cm on hard surfaces under controlled conditions
Horizontal RMSE: 2–5 cm
Orthomosaic GSD: 2–10 cm/pixel
Ground penetration: zero

Worker in a high-visibility jacket planting seedlings in a tilled row

LiDAR (DJI Zenmuse L2 with PPK)

Pulse rate: 240k points/sec single return, 1.2M points/sec multi-return
Vertical accuracy: 2 cm-class on hard, open ground, verified against ground control with an RMSE report; under canopy, reported separately as Vegetated Vertical Accuracy
Horizontal RMSE: 1–3 cm with RTK/PPK
Point density: 100–400+ pts/m² typical
Ground penetration: yes (multiple returns)

The rule of thumb

If you can stand at the project and look at exposed bare ground across most of the site, photogrammetry is probably the right answer and will cost less. If you can stand at the project and see vegetation, brush, low crops, or anything else between the air and the ground, LiDAR is the right answer for the terrain product.

There's a corollary that surprises people: if the site is fully bare but the deliverable needs the most defensible accuracy report — a regulatory submittal, a high-stakes design — LiDAR still wins because QL2-grade LiDAR is FEMA's current minimum for floodplain mapping work, and the formal ASPRS reporting framework is more rigorous on bare earth too.

How cost fits into the choice

Our 2026 cost guide covers what drives the cost of each method. The short version: photogrammetry is usually the more economical choice because the hardware costs less and acquisition is faster. The real question is whether that saving is worth the limits on terrain accuracy and vegetation penetration, and in vegetated SoCal terrain the answer is often no.