LiDAR vs Photogrammetry: Which to Use for Surveying and Mining

Lidar measures distance directly. The scanner sends brief pulses of laser light; the travel time of each reflection gives the range, and GNSS and an inertial navigation system give the scanner's position and orientation, so every reflection becomes a 3D point. One pulse can produce several returns — from the canopy, from branches and from the ground below — and the LAS format records which return each point was.

Photogrammetry measures indirectly. Overlapping photographs of the same ground are matched by structure-from-motion processing, which recovers where each photo was taken and then the 3D position of every matched feature. It measures what the camera sees — texture under light — and gives colour for free.

This is the biggest practical difference. Lidar pulses pass through gaps in the canopy, so some of them reach the ground; classifying those returns as ground is what produces a bare-earth DEM. Photogrammetry sees only the top surface: under continuous canopy it measures the leaves, and no processing can recover ground it never saw. The USGS map accuracy standards noted the same limit decades ago: areas of dense woodland are an exception, because aerial photographs cannot provide the detail needed there.

On bare sites — open pits, quarries, stockpile yards, earthworks — the difference mostly disappears, which is why photogrammetry dominates those surveys.

Photogrammetry needs texture. Uniform surfaces such as fresh asphalt, snow, sand or still water give the matching little to hold on to and come out noisy or with holes; moving objects and deep shadows cause trouble too.

Lidar needs a reflection. Water returns little or nothing to a topographic laser — which is why lidar DEMs have their water bodies flattened with breaklines — very dark or wet materials return weakly, and any error in the scanner's position or calibration goes straight into every point.

Both can deliver centimetre-level heights on a well-controlled survey, and both can be far out on a badly controlled one. For photogrammetry, accuracy follows ground sample distance, image overlap, ground control or RTK/PPK positioning and surface texture. For lidar it follows the trajectory solution, the calibration between scanner and inertial unit, and the range accuracy of the scanner. Judge either by independent checkpoints, not by the processing report.

Lidar delivers a point cloud, usually as LAS or LAZ, with intensity and — if a camera flew too — colour; it is classified into ground, vegetation and buildings, and bare-earth DTMs and surface models are derived from it. Photogrammetry delivers a dense point cloud, a surface model, an orthomosaic and often a textured mesh. A photogrammetry point cloud can be classified too, but under vegetation its ground points are simply missing.

For open pits, quarries and stockpile yards, photogrammetry usually gives the monthly survey everything it needs — a surface, volumes and an orthophoto to draw on — with less expensive equipment. Choose lidar where vegetation hides the ground: pre-strip surveys of forested ground, haul road corridors through bush, rehabilitation areas and drainage studies on vegetated slopes. Where both are available, many teams fly photogrammetry routinely and lidar where the canopy demands it.

STREAM does not process raw photos or lidar trajectories; it works with their outputs. It opens LAS and LAZ from either method, streams large files in chunks, and builds a bare-earth DEM from ground-classified points, a surface model from all points and an orthophoto from point colour. DEMs and orthophotos from photogrammetry load directly as GeoTIFF. From there, volumes, contours, cross-sections, slope and change between surveys work the same whichever sensor flew.