GIS, lidar, photogrammetry and drones

Traditional surveying measures one point at a time. Modern geomatics can also measure millions of points in minutes, from aircraft, drones, vehicles and tripod-mounted scanners, and it stores, analyzes and shares spatial data in geographic information systems. These technologies have changed how engineers see a site: instead of a few cross-sections, they work with complete three-dimensional models. This lesson introduces geographic information systems, photogrammetry and the ground sample distance, lidar and laser scanning, drone surveying and its regulations, and how the accuracy of mass data is checked.

Theo Martel runs this side of Coulee Geomatics. For the Coulee Ridge subdivision he flew the site with a drone to build an orthophoto and a surface model, combined the results with municipal utility records in the firm's GIS, and used the surface to compute earthwork volumes for comparison with the cross-section and grid methods of lesson 5.

Geographic information systems

A geographic information system (GIS) stores spatial data together with descriptive information and provides tools to query, analyze and map it. Spatial data come in two main forms. Vector data represent features as points, lines and polygons: a hydrant is a point, a water main is a line and a lot is a polygon. Each feature carries attributes, such as pipe material, diameter, installation date or lot owner, in a linked table. Raster data represent continuous surfaces or images as grids of cells, such as an orthophoto, a digital elevation model or a land cover classification.

GIS organizes data in layers that share a coordinate system, so that utilities, parcels, contours, zoning and aerial imagery can be overlaid and analyzed together. Typical engineering uses include finding all lots within a flood zone, tracing which valves isolate a broken water main, estimating the population served by a pump station, and checking design alternatives against environmental constraints. Municipalities maintain GIS databases of their infrastructure assets, and many publish open data portals from which engineers can download parcels, imagery and elevation data.

GIS data are only as good as their sources. A utility line digitized decades ago from a paper drawing may be several meters from its real position, while a survey-grade as-built may be good to a centimeter. Good GIS data carry metadata describing their source, date, coordinate system and accuracy, and surveyors and engineers must read it before using the data for design. On the Coulee Ridge project, the municipal water main location in the GIS was used for planning, but Grant's crew located the main in the field, with the utility owner's locate marks and test holes, before the design was finalized.

Photogrammetry

Photogrammetry is the science of making measurements from photographs. When the same ground point appears in two or more overlapping images taken from different positions, its three-dimensional position can be computed by intersecting the rays from the cameras, just as a total station intersection uses two directions. Modern photogrammetry uses structure-from-motion software, which automatically finds thousands of matching features in overlapping images, computes the camera positions and orientations, and builds a dense three-dimensional point cloud, a surface model and an orthophoto, which is a mosaic of images corrected so that it has a uniform scale like a map.

Photogrammetric surveys are planned with large overlaps between images: typically about 75 to 80 percent forward overlap along each flight line and 60 to 70 percent side overlap between lines. The resolution of the imagery is expressed as the ground sample distance (GSD), the size of one image pixel on the ground. It depends on the camera's sensor width, focal length and image width in pixels, and on the flying height above the ground:

As a rule of thumb, a well-controlled drone photogrammetry survey can achieve horizontal accuracy of about one to two times the GSD and vertical accuracy of about two to three times the GSD on open, hard surfaces, provided there is good ground control. Accuracy is much worse in tall grass, crops and shadows, and photogrammetry cannot see the ground under dense vegetation.

Ground control and checkpoints

Photogrammetric models must be tied to the project coordinate system. Traditionally this is done with ground control points: targets, often painted crosses or checkerboard panels, placed around the site and surveyed with GNSS or a total station before the flight. The software uses them to scale, orient and georeference the model. Drones with onboard RTK or post-processed kinematic (PPK) GNSS record accurate camera positions and need fewer control points, but at least a few are still recommended, especially to control the vertical.

Independent checkpoints are surveyed points that are not used in the processing. Comparing the model with the checkpoints gives an honest measure of accuracy, usually expressed as a root mean square error (RMSE). Accuracy standards, such as the American Society for Photogrammetry and Remote Sensing positional accuracy standards widely used in both countries, define accuracy classes in terms of RMSE and specify how many checkpoints are needed and how they should be distributed.

Lidar and laser scanning

Lidar, short for light detection and ranging, measures distances with laser pulses. Combined with precise positioning and orientation from GNSS and inertial measurement units, each returned pulse becomes a three-dimensional point. Airborne lidar flown from aircraft covers large areas and is the main source of modern regional elevation models; many provinces, states and the federal governments of both countries have released lidar elevation data. Mobile lidar mounted on vehicles maps road corridors at highway speeds. Terrestrial laser scanners on tripods capture buildings, bridges, plants and excavations in great detail, and many total stations now include scanning functions. Drone lidar brings airborne methods to small sites.

A key advantage of lidar over photogrammetry is that a single laser pulse can produce several returns as parts of it reflect off leaves, branches and finally the ground. The last returns under vegetation allow the bare ground to be mapped beneath tree canopy, which photogrammetry cannot do. Lidar points are classified, by software and by manual editing, into categories such as ground, low, medium and high vegetation, buildings and water. Lidar point density is expressed in points per square meter and ranges from a few points per square meter for regional mapping to hundreds or thousands for drone and mobile surveys.

Drone regulations

Drones used for surveying are regulated by the Federal Aviation Administration. Commercial operations of small drones, under 55 lb, are flown under the Part 107 rules, which require a remote pilot certificate, registration of the drone, broadcast Remote ID, operation within visual line of sight, generally no higher than 400 ft above the ground, and authorization before flying in controlled airspace near airports. Waivers allow some operations outside these limits. The FAA proposed a new Part 108 rule in 2025 to allow routine flights beyond visual line of sight; as of late 2026 the final rule was still in review. Operators must also respect state privacy and trespass laws, local ordinances and landowner permission for take-off and landing.

Beyond the regulations, drone surveys need practical planning: checking airspace and nearby airports, weather and wind limits, battery planning, a safe take-off and landing area, visual observers where needed, and communication with people on the site. On construction sites, drone flights must be coordinated with crane operations and site safety rules.

# Drone flight planning and checkpoint accuracy
import math

sensor_w_mm, focal_mm, img_w_px, img_h_px = 13.2, 8.8, 5472, 3648

def gsd(height_m):
    return sensor_w_mm * height_m / (focal_mm * img_w_px)

for h in (60, 73, 100, 120):
    g = gsd(h)
    across, along = g * img_w_px, g * img_h_px
    print(f"H {h:3d} m ({h / 0.3048:3.0f} ft): GSD {g * 100:.2f} cm, footprint {across:.0f} x {along:.0f} m, "
          f"line spacing {0.30 * across:.0f} m, photo spacing {0.20 * along:.0f} m")

# Number of flight lines and images for a 600 m x 450 m site at 100 m
g = gsd(100)
lines = math.ceil(600 / (0.30 * g * img_w_px)) + 1
photos_per_line = math.ceil(450 / (0.20 * g * img_h_px)) + 1
print(f"site 600 x 450 m: about {lines} lines x {photos_per_line} photos = {lines * photos_per_line} images")

# Checkpoint statistics
dz = [0.021, -0.034, 0.012, 0.028, -0.018, 0.040, -0.025, 0.009]
rmse = math.sqrt(sum(d * d for d in dz) / len(dz))
print(f"RMSEz = {rmse * 100:.1f} cm, mean error = {sum(dz) / len(dz) * 100:+.1f} cm, 95% ~ {1.96 * rmse * 100:.1f} cm")

H  60 m (197 ft): GSD 1.64 cm, footprint 90 x 60 m, line spacing 27 m, photo spacing 12 m
H  73 m (240 ft): GSD 2.00 cm, footprint 109 x 73 m, line spacing 33 m, photo spacing 15 m
H 100 m (328 ft): GSD 2.74 cm, footprint 150 x 100 m, line spacing 45 m, photo spacing 20 m
H 120 m (394 ft): GSD 3.29 cm, footprint 180 x 120 m, line spacing 54 m, photo spacing 24 m
site 600 x 450 m: about 15 lines x 24 photos = 360 images
RMSEz = 2.5 cm, mean error = +0.4 cm, 95% ~ 5.0 cm

Choosing a method

Each method has strengths. Total stations and GNSS give the highest accuracy on individual points and are essential for control, legal surveys, layout and checking. Drone photogrammetry is fast and economical for open sites, stockpiles and progress monitoring, and gives excellent imagery. Lidar penetrates vegetation, captures complex structures and works in poor light. Terrestrial scanning records existing buildings and plant rooms in detail for renovation design. GIS ties all of it together with records of ownership, utilities and assets. Most projects use a combination: on Coulee Ridge, total station and GNSS control, leveled benchmarks, a drone surface for earthwork, field shots of hard features and utilities, and GIS for planning and records.

Whatever the method, the deliverable must state its coordinate system, its accuracy and how that accuracy was verified. A beautiful three-dimensional model with an unknown datum or untested accuracy is a hazard, not an asset.