PRECISION FOR THE FUTUREWhat 3DEP and the quality levels are
The 3D Elevation Program (3DEP) is the U.S. Geological Survey’s national program to collect consistent lidar over the United States. Its technical rulebook is the Lidar Base Specification (LBS), which defines quality levels (QL0, QL1, QL2, QL3): bundles of pulse density and accuracy requirements a collection must meet.
The quality levels have become shorthand far beyond federal work. State programs, counties, engineering RFPs and utility corridors routinely say “QL1” or “QL2” because the numbers behind them are published and testable.
QL0 to QL3 compared
| Quality level | Pulse density | Pulse spacing | Vertical accuracy (NVA, RMSEV) | Smooth-surface precision | Swath overlap difference | DEM cell size |
|---|---|---|---|---|---|---|
| QL0 | ≥ 8.0 pls/m² | ≤ 0.35 m | ≤ 5.0 cm | ≤ 3 cm | ≤ 4 cm | 0.5 m (1 ft) |
| QL1 | ≥ 8.0 pls/m² | ≤ 0.35 m | ≤ 10.0 cm | ≤ 6 cm | ≤ 8 cm | 0.5 m (1 ft) |
| QL2 | ≥ 2.0 pls/m² | ≤ 0.71 m | ≤ 10.0 cm | ≤ 6 cm | ≤ 8 cm | 1 m (2 ft) |
| QL3 | ≥ 0.5 pls/m² | ≤ 1.41 m | ≤ 20.0 cm | ≤ 12 cm | ≤ 16 cm | 2 m (5 ft) |
Density and spacing are aggregate nominal values (all swaths combined). Smooth-surface precision and swath overlap difference are relative accuracy measures (RMSDZ); the vertical accuracy is absolute, tested against independent checkpoints in non-vegetated terrain.
What changed in the accuracy numbers
Since the 2024 revision the specification follows ASPRS Positional Accuracy Standards, Edition 2. Accuracy is stated as RMSE only: the older 95% confidence figures (such as 19.6 cm NVA for QL2) are gone. Non-vegetated vertical accuracy (NVA) must meet the table; vegetated vertical accuracy (VVA) is now reported rather than pass/fail. The checkpoint survey’s own error is included in the reported accuracy, and the report gives NVA and VVA for both the point cloud and the DEM.
Which quality level fits which job
- QL2 is the 3DEP baseline: county-wide terrain, floodplain mapping, regional hydrology and planning. It is the level most public lidar is collected at.
- QL1 keeps QL2’s vertical accuracy but at four times the density, so features are better defined: building footprints, forestry, corridor and transportation studies, urban areas.
- QL0 tightens accuracy as well as density: engineering-grade surfaces, detailed drainage design, critical infrastructure.
- QL3 is sparse and coarser; it is mostly encountered in older or legacy data.
Density and accuracy are different things. A dense cloud defines small features better, but its vertical accuracy depends on calibration, boresight, ground control and the GNSS/IMU trajectory. Check both numbers in the report, not just the points per square meter.
For site-scale engineering, a drone or terrestrial survey may beat any aerial quality level; see our guide to drone survey accuracy and ground control.
What a 3DEP-style deliverable includes
- Classified point cloud in LAS 1.4 with the required classes (see ASPRS classification codes), tiled, with a complete coordinate reference system in the header.
- Bare-earth DEM at the quality level’s cell size, hydro-flattened so lakes are flat and rivers run downhill, using breaklines.
- Breaklines for the hydro-flattening, as vector data.
- Accuracy reporting: NVA and VVA for the point cloud and the DEM, tested against independent checkpoints (calibration control may not be reused as checkpoints), plus the relative accuracy measures.
- Metadata: project and collection reports, flight logs, calibration and control reports, and FGDC-compliant metadata.
Where QL projects usually fail QA
- Swath misalignment between flight lines that shows up as ridges in the DEM.
- Ground misclassified under dense vegetation, which shows up as poor vegetated vertical accuracy (VVA) in the report.
- Hydro-flattening gaps: water bodies that are not flat, or streams that do not flow downhill through bridges and culverts.
- Bridge decks left in ground, damming the DEM.
- Mixed or missing vertical datum and geoid information in the headers and metadata.