Turn laser returns into structural evidence

ACTION

1. State the biological or terrain question.
2. Define horizontal support and vertical reference.
3. Name the field quantity, if any, used for validation.
4. Predeclare places where returns may be absent or unreliable.

INPUT

• Scientific question
• AOI
• Target support
• Validation plan

OUTPUT

• Measurement statement
• Support specification
• Non-claims

CHECK

• The target is not simply ‘LiDAR height’.
• The support matches the intended ecological comparison.

ACTION

1. Record file version and point format.
2. Verify horizontal and vertical CRS, units and datum/geoid information.
3. Record acquisition date, sensor, nominal pulse spacing or density and classification scheme.
4. Inventory tiles, extents, overlaps and missing files.

INPUT

• LAS/LAZ files
• Metadata
• Collection report
• Tile index

OUTPUT

• Point-cloud inventory
• CRS/units record
• Coverage map

CHECK

• Coordinates fall in a plausible range.
• Vertical units and reference are explicit.
• Tile count matches the index.

ACTION

1. Map point count or density by cell.
2. Separate zero-density gaps from water or masked areas.
3. Inspect scan angle and source/flightline identifiers.
4. Examine overlap areas for vertical disagreement.

INPUT

• Source points
• AOI
• Flightline/source IDs

OUTPUT

• Density raster
• Gap mask
• Overlap QA

CHECK

• Density is reported as a distribution, not one project-wide average.
• Striping, gaps and edge effects are investigated.

ACTION

1. Inspect elevation distributions and vertical profiles.
2. Apply documented range, isolation or neighborhood checks.
3. Use classification/withheld flags rather than deleting source points.
4. Review bridges, wires, birds and high vegetation before calling them noise.

INPUT

• Point cloud
• Existing flags/classes
• Neighborhood parameters

OUTPUT

• Flagged working points
• Outlier report
• Unchanged source copy

CHECK

• Every filter parameter is recorded.
• Removed/withheld points can be recovered.
• Tall valid structures were not clipped by a global maximum.

ACTION

1. Inspect ground and non-ground classes in cross-section.
2. Check slopes, banks, low vegetation, buildings and water edges.
3. Correct or flag systematic errors.
4. Keep return number and classification conceptually separate.

INPUT

• Flagged points
• Classification codes
• Reference terrain context

OUTPUT

• Accepted ground subset
• Classification QA map
• Correction log

CHECK

• Ground points form a plausible surface.
• Vegetation has not been absorbed into ground.
• First/last return is not treated as a land-cover class.

ACTION

1. Choose cell size from point spacing, terrain complexity and purpose.
2. Interpolate ground returns with a documented method.
3. Retain void and extrapolation masks.
4. Compare the DTM with independent open-ground checkpoints.

INPUT

• Accepted ground points
• Grid definition
• Checkpoints

OUTPUT

• DTM
• Interpolation/void mask
• Vertical-accuracy report

CHECK

• RMSE and signed bias are reported with sample count.
• Terrain edges and sparse areas are reviewed.

ACTION

1. Interpolate ground elevation at each accepted point.
2. Calculate height_above_ground = z − ground_z.
3. Inspect negative and implausibly high normalized values.
4. Keep absolute z and normalized height as separate attributes.

INPUT

• Accepted points
• Validated terrain surface

OUTPUT

• Height-normalized point cloud
• Normalization QA table

CHECK

• Ground returns cluster near zero.
• Negative heights are explained or flagged.
• No vertical datum conversion is hidden in the subtraction.

ACTION

1. Choose cell or plot support before calculation.
2. Calculate count, maximum, percentiles and cover-above-threshold only where meaningful.
3. Record minimum point count and empty-cell behavior.
4. Create DSM/CHM with a documented gridding method when required.

INPUT

• Normalized points
• Grid or plot polygons
• Metric definitions

OUTPUT

• Structural-metric table
• CHM or DSM where justified
• Support mask

CHECK

• Metric units and denominator are explicit.
• Outlier sensitivity is tested.
• CHM = surface estimate − terrain estimate is labelled derived.

ACTION

1. Match field and LiDAR acquisition timing as closely as the claim requires.
2. Align field locations and plot footprints.
3. Compare height metrics with field measurements using bias, error and scatter.
4. Stratify residuals by vegetation structure and point density when possible.

INPUT

• Held-out field measurements
• Derived metrics
• Support and timing metadata

OUTPUT

• Validation table
• Residual plots
• Applicability statement

CHECK

• Field height protocol and LiDAR metric describe compatible quantities.
• Validation plots were not used to tune filters.

ACTION

1. Export accepted/flagged LAZ, DTM, CHM and metrics in open formats.
2. Save CRS, vertical reference, grid, filter, classification and interpolation records.
3. Include density, void, accuracy and applicability layers.
4. Write supported claims and explicit non-claims.

INPUT

• Accepted points
• Derived surfaces/metrics
• QA and validation records

OUTPUT

• LAZ point package
• GeoTIFF/COG rasters
• Metric table
• Manifest and report

CHECK

• Every raster aligns to the declared grid.
• Point and raster products retain units and vertical reference.
• Files open outside the production software.