such that every LiDAR posting is assigned a geographic position. To ensure
consistent and accurate DGPS positioning throughout the survey, the GPS basestations on the ground, of which there may be several, must be checked for their
broadcast quality. It is also typical that an expert operator accompanies the
instrument during flight to provide real-time monitoring of overall data accuracy
and quality, which is done in an effort to avoid large and unexpected errors
(Mohammadzadeh and Valadan Zoej 2008). The operator is also tasked with
monitoring the survey-coverage to avoid data gaps.
Upon landing, the raw data are promptly downloaded and converted to a format
readable by LiDAR processing software for further inspection and quality control.
At this stage, outliers may be filtered out to deliver a more realistic dataset and the
point-cloud is transformed to the desired projection system. Despite rigorous
calibration, inaccuracies arise from sources beyond the control of the operator.
These include the necessity to fly a survey at a sub-optimum altitude because of air
traffic control restrictions and turbulence, or because of unavoidable poor sea-state
conditions. To account fully for the influence of all potential error sources, the 1
sigma vertical accuracy of topographic LiDAR is generally quoted to be ±0.15 m
(Wozencraft 2003). For bathymetric data, this sigma is likely to be slightly greater.
Validation of a bathymetric LiDAR survey can be conducted against an
ancillary dataset of depth soundings, which is commonly acquired by sonar or
multi-beam (Chaps. 8–10). Optical depth-extraction from satellite imagery is not
of sufficient accuracy to serve as a validation set. LiDAR, like sonar, delivers spot
postings of depth and hence a direct spot-by-spot validation is not feasible since
the two surveys will not be exactly coincident. Hence, the vessel-acquired acoustic
data must be gridded prior to attempting a validation in which the LiDAR points
are compared to the grid. It is typical that multi-beam soundings are of higher
density than LiDAR. A further complication is that the ground-verified bathymetry
contains a tidal signal. By contrast, LiDAR postings may be collected with reference to the ellipsoid and are not tidally influenced. If the multi-beam data, like
the LiDAR, are acquired using post-processed kinematic techniques, the tidal
offset is mitigated. Another discrepancy between bathymetric LiDAR and sonar is
that, since laser profilers are designed primarily to provide data for hydrographic
charting, they typically return the depth of the highest object within the bounds of
the laser spot. For this reason, bathymetric LiDAR is referred to as being ‘shoal
biased’ (Quadros et al. 2008). Because the laser spot is broad in comparison to a
sonar sounding, the validation of submarine LiDAR with sonar is also problematic
when seabed terrain is rough. For dual-laser LiDAR systems that acquire data
across the intertidal, there may exist the opportunity for validation between the
marine and topographic datasets. Caution must be applied to this approach as the
size of the laser spot for a marine LiDAR is generally larger than that of a
topographic instrument. Because of the aforementioned problem with shoal biasing, the bathymetric LiDAR will return a terrain height above that of a topographic
laser scanner. Despite all of these concerns, acoustic data remains the best source
of validation data for verifying LiDAR accuracy.
5 LiDAR Overview
137
consistent and accurate DGPS positioning throughout the survey, the GPS basestations on the ground, of which there may be several, must be checked for their
broadcast quality. It is also typical that an expert operator accompanies the
instrument during flight to provide real-time monitoring of overall data accuracy
and quality, which is done in an effort to avoid large and unexpected errors
(Mohammadzadeh and Valadan Zoej 2008). The operator is also tasked with
monitoring the survey-coverage to avoid data gaps.
Upon landing, the raw data are promptly downloaded and converted to a format
readable by LiDAR processing software for further inspection and quality control.
At this stage, outliers may be filtered out to deliver a more realistic dataset and the
point-cloud is transformed to the desired projection system. Despite rigorous
calibration, inaccuracies arise from sources beyond the control of the operator.
These include the necessity to fly a survey at a sub-optimum altitude because of air
traffic control restrictions and turbulence, or because of unavoidable poor sea-state
conditions. To account fully for the influence of all potential error sources, the 1
sigma vertical accuracy of topographic LiDAR is generally quoted to be ±0.15 m
(Wozencraft 2003). For bathymetric data, this sigma is likely to be slightly greater.
Validation of a bathymetric LiDAR survey can be conducted against an
ancillary dataset of depth soundings, which is commonly acquired by sonar or
multi-beam (Chaps. 8–10). Optical depth-extraction from satellite imagery is not
of sufficient accuracy to serve as a validation set. LiDAR, like sonar, delivers spot
postings of depth and hence a direct spot-by-spot validation is not feasible since
the two surveys will not be exactly coincident. Hence, the vessel-acquired acoustic
data must be gridded prior to attempting a validation in which the LiDAR points
are compared to the grid. It is typical that multi-beam soundings are of higher
density than LiDAR. A further complication is that the ground-verified bathymetry
contains a tidal signal. By contrast, LiDAR postings may be collected with reference to the ellipsoid and are not tidally influenced. If the multi-beam data, like
the LiDAR, are acquired using post-processed kinematic techniques, the tidal
offset is mitigated. Another discrepancy between bathymetric LiDAR and sonar is
that, since laser profilers are designed primarily to provide data for hydrographic
charting, they typically return the depth of the highest object within the bounds of
the laser spot. For this reason, bathymetric LiDAR is referred to as being ‘shoal
biased’ (Quadros et al. 2008). Because the laser spot is broad in comparison to a
sonar sounding, the validation of submarine LiDAR with sonar is also problematic
when seabed terrain is rough. For dual-laser LiDAR systems that acquire data
across the intertidal, there may exist the opportunity for validation between the
marine and topographic datasets. Caution must be applied to this approach as the
size of the laser spot for a marine LiDAR is generally larger than that of a
topographic instrument. Because of the aforementioned problem with shoal biasing, the bathymetric LiDAR will return a terrain height above that of a topographic
laser scanner. Despite all of these concerns, acoustic data remains the best source
of validation data for verifying LiDAR accuracy.
5 LiDAR Overview
137