The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
179
full oscillating cycles, each cycle consisting of two scan lines
(Figure 2b). In other words, an oscillating mirror at 100 Hz scan
frequency generates 200 scan lines on the ground. Not knowing
this quantitative difference when comparing distinct scanners
with the “same” scan frequency may lead to misinterpretation
of system capabilities and miscalculation of point spacing for a
survey mission.
Scan field of view (FOV) versus scan rate (or scan
frequency): In the case of an oscillating mirror, two
parameters—maximum scan rate and maximum scan angle—
are not only interrelated but also inversely proportional to each
other. Their product determines the maximum scanner velocity
that a particular scanner can practically achieve, characterized
by the maximum scan product (SP). The maximum SP
represents the real physical limitation of an oscillating mirror
scanner and, in combination with the scan pattern driving signal,
determines the maximum load allowed for the scanner. Since
the maximum SP characterizes the maximum achievable
scanner velocity and simultaneously accounts for both the
highest scan rate and the maximum scan FOV, it also
determines the maximum possible area coverage rate for a lidar
system.
It was shown (Ussyshkin et al., 2008b) that the maximum scan
rate (or frequency) available for a particular lidar system may
have limited practical advantage if the maximum scan angle
available at this scan rate reduces the scanner FOV to
impractical limits. On the other hand, given an equivalent scan
pattern, a higher SP indicates a wider scan FOV available for
the maximum scan rate and consequently a scanner that can
operate at a higher scanner velocity to complete the job faster.
However, lidar system users should remember that SP values
calculated for different types of scan patterns are derived
dissimilarly and should never be compared directly as
counterparts.
In summary, for any type of oscillating mirror, regardless of
scan driving signal differences, the maximum scan rate (that is,
frequency) is always linked to the maximum scan angle
available for this frequency. That is why the seeming advantage
of large numbers on the specification sheet may not equate to
any actual benefit, and users should always examine the
numbers by considering real operational scenarios and practical
limitations.
3. ACHIEVABLE ACCURACY VERSUS
ACCURACY SPECIFICATIONS
Of particular importance are numbers on a specification sheet
characterizing lidar data accuracy. These numbers represent one
of the most important system specifications. However, these
numbers can be very misleading, if the context of the reference
conditions and deriving methodologies are not taken into
account. While instrument accuracy specifications are provided
by the manufacturers, translating the specification numbers to
real-world achievable accuracy is a challenge usually left to the
end user, and it has long been a subject of different
interpretations (Ussyshkin et al., 2006a). Moreover, without
widely accepted guidelines for deriving accuracy numbers, lidar
system manufacturers typically use different methodologies for
accuracy specifications.
Owing to the nature of lidar data collection, many factors affect
the real-world accuracy of lidar data, including extreme
operational parameters (such as a very wide scan FOV and very
high flying altitudes), strong variations in the target physical
properties (such as size, slope, and reflectivity), and so forth.
While some of these factors may be defined and described on a
specification sheet, not all of them can be accounted for even in
the most detailed document, and that is why the impact of some
of these factors on data accuracy is sometimes either ignored or
underestimated. However, it is very important for the user to
estimate the influence of these factors on achievable data
accuracy. We will give several examples showing the
relationships between unexpected or underestimated factors and
their impact on lidar data accuracy.
3.1 Dynamic Range of Intensities and Data Accuracy
Though usually interpreted as essential to intensity data and
image quality, the dynamic range of intensity that a lidar system
can accommodate (also known as the “intensity digitization
specification”) may be extremely important for the achievable
range data accuracy in surveys where strong variations in the
returned signal are expected due to the highly variable
reflective properties of the terrain and/or the size and shape of
the objects on the ground (Ussyshkin et al., 2007). Examples of
such surveys are corridor projects over highways covered by
dark asphalt with white painting on top, or power transmission
line corridors where the signal strength from thin wires is very
weak compared to that from the ground. In these cases, if the
receiver’s dynamic range is limited and cannot accommodate a
wide range of signals, weak signals could be lost, or strong
signals could saturate the receiver, consequently compromising
range data accuracy (
Figure 4). Range data accuracy may even worsen when small-
size surveyed targets are suspended over terrains with highly
variable reflective properties (black/white roads or
snow/wetland).
Figure 4. Simplified illustration of a possible error due to
limited dynamic range of the lidar receiver: If the signal
variations exceed the receiver’s signal dynamic range, range
measurement accuracy may be compromised.
On the other hand, lidar system manufacturers typically
characterize lidar performance for the most general case of
targets: that is, flat open terrain with uniform reflective and
physical properties and no strong signal variations within a
single mission. Thus, the accuracy numbers presented on a
system specification sheet may be inapplicable to many real-life
operating scenarios in which strong signal variations occurring
on a microsecond time scale may challenge the lidar receiver’s
dynamic range.