IEEE Geoscience and Remote Sensing Magazine - September 2022 - 60

Applying MS and RGB cameras to estimate forest structure
[244] and biomass [160] has achieved limited success. Both
of those studies indicated that MS data did not provide added
value in comparison to RGB camera-based data. The authors
of [245] studied the biodiversity of boreal forests, and
their results showed that HS and photogrammetric data
could be used, especially for studying structural diversity.
In [246], the authors used RGB, MS, and TIR sensors for the
phenotyping of forest genetic trials. They concluded that
the combination of spectral indices and canopy temperature
accounted for about 60% of the population variability
in the stem volume of Pinus halepensis. Since these findings
are based only on a few specific studies, further research is
required to obtain general conclusions.
In general, the door for deriving more tree and forest attributes
through close-range remote sensing has been opened.
The reliability of the feature estimations largely determines
their application potentials. At this moment, challenges exist
in deriving critical tree and forest attributes as well as in
the application of such techniques in different forest conditions.
One example is the applicability of the latest technologies
in tropical areas. Since the top canopy of all the trees is
merged, the fundamental processing of geometrically locating
and reconstructing individual trees from ULS becomes
more difficult than in boreal areas. Further research needs
to answer questions about proper sensors and data acquisition
protocols for challenging tropical conditions.
HARDWARE AND SOFTWARE
Hardware and software development supports the advance
of close-range remote sensing for forests. Among the advances
in hardware, the capacity for capturing MS information
from active sensors has been one of the most significant
developments in the past decade, which is expected
to solve many challenges that cannot be addressed using
spatial features only. Currently, the utilization of MS-LS is
still at a very early stage in forest research. Before MS-LS
can be practically used, a number of challenges have to be
resolved. One is to decide the best wavelengths and their
number and the required special resolution for particular
applications. Technically, fewer wavelengths make instrument
design more straightforward, and it becomes more
feasible to produce high-quality point clouds in terms of
the spatial resolution, point collection frequency, and scanning
range. A higher number of wavelengths requires specific
consideration in the optic design and compromises the
point cloud quality. Therefore, sensor properties should be
carefully designed depending on use cases.
The second challenge is the lack of applicable calibration
solutions, especially for the incidence angle and range
effect. Methods have been developed for correcting the
incidence angle and range effect on lidar intensity. The
lidar equation was found to be available for distance correction
under the premise that the target-sensor distance
was greater than 10-15 m for the FARO LS880 and Leica
HDS 6100 scanners since both instruments were equipped
60
with a brightness reducer for near distances. The most effective
method for near distance correction was applying a
reference table. For incidence angle correction, an empirical
function was developed [137]. However, the relationship
between range and intensity changes when the target is
smaller than the laser footprint since some of the reflected
laser intensity is lost due to multiple scattering [176]. A priori
information about the illuminated footprint is required
unless it could be assumed that the target width and density
can be approximated from other properties, such as tree
forms and environmental conditions. However, the stability
of intensity features with varied scanner-target distances
and incident angles need to be examined.
Studies of the intensity responses of the stems and
canopies of various tree species should be conducted to
take advantage of the increasing availability of MS active
sensors and to take a step toward automated methods
for the collection of field data. Perfectly collimated laser
beams with the same beam diameter at the output and
beam divergence could solve the challenges of varying
intensity-distance relationships with small targets. When
the amount of multiple scattering is similar for each wavelength,
a normalized index of the wavelengths should
remain stable, although the distance will vary. Rigorous
calibration models should also be developed for all shapes
and forms of trees.
Applications of HS systems in the field are very limited
at this moment. Safety issues are the main concern
that arises from the utilization of a supercontinuum laser
source, due to the high power of the laser, which causes immediate
damage to the eye when exposed. The eye safety
issue should be tackled before commercial manufacturers
are interested in bringing HS and LS to the market. The
weight, e.g., 15 kg, also limits field operability, especially
in rugged terrain, whereas single-wavelength commercial
TLSs weigh 5 kg or less. In addition, the generally larger
beam divergences in HS TLS systems compared to singlewavelength
systems result in wider laser footprints that
may cause errors at longer distances, due to edge effects
from targets. In addition to hardware development in controlled
environments, more research should be done in the
field to provide practical instructions for using MS and HS
LS. The advantages of the additional spectral information
provided by an HS system need to be jointly considered
with economic feasibility.
The development of data science has significantly
pushed forward the development of close-range remote
sensing. The most impressive progress is probably the dense
matching that achieved a breakthrough in the early 2000s
and today is been routinely used in various terrestrial and
aerial imagery observations through commercial software.
Yet, there are still many unsolved challenges that require
new algorithms and software. Geometrical inconsistency
in the data is currently the most significant limiting factor
for mobile applications. As long as geometrical inconsistency
exists, mobile mapping is suitable for a few applications
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE SEPTEMBER 2022
IEEE Geoscience and Remote Sensing Magazine - September 2022

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - September 2022
