IEEE Geoscience and Remote Sensing Magazine - September 2022 - 57

to detect trees and measure the DBH and tree height as a
reference. Forest and tree attributes, i.e., the tree positions,
DBH, tree height, stem volume, stem curve, and total tree
biomass, were benchmarked among ULS, TLS, and MLS/
PLS using the same automatic processing method. The
RMSE percentage of the DBH, tree height, stem curve, biomass,
and total tree volume across three densities when
the automatic method was used were 16-28, 9-27, 28-55,
60-220, and 35-175%, respectively. The highest accuracy
was achieved within the least stem-dense plots, and it was
the lowest within the densest plots.
When manual measurement was used, the accuracy of
DBH estimates from ULS was comparable to MLS (RMSE
percentage of 15-30%) but twice as much as the RMSE percentage
of TLS. At the same time, tree height estimates had
higher accuracy than MLS and TLS, especially for mediumand
high-density plots. ULS was previously assumed to have
higher accuracy in stem curve estimates in comparison with
terrestrial-based technologies, especially for upper parts of
trunks, due to the above-canopy perspective. On the contrary,
the accuracy was modest due to the lack of points from the
middle section of trunks, which led to less accurate stem curve
estimations. The ULS-based PHC stem curve estimate was, in
general, higher than that of MLS but lower than that of TLS.
DISCUSSION AND OUTLOOK
Field observation is an inevitable part of forest investigations.
Conventional field inventories are labor-intensive,
time-consuming, and expensive. They are limited to small
areas and have a low temporal resolution. Small sample
sizes may lack representatives. Thus, errors in field observations
propagate to a large area through airborne and satellite
remote sensing. Foresters and researchers are constantly
looking for efficient solutions for acquiring reliable forest
attributes and conditions.
Significant attention has been paid to close-range remote
sensing solutions in recent decades. Research and
progress demonstrated promising potential, brought new
knowledge and possibilities, and raised questions that need
to be answered before close-range remote sensing can be
practically applied. Meanwhile, developments also challenge
knowledge about the applicability, performance, and
potential of existing methods for forest observations. This
section discusses the challenges that close-range remote
sensing faces, especially data quality, which directly impacts
the performance of forest attribute estimation. Trends
and opportunities for applying close-range remote sensing
in forest observations are further explored.
CHALLENGES
The quality of the input data, e.g., the completeness and
accuracy, fundamentally determines the quantity and reliability
of the information that can be derived. Sophisticated
processing methods can calibrate, compensate, and correct
errors and inaccuracies inherent in raw data, but they introduce
IEEE Geoscience and Remote Sensing Magazine - September 2022

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