Terrestrial Laser Scanning for Forest Structural Parameters Retrieval and 3D Forest Modelling
Vaccari, Simone1; Calders, Kim1; Herold, Martin1; Bartholomeus, Harm1; van Leeuwen, Martin2; Coops, Nicholas C.2
1Wageningen University, NETHERLANDS; 2University of British Columbia, CANADA
More than one third of the world's land surface is covered by forests which biomass, biophysical characteristics and environmental conditions keep changing over time. These dynamics raise the need to keep such areas monitored. Active remote sensing (RS) sensors called light detection and ranging (LiDAR) are capable to describe forest structural attributes by measuring interceptions of emitted laser pulses with the canopy. The spectral properties of forest constituents can be estimated by investigating optical data acquired from passive RS sensors. The challenge and objective of this study is to combine structural and radiometric characteristics of a forest in order to simulate its reflective properties by using three-dimensional (3D) radiative transfer modeling (RTM) ().
Terrestrial laser scanning (TLS) is an active sensor technology that uses a LiDAR scanner mounted on a tripod on the ground and is able to record the spatial distribution of the canopy constituents in the three dimensions. By standing at below-canopy level and having high sampling characteristics, TLS offers the possibility to more accurately describe vertical forest properties compared to airborne and spaceborne RS techniques. On the other hand, orbiting at higher altitudes, aerial/spaceborne optical RS devices are capable to cover areas with larger extents and acquire their spectral information by studying target's reflective properties. The challenge is to combine the two techniques in order to benefit from both methods' advantages.
Terrestrial and airborne/spaceborne data can be combined with each other by use of a 3D Radiative Transfer Modeling (RTM). The spatial distribution of crown constituents in the 3D space in combination with the radiometric properties of the simulated optical data influence the radiation interception, emission and scattering by canopy elements. In this study, the Monte Carlo Ray Tracing (MCRT) RTM model librat is used with the purpose of simulating spectral reflectance values of a virtual forest based on radiometric characteristics derived from Landsat 5 data. Forest biophysical characteristics include tree locations and heights, crown height and width, diameter at breast height and leaf area index. The analyzed forest is located in The Netherlands and is dominated by Pinus sylvestris trees, which structural parameters were retrieved from TLS data.
Our findings confirmed TLS to be a valid technique to describe forest structure attributes, which estimation was more accurate and precise compared to other techniques (e.g. airborne laser scanning, hemispherical photography and field data inventory). Forest 3D structure in combination with radiometric proprieties proved to successfully simulate Landsat 5 data through RTMs, obtaining an overall coefficient of determination (R2) of 53.42% when the RTM-simulated and Landsat-retrieved spectral profiles were compared. To further enhance the reliability of the simulations, improvements can be made by including topography and forest understory in the 3D model.
This research highlights the potential of TLS in describing forest structural attributes with the combination perspective with radiometric information obtained from optical imagery. The estimation of reflectance properties from 3D information rather than common techniques based on optical 2D data is therefore achievable and offers promising prospective for fusion of data from both optical and LiDAR sensors.