LIDAR is an active remote sensing technique that measures distances with unprecedented accuracy. Similar to RADAR, it is typically an airborne optical remote-sensing technique, although ground-based and space-borne systems do exist. Shorter wavelengths in the electromagnetic spectrum, typically in the ultraviolet, visible or near-infrared are transmitted at frequencies up to 150 kilohertz (kHz). The return signal is then recorded either as discrete values or in full waveform over multiple returns. This allows mapping of the forest canopy surface, the tree structure, as well as the underlying topography.
LIDAR is an active remote sensing technique that measures distances with unprecedented accuracy. Similar to RADAR, it is typically an airborne optical remote-sensing technique, although ground-based and space-borne systems do exist. Shorter wavelengths in the electromagnetic spectrum, typically in the ultraviolet, visible or near-infrared are transmitted at frequencies up to 150 kilohertz (kHz). The return signal is then recorded either as discrete values or in full waveform over multiple returns. This allows mapping of the forest canopy surface, the tree structure, as well as the underlying topography.
LIDAR’s principle advantage lies in its ability to capture data in a dense field, capturing up to five points per meter with the ability to penetrate forest canopies and map topography as well. This capability is unique to LIDAR and cannot be reproduced using other remote sensing techniques. The bottom line is that LIDAR produces canopy height, tree volume and topography estimations with unprecedented accuracy.
LIDAR data is comprised of point clouds, a dense field of data points over a three-dimensional coordinate system. This is in part due to the high sampling frequency coupled with the multiple returns measured. It is imperative to filter data and identify anomalies, whether it’s due to irregular sampling or abnormal point spacing which could be indicative of missing data or improper feature classification. This is a critical step when extracting value from the raw data.
The filtered data still poses data processing challenges as the amount of data collected by LIDAR surveys is significantly greater when compared to other remote sensing techniques. For example, a surveyed area encompassing 5 hectares produces 1.3 kB of data with Landsat ETM satellite imagery with 15 m resolution and 390 kB of data with IKONOS satellite imagery with 1 m resolution, versus 5 MM of data for a LIDAR system with 1 pulse per meter and four return signals and 50 MB of data for a LIDAR system with 10 pulses per meter and four return signals.
To overcome data management issues raw filtered data can be converted to a multipoint feature class in a geodatabase. Among other advantages, the use of a geodatabase allows users to perform data analysis, share data with the end user, and produce graphical representations of the data set for end users.
To visualize LIDAR data with no outside reference data the data points can be organized in an elevation raster file. This does slightly reduce processing time and may suffice for some applications, but in order to produce the highest accuracy it is recommended to incorporate the multipoint feature class into a geodatabase terrain.
Geodatabase terrain maps blend multiple data sources using ground control points with known elevation such as roads, lakes, rivers or coastlines. This step allows for geometric corrections and calibration of the raw elevation raster file.
Digital surface models (DSM) and digital elevation models (DEM) are the graphical representations of the processed data. They are terrain maps representative of the canopy surface and terrain beneath the tree, respectively. Through careful interpretation foresters use these maps to extract detailed information on forest structure across large spatial fields.
LIDAR allows foresters unprecedented accuracy when gauging tree volume and canopy height information over large spatial fields. Aiding in growth analysis, fertilization regimes and logging operations, DSMs and DEMs are produced with a RMSE (Root Mean Square Error) as low as 15 to 20 cm vertically and 20 to 30 cm horizontally. Traditional methods that included labor-intensive field surveys, topographical maps and data extrapolation techniques or even satellite imagery could never match the detailed physical data captured by LIDAR.
Further advances in the development of LIDAR systems and associated data processing techniques are expected to reduce costs and improve process time. As the limitations of LIDAR data sets are overcome, its ability to produce unprecedented detail when calculating canopy height, tree density, volume estimates and underlying topography enable a bright future in the field of forestry management and related operational activities.