Towards Radiometrically-calibrated LiDAR Intensity: Analysis of Range-corrected Reflectance compared to Hyperspectral Data
Lidar intensity represents an element of information often used visually, but not always suitable for e.g. classification as the intensity is highly variable even over similar surfaces. The differences are due to various instruments having different laser power, detector sensitivity and gain, but also due to flight altitude and field of view, since they vary the distance the light needs to travel through the atmosphere. There is also a variation dependent of angle of incidence, since the reflectance off a surface is not perfectly diffuse, which would have given even light distribution for all angles. As there has been a demand for a stable and predictable intensity distribution of the data, especially for forestry analysis, several suppliers have offered solutions that move away from the earlier automatic gain control in the detector circuit, aiming to increase the lidar range maximally, but at the same time making the intensity highly unpredictable. Riegl Scanners offer an option to choose between ?reflectance? and ?amplitude? in the processing of raw data to a lidar point cloud. The ?amplitude? is purely a number directly linked to the electric signal in the detector circuit, whilst the ?reflectance? is a range-corrected number meant to a degree to represent the true surface reflectivity. This reflectance value should in theory be corrected for range differences due to laser power, flight altitude and field of view. In collaboration with the Norwegian Mapping Authorities, we have carried out a test project to analyse whether there are advantages switching to “reflectance” rather than “amplitude” in our current nationwide lidar mapping contract for Norway. The presentation will unveil interesting results from multiple examples showing the difference between amplitude and reflectance data. We will present obtained reflectance values compared to known standard values for various surfaces (grass, asphalt, gravel etc.) as well as atmospherically corrected hyperspectral data over the same area. An analysis over the same sample area with variable laser power and flight altitude will also be presented, as well as some promising first results using range-corrected lidar reflectance in forestry analysis.