Evaluation of different scan configurations for an effective field procedure on a Terrestrial LiDAR Scanner

Organised by Laboratory of Geo-information Science and Remote Sensing

Wed 12 March 2014 13:30 to 14:00

Venue Gaia, gebouwnummer 101
Room 1

By Alvaro Ivan Lau Sarmiento (Peru)


Despite the advances in forest measurements, characterizing parameters using a terrestrial laser scanner (TLS) still remains a challenge. This challenge is mostly due to the quality of the scan, technical constraints of the scanning system and adverse environmental conditions. A wide range of laser scan systems are commercially available today, and among those, partial hemispherical systems are the most common. These systems do not allow you to scan on a zenith angle and; therefore a second scan, on a tilted position, is necessary. A full hemispherical scan can be achieved by merging both scans. However, the use of a second scan increases project’s budget. Knowing in advance the scan configuration needed for specific parameters, will benefit the fieldwork procedure. This study investigates the need of the second scan for three parameters, digital terrain model (DTM), forest height (FST) and top of canopy (TOC). In this study different setups (discrete-return and multiple-return) and different resolutions (0.5 m and 1.0 m) were analyzed. First, RMSE residuals of up to 2.3 m proved that discrete-return scans could not achieve the same accuracy as multiple-return scans. Then, visual assessment of the rasters indicated the presence of outliers at 0.5 m, compared to 1 m resolution. This was confirmed by standard deviation; where 1 m resolution raster showed less dispersion than 0.5 m resolution raster (e.g. 9.23 % less dispersion for FST parameter). Finally, RMSE was calculated for the different parameters at 1 m resolution. For the DTM parameter, upright scans showed 78 % more accuracy than tilted ones, with a RMSE below 0.5m. However, for the FST parameter, upright scans Is only 6 % more accurate than tilted scans. For the TOC parameter the tilted scan is 12 % more accurate than the upright scan. Despite of the results, RMSE residuals are still too large (around 1 meter difference) to confirm the exclusive use of one scan (upright or tilted) to calculate these parameters. This study proved that using both scans, upright and tilted, for these parameters is the most effective way to have accurate forest measurements.