Calibration of LiDAR histogram databases using a non-linear mathematical model
Loading...
Date
2011
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Dar es Salaam
Abstract
The airborne laser scanning technology (also known as Light Detection and Ranging-
LiDAR) in Finland has been proved to be the efficient remote sensing technique used
for forest data collection. The data obtained from this technology is used for pre
diction of forest, stand characteristics for different forest inventories. Using airborne
Laser scanning for the new inventory is very expensive. However as different forests
have been laser scanned there is a need of utilizing such previous databases in new in
ventory areas. Since the new site and the database sites have been laser scanned with
different scanning instruments with different flying altitudes, there is a need of cali
brating the databases to match with the new site LiDAR histograms. Therefore, the
aim of this study is to the calibrate the LiDAR histogram databases by a non-linear
mathematical model using the mean values of the LiDAR histograms. The accuracy
of the calibrated LiDAR data is verified by predicting the forest stand characteristics
using sparse Bayesian regression. The results obtained show that it is possible to
calibrated histogram databases and the calibrated LiDAR histograms can be used
in new inventory areas for the forest stand parameters estimation. It is observed
that when the small number of calibration set from the new area (approximately
50 plots) is combined with the calibrated plots from the database, the estimation
accuracy is almost equal to that when using the whole new area plots. This process
is cost effective since instead of scanning the whole new inventory area, only ran-
domly selected plots (approximately 50) can be scanned and be complemented with
calibrated database plots for prediction of the forest stand characteristics. In some
forests the geographical location does not support the aeroplanes to scan the forests,
hence using calibrated databases can help to predict the forest characteristics of such
areas.
Description
Dissertation
Keywords
Airborne Laser Scanning Technology, Mathematical modelling