OF THE FINNISH GEODETIC INSTITUTE N : o 137 METHODS AND TECHNIQUES FOR FOREST CHANGE DETECTION AND GROWTH ESTIMATION USING AIRBORNE LASER SCANNING DATA

Xiaowei Yu
Abstract:Airborne laser scanning has been used increasingly for extracting and estimating forest parameters. Experiences in Nordic countries and Canada have shown that retrieval of stem volume and mean tree height on a tree or stand level from laser scanner data performs as well as, or better than, photogrammetric methods, and better than other remote sensing methods. The increasing interest in laser data for forestry applications has led to the present research, which quantifies forest growth and detects possible changes over time using repeated multi-temporal laser surveys over boreal forests. For the thesis, methods and techniques were developed for detecting change automatically and estimating forest growth using multi-temporal airborne laser scanning. The performance of these methods was evaluated based on the field measurements consisting of individual trees or sample plots. All the component studies were carried out in boreal forest at a test site in southern Finland. For the detection of change, e.g. harvested or fallen trees, an automatic method was developed based on the image differencing technique applied to digital canopy height models generated from laser data from different dates. New scientific approaches developed for height and volume growth estimation were the individual tree-top differencing method, digital surface differencing and canopy height distribution based analysis. In the individual tree-top differencing method, growth estimation was based on individual tree identification and a tree-to-tree matching algorithm. The digital surface differencing method was based on the difference image of digital surface models. In the analysis based on canopy height distribution, growth was determined as a function of the difference in corresponding percentiles of the canopy height distribution between different laser acquisitions. These methods can be applied at both the individual tree level and the plot/stand level. The findings reported in this thesis indicated that multi-temporal airborne laser scanner data can be used for estimating or predicting growth and detecting harvested area and fallen trees with an acceptable level of accuracy (an RMSE of less than 0.5m for individual tree height growth, a standard deviation of about 6.7 mha (26.8%) for volume growth and 0.15 m for mean height growth, and a detection accuracy of 80% for harvested trees). The methods developed could be used to complement field measurements, to improve predictions from a growth model and to develop newgeneration forest growth models.
What problem does this paper attempt to address?