3. Doktoritööd
Selle kollektsiooni püsiv URIhttp://hdl.handle.net/10492/7081
Sirvi
Sirvi 3. Doktoritööd Autor "Arumäe, Tauri" järgi
Nüüd näidatakse 1 - 1 1
Tulemused lehekülje kohta
Sorteerimisvalikud
Kirje Estimating forest variables using airborne lidar measurements in hemi-boreal forests(Eesti Maaülikool, 2020) Arumäe, Tauri; Lang, Mait; Packalén, Petteri (opponent)Forest management and planning requires up-to-date data, which commonly is acquired using field experts and ground measurements. Nowadays, more and more of data about forest stands is measured using remotely sensing methods. Most common methods include aerial photography and laser scanning from airplanes, also spectral measurements from satellites or even drone images and applications. This doctoral thesis focuses on developing applications and methods for utilising the airborne laser scanning (ALS) data that is freely available for the whole Estonia. The ALS measurements are carried out by the Estonian Land Board on a routine basis twice a year – in spring and summer. The first variable that was studied in this thesis was forest height. Based on the thesis, the most reliable method for forest height assessment was using the ALS point-cloud 80th height percentile (HP80). The small circular plot (radius of 15…30 m) and stand based studies showed high correlations with the field-measured forest heights and with great confidence it can be said, that ALS-based forest height estimations are close or even with higher accuracy, than field inspected. The second studied variable was standing wood volume. The ALS-based methods and models that were developed throughout this thesis used the idea, that standing wood volume is based on forest height and density. For this the HP80 and a threshold-based point count ratio was used (canopy cover - CC). ALS-based CC (CCALS) estimates were studied and compared with digital hemispherical photo based measurements. The results showed similar errors as were shown in other similar studies, with around 10-15% root mean square error (RMSE). The strongest correlation was shown using all echoes above a 1.3 metre threshold. Combining the CCALS and HP80 showed standing wood volume estimates with a similar error as we would receive from field measurements (<20%). The freely available multitemporal ALS data showed promising results for forest height growth monitoring and detecting small-scale disturbances. CCALS was shown to have strong predictive value, when compared with a four year difference in thinned and unthinned stands. The nation-wide ALS data can also be combined with forest height predictions from satellites, providing a faster update compared to the ALS data. Promising results were shown using the interferometric synthetic aperture radar (InSAR). Stand species maps generated using self-learning algorithms and satellite based spectral data can be used for developing species specific models of standing wood volume prediction. By combining these different datasets we can construct a nation-wide forest resource to help make better decisions for forest management and targeting fieldwork.