Tree species identification constitutes a bottleneck in remote sensing applications. Waveform LiDAR has been shown to offer potential over discrete-return observations, and we assessed if the combination of two-wavelength waveform data can lead to further improvements. A total of 2532 trees representing seven living and dead conifer and deciduous species classes found in Hyytiälä forests in southern Finland were included in the experiments. LiDAR data was acquired by two single-wavelength sensors. The 1064-nm and 1550-nm data were radiometrically corrected to enable range-normalization using the radar equation. Pulses were traced through the canopy, and by applying 3D crown models, the return waveforms were assigned to individual trees. Crown models and a terrain model enabled a further split of the waveforms to strata representing the crown, understory and ground segments. Different geometric and radiometric waveform attributes were extracted per return pulse and aggregated to tree-level mean and standard deviation features. We analyzed the effect of tree size on the features, the correlation between features and the between-species differences of the waveform features. Feature importance for species classification was derived using F-test and the Random Forest algorithm. Classification tests showed significant improvement in overall accuracy (74→83% with 7 classes, 88→91% with 4 classes) when the 1064-nm and 1550-nm features were merged. Most features were not invariant to tree size, and the dependencies differed between species and LiDAR wavelength. The differences were likely driven by factors such as bark reflectance, height growth induced structural changes near the treetop as well as foliage density in old trees.

The objective of this study was to develop a method for estimation of forest stand variables and updating the forest resource data, based on a well known and widely used method among forest sector, aerial photography. The second objective was to produce information of cost-effectiveness and accuracy of digital surface model (DSM) generated from very high resolution aerial images in comparison of methods based on aerial laser scanning (ALS). The study area covering circa 2000 hectares is located in state owned forest in Hämeenlinna, Southern Finland. The study material consisted of 85 digitised and orthorectified colour-infrared (CIR) aerial photographs, LiDAR measurements of the corresponding area and field measurements of 402 concentric circular plots. Both the remote sensing data and the field measurements were acquired in 2009. In this study, the accuracy of DSM generated from very high resolution CIR - aerial images was examined in the estimation of forest stand variables. Estimation of forest stand variables was made using non-parametric k-nearest neighbour method. Sequential forward selection was used for selecting features from remote sensing data and the examination of accuracy was done with cross validation. The variables examined were mean diameter, basal area, mean height, dominant height and mean volume. Relative RMSE -values of DMS estimation were at the best with mean diameter, basal area, mean height, dominant height and mean volume 33,67 %, 36,23 %, 25,33 %, 23,53 % and 40,39 %. For the reference ALS-data, relative RMSE-values were 25,26 %, 27,89 %, 19,94 %, 16,76 % ja 31,26 %. Photogrammetric DSM was best suited for estimating dominant and mean height and produced estimates slightly more inaccurate than those of reference ALS-data. When estimating mean diameter, photogrammetric DSM was slightly better, but at mean volume estimation, ALS-data proved again to be a little more a accurate than photogrammetric DSM. At basal area estimation, ALS-data gave considerably better results than photogrammetric DSM. This research showed that the photogrammetric DSM suits well for updating the forest resource data, and also satisfies the requirements in a more economic way.

The pulse density of airborne Light Detection and Ranging (LiDAR) is increasing due to technical developments. The trade-offs between pulse density, inventory costs, and forest attribute measurement accuracy are extensively studied, but the possibilities of high-density airborne LiDAR in stream extraction and soil wetness mapping are unknown. This study aimed to refine the best practices for generating a hydrologically conditioned digital elevation model (DEM) from an airborne LiDAR -derived 3D point cloud. Depressionless DEMs were processed using a stepwise breaching-filling method, and the performance of overland flow routing was studied in relation to a pulse density, an interpolation method, and a raster cell size. The study area was situated on a densely ditched forestry site in Parkano municipality, for which LiDAR data with a pulse density of 5 m–2 were available. Stream networks and a topographic wetness index (TWI) were derived from altogether 12 DEM versions. The topological database of Finland was used as a ground reference in comparison, in addition to 40 selected main flow routes within the catchment. The results show improved performance of overland flow modeling due to increased data density. In addition, commonly used triangulated irregular networks were clearly outperformed by universal kriging and inverse-distance weighting in DEM interpolation. However, the TWI proved to be more sensitive to pulse density than an interpolation method. Improved overland flow routing contributes to enhanced forest resource planning at detailed spatial scales.

Determining the market value of forest properties is needed for several purposes. In Finland the most used methods for valuation of forests are summation approach, income approach and market approach. Real estate valuation methods are standardized by the International Valuation Standards Council (IVSC). The council publishes standards that have been the premise for real estate valuation also in Finland. However, standard for the valuation of forests doesn’t exist due to significant problems in every method used. From International Valuation Standards Councils perspective valuation should always be market-based. Figures for the calculations should be derived from the market. This has been problematic for forest properties as the specific forest inventory data has been too expensive and difficult to collect. The new forest inventory data system of The Finnish Forest Centre, which is based on airborne laser scanning, creates new possibilities for combining the data with the market prices. This enables a better examination of the valuation methods used and a possibility for the creation of a standardized method. The purpose of this study was to compare the attributes and suitability of the most used valuation methods when determining the market value of forest properties. Research material of this study consists of 15 representative forest property transactions (areas over 10 hectares) from Central Finland and the laser scanning based inventory data of these properties. The attributes investigated were the size of correction from total value when using summation approach, the internal rate of return in the income approach, possible net income for the near future and the accuracy of these valuation methods. In addition, taxation of forest revenues, transfer taxes, administration costs of forests and trade costs were applied in examination of these methods. Processing and calculations of data were carried out with MELA, Motti, Tforest and Excel programs. The average internal rate of return was 5,3 percent and median 4,3 percent, which is similar to previous studies. Investments in forest properties are categorized to an average risk-return investment class. The correction from total value when using summation approach was similar to previous studies as it varied from -2 to -60 percent and was -26 percent on average (and -13 percent when expectation values were left out). The possible net income from these forest properties from the period of five years could cover 64 percent of market prices; however notable differences were among properties. When taking taxes, administration and trade costs into account the average internal rate of return sets down between 3 - 4 percent. The total value correction in summation approach is only -4,5 percent on average (+12,9 percent without expectation values). The problems of the valuation methods can be seen when looking at the accuracy of the methods. Standard deviation of every method varies from 25 - 35 percent when comparing them to the market values. Notable is that with a very simple method; multiplying the growing stock with the average stumpage prices, the results are as accurate as with more complex methods. The most accurate results for the whole research material were calculated with the income approach using 5 percent interest rate. Also using the summation approach and taking taxes, administration and trade costs into account was very accurate. More research is still needed for every method. Perhaps in practical valuation tasks the market value of forest properties should be investigated using multiple methods side by side, as IVSC has proposed. The results of this study are similar to previous studies and therefore support the intention for combining the new laser scanning based forest inventory data to the market prices. The use and research of extensive and up-to-date market data of forest properties could also open new possibilities in valuation of non-market benefits.

Accurate and economical remote sensing method with good temporal resolution is required for mapping up-to-date information about the forest resources. Detecting forests by optical satellite images is an inaccurate procedure with the saturation problem. Airborne laser scanning (ALS) is a precise application, but the inventory process is slow and expensive. Recently the new synthetic aperture radar (SAR) satellites with a high spatial resolution have caused a renaissance of radar-based remote sensing. The purpose of the master’s thesis was to investigate the accuracy of forest mapping by radargrammetric processing of TerraSAR-X satellite images. The radargrammetry is based on stereoscopic measurement, which calculates 3D coordinates for corresponding points of the SAR image pair. In the research an area-based approach (ABA) was utilized to estimate forest attributes from the 3D points, and digital terrain model (DTM) produced by ALS was used to calculate height of the corresponding points. In plot-level the relative RMSEs for stem volume, biomass, basal area and mean height were 40.3 %, 39.9 %, 34.0 % and 15.9 %. In stands larger than 2 hectares the corresponding RMSEs were 20.2 %, 20.4 %, 36.1 % and 6.9 %. It’s notable that the estimation of basal area didn’t improve in stand-level at all. According to the research SAR radargrammetry is a precise technology to estimate forest canopy height, but the mapping of forest density is very unclear. Nevertheless the results about the estimation accuracy of forest stem volume and biomass by SAR radargrammetry were clearly better than the comparable estimation accuracy of optical satellite images.

There is a growing demand for high-quality spatial data and for efficient methods of updating spatial databases. In the present study, automated object-based interpretation methods were developed and tested for coarse land use mapping, detailed land cover and building mapping, and change detection of buildings. Various modern remotely sensed datasets were used in the study. An automatic classification tree method was applied to building detection and land cover classification to automate the development of classification rules. A combination of a permanent land cover classification test field and the classification tree method was suggested and tested to allow rapid analysis and comparison of new datasets. The classification and change detection results were compared with up-to-date map data or reference points to evaluate their quality. The combined use of airborne laser scanner data and digital aerial imagery gave promising results considering topographic mapping. In automated building detection using laser scanner and aerial image data, 96% of all buildings larger than 60 m2 were correctly detected. This accuracy level (96%) is compatible with operational quality requirements. In automated change detection, about 80% of all reference buildings were correctly classified. The overall accuracy of a land cover classification into buildings, trees, vegetated ground and non-vegetated ground using laser scanner and aerial image data was 97% compared with reference points. When aerial image data alone were used, the accuracy was 74%. A comparison between first pulse and last pulse laser scanner data in building detection was also carried out. The comparison showed that the use of last pulse data instead of first pulse data can improve the building detection results. The results yielded by automated interpretation methods could be helpful in the manual updating process of a topographic database. The results could also be used as the basis for further automated processing steps to delineate and reconstruct objects. The synthetic aperture radar (SAR) and optical satellite image data used in the study have their main potential in land cover monitoring applications. The coarse land use classification of a multitemporal interferometric SAR dataset into built-up areas, forests and open areas lead to an overall accuracy of 97% when compared with reference points. This dataset also appeared to be promising for classifying built-up areas into subclasses according to building density. Important topics for further research include more advanced interpretation methods, new and multitemporal datasets, optimal combinations of the datasets, and wider sets of objects and classes. From the practical point of view, work is needed in fitting automated interpretation methods in operational mapping processes and in further testing of the methods. Laadukkaan paikkatiedon tarve kasvaa jatkuvasti, ja paikkatietokantojen ajantasaistukseen tarvitaan tehokkaita menetelmiä. Tässä tutkimuksessa käytettiin useita uudenaikaisia kaukokartoitusaineistoja. Niiden pohjalta kehitettiin ja testattiin automaattisia, objektipohjaisia tulkintamenetelmiä yleispiirteiseen maankäytön luokitteluun, yksityiskohtaiseen maanpeitteen ja rakennusten kartoitukseen sekä rakennusten muutostulkintaan. Rakennusten tulkintaan ja maanpeiteluokitteluun sovellettiin automaattista luokittelupuumenetelmää, jonka avulla voidaan automatisoida luokittelusääntöjen kehittäminen. Uusia aineistoja voidaan analysoida ja vertailla nopeasti, kun luokittelupuumenetelmää käytetään yhdessä pysyvän maanpeiteluokittelutestikentän kanssa. Luokittelu- ja muutostulkintatuloksia verrattiin niiden laadun arvioimiseksi ajantasaiseen kartta-aineistoon tai referenssipisteisiin. Ilmalaserkeilausaineisto ja digitaalinen ilmakuva-aineisto yhdessä antoivat lupaavia tuloksia maastotietojen kartoitusta ajatellen. Automaattisessa rakennusten tulkinnassa 96 % kaikista yli 60 m2:n rakennuksista tunnistettiin oikein. Tämä tarkkuustaso (96 %) vastaa käytännön laatuvaatimuksia. Automaattisessa muutostulkinnassa noin 80 % kaikista referenssirakennuksista luokiteltiin oikein. Maanpeiteluokittelussa neljään luokkaan saavutettiin laserkeilaus- ja ilmakuva-aineistoa käyttäen 97 %:n kokonaistarkkuus referenssipisteisiin verrattuna. Pelkkää ilmakuva-aineistoa käytettäessä tarkkuus oli 74 %. Tutkimuksessa verrattiin myös ensimmäiseen ja viimeiseen paluupulssiin perustuvia laserkeilausaineistoja rakennusten tulkinnassa. Vertailu osoitti, että viimeisen paluupulssin käyttö ensimmäisen sijasta voi parantaa tulkintatuloksia. Automaattisten tulkintamenetelmien tuloksista voisi olla hyötyä maastotietojen manuaalisessa ajantasaistusprosessissa tai lähtötietoina kohteiden automaattisessa rajauksessa ja mallinnuksessa. Tutkimuksessa käytettyjen synteettisen apertuurin tutkan (SAR) tuottamien kuvien ja optisen satelliittikuvan tärkeimmät hyödyntämismahdollisuudet liittyvät maanpeitteen kartoitukseen. Yleispiirteisessä maankäyttöluokittelussa kolmeen luokkaan saavutettiin moniaikaista interferometrista SAR-aineistoa käyttäen 97 %:n kokonaistarkkuus referenssipisteisiin verrattuna. Aineisto osoittautui lupaavaksi myös rakennettujen alueiden jatkoluokitteluun rakennustiheyden perusteella. Jatkotutkimusten kannalta tärkeitä aiheita ovat edistyneemmät tulkintamenetelmät, uudet ja moniaikaiset aineistot, eri aineistojen optimaalinen yhdistäminen sekä useampien kohteiden ja luokkien tarkastelu. Käytännön näkökulmasta työtä tarvitaan automaattisten tulkintamenetelmien sovittamiseksi operatiivisiin kartoitusprosesseihin. Myös menetelmien testausta on jatkettava.

Multispectral airborne laser scanning (MS-ALS) sensors are a new promising source of data for auto-mated mapping methods. Finding an optimal time for data acquisition is important in all mapping applica-tions based on remotely sensed datasets. In this study, three MS-ALS datasets acquired at different times of the growing season were compared for automated land cover mapping and road detection in a suburban area. In addition, changes in the intensity were studied. An object-based random forest classi-fication was carried out using reference points. The overall accuracy of the land cover classification was 93.9% (May dataset), 96.4% (June) and 95.9% (August). The use of the May dataset acquired under leafless conditions resulted in more complete roads than the other datasets acquired when trees were in leaf. It was concluded that all datasets used in the study are applicable for suburban land cover map-ping, however small differences in accuracies between land cover classes exist.