The doctoral dissertation in the field of Forestry will be examined at the Faculty of Science, Forestry and Technology, Joensuu campus and online.
What is the topic of your doctoral research? Why is it important to study the topic?
The assessment of forests' structural biodiversity is a timely topic as forests play a significant role in biodiversity-related decision-making. Forests support approximately 80% of the world’s terrestrial biodiversity. Forest vegetation structure and their complexity affect local biodiversity by providing shelter and breeding sites and affect the distribution and availability of resources. In assessment of forest vegetation structure, remote sensing data, such as airborne laser scanning (ALS) data and optical data, are widely used. Remote sensing -based assessment of forests makes it possible to produce detailed information at both fine and broad spatial scales.
What are the key findings or observations of your doctoral research?
Based on this thesis, research on ALS-based assessment of forest biodiversity has been clustered on the European and North American continents, and most of the research currently has focused on animal ecology, tree species richness/diversity measures and the assessment of dead wood. There are no ALS metrics that would suit all the various assessments of forest biodiversity, although some ALS metrics have been used more often than others. It was also discovered in this thesis how important it is to utilise as representative data as possible when dealing with the rare phenomena. This was demonstrated by mapping ecologically important, but relatively rare European aspen trees. SMOTE data augmentation was utilised to tackle issues caused by imbalance in the training data. This thesis also reported that the categorical variables describing site conditions (e.g., main type) were important in the prediction of forest age at plot level.
How can the results of your doctoral research be utilised in practice?
The most commonly studied topics and the most often utilised ALS metrics presented in this thesis could drive the research of remote sensing based assessment of forest biodiversity towards less studied topics, which were accounted in the first paper. The results presented in the second paper will invite reflection on how important it is to utilise as realistic data as possible (e.g., tree species proportion) when dealing with rare phenomena. This means that the utilisation of data where rare phenomena is over-represented, should not be favoured. The issue can be tackled via statistical methods, such as SMOTE data augmentation. GPBoost method utilised in the third paper is relatively new machine learning algorithm, which is not utilised in forestry applications to the best of my knowledge. The method could be further tested with other forest attributes than forest age.
What are the key research methods and materials used in your doctoral research?
The first paper of the thesis is a literature review for which the data (research articles) were gathered by the doctoral student. Suitable articles were read thoroughly, and they were divided as the chapters of the first paper based on their topics.
In the second and the third paper, remote sensing data (airborne laser scanning, aerial images, satellite images) and field plots measured by Finnish Forest Centre and Natural Resources Institute Finland, were utilised. The airborne laser scanning data were the newest available dataset from study areas which is high pulse density data which is collected by National Land Survey of Finland based on national acquisition programme. Aerial images were also collected by the National Land Survey of Finland.
The third paper utilised Sentinel-2 satellite images along with laser scanning data. In the second paper, large European aspens were detected at individual tree level with Random Forest classifier by utilising metrics calculated from remote sensing data. As large aspens were rare, their proportion was artificially increased by SMOTE data augmentation which created new synthetical observations of large aspen trees. In the third paper, forest plot age was predicted with area-based approach by utilising remote sensing and field data -based predictors. Age predictions were compared between linear mixed effects model and GPBoost algorithm.
The doctoral dissertation of Janne Toivonen, MSc (Agr & For), entitled Assessing the structural biodiversity of forests with airborne laser scanning and optical data (Metsien rakenteellisen monimuotoisuuden arviointi lentolaserkeilauksen ja optisen datan avulla) will be examined at the Faculty of Science, Forestry and Technology, Joensuu Campus and online. The opponent will be Senior Lecturer Eva Lindberg, Swedish University of Agricultural Sciences, and the custos will be Research Professor Petteri Packalen, Natural Resources Institute Finland. Language of the public defence is English.
For further information, please contact:
Janne Toivonen, jantoi@uef.fi, p. 044 300 5530
