TUT Inverse Problems research group is organizing a very informal small-scale workshop for discussing new interdisciplinary possibilities in tree/forest modelling. The idea is to combine advanced instrumentation, mathematical methods and plant models. The participants include experts from the fields of laser scanning, functional-structural plant modelling, forest science and mathematical modelling.
The agenda of the workshop will be built around a list of research interest and open research questions suggested by the participants. The first day of the workshop will consist of gathering such a list, but we encourage participants to send questions in advance with the form below. Suggestions received before the workshop will be listed here:
- How can Quantitative Structure Models (QSMs) be utilized in the future, e.g., on large areas?
- QSMs enable fully automatic species recognition from TLS data. How can this be utilized?
- Should we store QSMs in a public database, and if so, what would be the applications of such a database?
- Faster/simpler/non-realistic models of tree growth for more effective calibration/tuning of the algorithms (Bayes Forest) producing Stochastic structure models (SSM).
- Genetic clone studies: relation between SSM (computer tree clones generator) and real genetic tree clones. How much structural variability is possible in computer and real clones? Calibration of SSM’s.
- Methods for estimating the leaf area density/leaf area index at individual tree scale and forest plot scale from TLS/ALS data.
- Issues with LAI and foliage profile estimation with TLS.
- Retrieving growth development parameters from crown analysis (rhythm ratio, dormancy, branching distribution laws) with ALS.
- How can we estimate tree development status (as defined by architectural botany) within heterogeneous forests from Aerial Lidar range data acquisitions?
More extensive information about the topics can be found below in the Material section. If you are a presenter or you otherwise feel that the participants should read some material before the workshop, please contact the organizers and we will add your material to the list.
There are currently 20+ people confirmed to participate.
Mathematical modelling / Organizers
Information on how to contact the organizers can be found on the front page.
Forest remote sensing
Below is a list of some of the hotels in the Tampere centre. They are in no particular order, and are all a walking distance away from the venue, Pikkupalatsi.
When travelling from abroad your flight will most likely take you to Helsinki. From there you'll have the option to travel the remaining 200 km to Tampere either by train, bus or a plane.
The venue for the workshop is in the centre of Tampere in a building called Pikkupalatsi. This property was originally build as a modest residence for the first superintendent of the Finlayson factory back in 1897. Since then the building has been used an apartment building and even office space.
Nowadays the top floor is used as an apartment, and the first and basement floors can be rented for various types of events. The first floor has been restored to its original appearance and consists of several individual rooms. The basement houses the sauna and pool area and also a pool room (biljards, not swimming) with a table more than a hundred years old.
The street address of the Pikkupalatsi is Hämeenpuisto 7 which is just a walking distance away from the city centre. The bus line 3 drives though the main street, Hämeenkatu, and stops at Pikkupalatsi, and the name of the bus stop is appropriately Pikkupalatsi.
We suggest that every field gives a general introduction and an overview of open questions in a short free-form presentation of < 30 minutes. Proposed introductions include the following:
|Topic / Field||Presenter(s)||Slides||Size|
|Tree modelling: TLS, QSM, applications||Pasi||PPTX||10.2|
|Stochastic structural plant models||Ilya||PPTX||8.4|
|forest laser scanning instruments||Kim & Alan||27.0|
|multiwavelength scanning||Mark & Fadal||PPTX||8.5|
|plant morphogenesis and architecture||Frederic||17.9|
|forest genetics||Juha & Kaisa||Coming soon|
|forest growth models||Mikko P.||PPTX||2.0|
|forest ecology, carbon issues||Raisa||PPTX||7.8|
The presentations will not be the main focus of the workshop. They should only offer a compact introduction to the field, and thus we encourage people to collaborate on the presentations to minimize their number. Each presentation should end on a list of open research questions, which will be collected and used as the agenda for Wednesday.
|Start||Tuesday, 7th June||End|
|09:45||Morning coffee at Pikkupalatsi||10:00|
|10:00||Opening of the workshop||10:30|
|15:30||Question list refining||16:30|
|19:00||Self-planned evening program||19:00|
|Wednesday, 8th June|
|08:45||Morning coffee at Pikkupalatsi||09:00|
|09:00||Information about the day||09:15|
|09:15||Discussions in groups||12:00|
|13:30||Group discussions continue||14:30|
|14:30||Travel to Viikinsaari||15:20|
|16:00||Group discussions continue||19:00|
|19:00||Dinner in Ravintola Viikinsaari||21:00|
|21:30||Return by ferry to Laukontori||21:50|
|Thursday, 9th June|
|08:45||Morning coffee at Pikkupalatsi||09:00|
|09:00||Summary of discussions||10:00|
|15:30||Ending of the workshop||15:45|
Material on topics
This section lists background information on topics to be discussed.
QSMs: large areas; applications and prospects
Quantitative Structure Models (QSM) are cylinder based models reconstructed from terrestrial laser scanning data. These models contain the geometric properties of a tree as well as the topological branching structure. For a simple overview of QSMs, please see the 3D Forest Information animation. For interactive examples of the models, visit the online Tree Gallery. For more details, see the paper Fast Automatic Precision Tree Models from Terrestrial Laser Scanner Data.
We are interested in how QSMs could be utilized with large areas, and what the future applications could be. For applications where illustration is key, we've developed alternative ways to visualize QSMs not as cylinders but continuous Bézier surfaces. Visualizing reconstructed tree models video demonstrates visualization with these surfaces with textures and leaves sampled from distributions based on the QSM.
Stochastic structure models
Stochastic structure models (SSM) are structure models augmented with stochastic rules. In particular, we've augmented the LIGNUM model with stochastics received from reconstructed quantitative structure models. For details, see the paper Data-based stochastic modeling of tree growth and structure formation.
We've found that SSMs can be used to create realistic plant models based on data and thus accounting for stochastic influences. SSMs allow modelling of plant growth which is one of our open research areas.
Tree model database
QSMs offer a compact representation of the structure of a tree, and thus are ideal to be stored in relational databases. We've been working on a demonstration on such a database. Currently we store information in forest plot, tree, tree model, branch and cylinder tables, and enable the computation of, e.g., volumes, areas, and size distributions.
Is all the relevant information in the database, should something be added? Where would you utilize a tree database? What are the likely use cases?
Below is a list of plans that were made during the workshop. The list only contains the plans that were reported with the provided template. If you would like to add something to the list, please contact the organizers.
Better (breeding) features (more wood) by changing genetics.
Find a mutant (lack of apical dominance for an example) phenotype differs from normal. Test that mutation is inherited by mendelian way (genotype is reason for strange phenotype). MAP mutated gene(s). Make marker for gene. Use elite form for breeding to make more valuable/higher biomass forest.
Objectives and timeline
- Effective search of wrong/bad mutants in the forest (ALS).
- Models calibration to the known mutants.
- Existing birch mutants can be already utilized for modelling and proper phenotyping. Modelling normal and mutant can be done in close future. National screening for new apical dominance mutants could be tried much later distant future.
- Clone plot QSM analysis
Total: 5-10 years
- Clone birches sites
- Known described mutants
- Phenotyping of known mutants - structural characteristics
- Characterizing physiological traits of mutants
- Develop v. good predictive model incorporating deep physiology (potentially ensemble of multiple models)
- Incorporate forest management model to predict stand development
- Test models on known stand development data, then incorporate mutant/variant traits.