120 likes | 140 Views
Explore a statistical clone generator for plant models, bridging the gap between various experiments and model variations to account for morphological diversity in forests. Discover the compatibility with Quantitative Structural tree Model (QSM) data and the intraspecific diversity measurements to assess object distinctions.
E N D
Stochastic Structural Plant Models a platform for morphological diversity Ilya Potapov Tampere University of Technology
Closer Data-Model Interaction • Usually Functional-Structural Plant Models (FSPM) utilize fixed parameters obtained from “average” experiment. • Experimentally defined parameters imply certain conditions that are not universal. • Different objects and experimentations are referred to for the same model. • Simple functions assumed. • Data are not used every time the model runs.
Statistical clone generator • Stochastic Structural Model (SSM) is a statistical “clone” generator… • That is, SSM produces trees with statistically similar structural characteristics, but not exact copies of each other.
Forest applications • Single-species defined forest • Two-species forests
Data = QSM Quantitative Structural tree Model (QSM) Laser scanning 3D point cloud • QSM possesses all geometrical and topological information • Cylinder-based model (or any other geometrical primitive) PasiRaumonen et al., Remote Sens.5(2), 491-520, 2013.
SSM = a structure model with stochasticity • Full description compatibility with QSM (i.e. the same geometrical and topological information). • Stochastic rules of growth to account for variability.
SSM compatible with Data DATA MODEL Goal: Find the maximum correspondence between DATA and MODEL data sets OR: Find maximum overlapping of the distributions
Intraspecific diversity What makes two trees different? How much are they different? Particularly: How much do the (genetic) clones differ?
Bio-diversity: quantitative aspect Ds(cypress_1,cypress_2) = ? Ds(frog, cypress_1) = ? What is the mutual (structural) distanceDs between objects?
Construction of a (structural) distance • Characteristics to measure. • Algorithm to quantitatively compare the measured characteristics from two sources. • Ds(DATA,MODEL) = 0 Full correspondence between DATA and MODEL.
Open questions • Faster/simpler SSM candidates to facilitate the calibration of the “Bayes Forest” algorithm. • Quantitative relation between real (genetic) clones and SSM-generated “clones”.