Correspondence: ilya.potapov@tut.fi 7 Keywords: stochastic structure tree model; quantitative structure tree model; morphological clone; 8 terrestrial laser scanning 9 10 Summary Statement. 11We present an algorithmic framework, based on the Bayesian inference, for generating morphological 12 tree clones using a combination of stochastic growth models and experimentally derived tree 13 structures. 14 15
Abstract. 16Detailed and realistic tree form generators have numerous applications in ecology and forestry. Here, 17we present an algorithm for generating morphological tree "clones" based on the detailed 18 reconstruction of the laser scanning data, statistical measure of similarity, and a plant growth algorithm 19 with simple stochastic rules. The algorithm is designed to produce tree forms, i.e. morphological 20 clones, similar as a whole (coarse-grain scale), but varying in minute details of organization (fine-grain 21 scale). We present a general procedure for obtaining these morphological clones. Although we opted 22 vs. self-organizing (Sachs and Novoplansky, 1995;Palubicki et al., 2009) character of architecture 48 development). 49 50 However, the most promising plant architectural models are so called functional-structural plant 51 models (FSPM), also known as "virtual plants" (Room et al., 1996;Sievänen et al., 2000; Godin et al., 52 2004), because this type of models allows for a balanced description between morphological and 53 functional/physiological properties of a plant. Thus, it is capable of connecting the external abiotic 54 factors (e.g. radiation, temperature and soil) and the most vital functions of a plant organism (such as 55 photosynthesis, respiration, and water and salts uptake) with its structural characteristics 56 (Prusinkiewicz, 2004;Fourcaud et al., 2008). 57 58 Nevertheless, biologically relevant architectural plant models rely on data in a form of empirically 59 fitted functions and parameters that correspond to a particular species and/or certain site conditions 60 (Mäkelä and Hari, 1986;Rauscher et al, 1990;Perttunen et al., 1996;Lacointe, 2000). Thus, the 61 change in these conditions requires re-calibration of the models, which is done in a manual fashion 62 every time the model is simulated for the new conditions. Strong dependence on data, where each 63 simulation would be calibrated automatically by data, is limited by both computation time and lack of 64 the fast measurement and processing systems allowing for a detailed 3D morphological reconstruction 65 of the real plant/tree. 66Recently, we have reported a proof-of-concept study where we used reconstruction of a pine tree and 100 the corresponding FSPM (named LIGNUM (Perttunen et al., 1996; Sievanen et al., 2008)) to 101 demonstrate the practical feasibility of the approach (Potapov et al., 2016). In this work, however, we 102 develop a unifying interface for our procedure and use general-purpose fast procedural tree growth 103 model from (Palubicki et al., 2009), since such a simple procedural model is easier to ada...