We present a new mathematical model of botanical trees capable of simu
lating the combinatorial structure of specific species based on their
bifurcation ratios. We first describe a general combinatorial model of
botanical trees for the purposes of synthetic imagery. We apply techn
iques from probabilistic analysis to generate random combinatorial tre
es and then model them as three-dimensional geometric trees. We choose
modelling functions that are partially based on results from theoreti
cal biology. By changing the underlying distribution used to generate
the random combinatorial trees, we are able to produce images of a wid
e variety of botanical trees. For example, just one parameter controls
the branching from dichotomous to monopodial. We then parameterize th
e model acording to Horton's first law. We implement our algorithm in
L-systems, a popular botanical modelling language. (C) 1998 Academic P
ress Limited.