| dc.description.abstract | 3D visualisation of vegetation change can communicate for instance the loss of rare
plant species, vegetation stress or vegetation risks in order to raise awareness. 3D
visualisation might also be helpful in the nature management process, by visualising
the effects of certain decisions, such as removing biomass from a vegetation area. In
order to make a 3D visualisation of a vegetation model, it is necessary that this model
has a map as output together with more information which describes the composition
of this vegetation. These will serve as the input for distribution modelling of the
individual plant species. 3D models of the plants can be placed at the point
distribution which is created by this distribution tool. To get these 3D models of the
plants, the plants which contribute to the grain, colour and structure of the vegetation
type will have to be selected and 3D modelled. This point distribution of the plants
and the 3D models can be brought together in a 3D simulation, together with a DEM
and an aerial photo to model height and to give the substrate a natural colour
There are several models for vegetation modelling, plant distribution modelling and
3D plant modelling. The vegetation modelling is done using vegetation mapping
models or vegetation succession models. The mapping can be done using
geostatistical interpolation, generalized linear networks, artificial neural networks and
classification trees. These models result in a map, and in order to be used as input for
the plant distribution tool, a description of the abundance of each plant species within
the vegetation types is necessary. However, when the species abundance is described
using the Braun-Blanquet scales as a cover percentage, it is not possible to use it as
input for a computer generated distribution, because this requires the plants per square
meter as an input. A conversion is possible, but requires accurate modelled 3D plant
models. The vegetation succession modelling models the transition from the one
vegetation type to another vegetation type. This change is driven by changing
circumstances for the plants and the output as biomass is calculated for five different
layers in the vegetation. In order to model this layered output correctly, it is necessary
that the 3D plant models give a correct representation of the biomass of the plants and
it therefore requires accurate 3D growth models of the plant species.
There methodologies for creating a plant distribution vary a lot in complexity: the
complex Agent Based Models and Cellular Automata can model competition for
resources and dispersion, but also require a lot of research, calibration and validation.
A simple computer generated distribution on the other side does only require the
number of plants per square meter per vegetation type as input.
The 3D plant models can be divided into two types: the accurate models and the
sketch based models. Accurate modelling techniques are AMAP and L-systems and it
requires scientific measurements of the plants growth, size and shape to model a plant
in 3D and the modelling needs to be calibrated and validated. These accurate 3D
models are necessary to model vegetation succession or to make a conversion from a
cover percentage to the number of plants per square metre. The accurate modelling
techniques can also be used sketch based, but for instance Xfrog works faster if a
plant has to be 3D modelled. Photographs from different angles of the plant can serve
as the input for the sketch based modelling process.
This research has shown that 3D visualising a vegetation map with a computer
generated distribution and sketch based 3D plant models is possible. However,
vegetation maps with the plant species abundance described as cover percentage are
unsuitable for 3D visualisation, unless an easy method is developed to convert this
percentage to the number of plants per square meter. | |
