Plant (growth) models in the context of citizens observation networks.

Problem definition: How to apply knowledge based interpretation to citizens observations of plant phenology.

Key words:

3d plant models, time development, model inversion, Bayesian parameter estimation. Biomathematics. Biochemistry, Temperature, Moisture .

In the context of link , which illustrates datacollection about plant phenology, a friend mentioned statistical analysis of the [space, time , temperature, phenology] data.

As a proponent of model based data analysis as a way of making use of knowledge in science, engineering and domain expertise, I started to collect some literature or plant modelling.

The basic approach (around 1985) was to model plants in 3D and time, produce ray tracing models, given camera models, including satellite multispectral scanners thus producing predicted image data given model parameters.

The likely actual parameters of objects, processes, sensors ,atmosphere, environment , … were calculated by Bayesian model inversion.
By adjusting the parameters in a maximum likelihood sense, the model is fit to all available model knowledge , measurements and observations.

At the time (1980 .. 1990) it was not trivial to obtain the required software or get access to “cpu time” . Meanwhile this problem does not exist . For a variety of plants, 3d models at various stages of growth can be ordered.

Plant models :
Ref. author: Przemyslaw Prusinkiewicz.
Keywords: Lindenmayer Systems, L-systems, Fractals, Plants, botanical trees,Graph theory, flowering, inflorescences, racemes, cymes, stochastic state transition, context sensitive L-systems .

Book: Przemyslaw Prusinkiewicz & James Hanan, Lindenmayer Systems, Fractals and plants, Lecture Notes in Biomathematics, Springer Verlag, 1989 .

Research proposal:
Relate the model parameters of L-system 4D plant models to environmental factors such as temperature, available photon flux, soil moisture and stress factors such as insect infestation.

Following selected plants in detail , using recording weather stations, spectroCams *, soil moisture sensors, air humidity sensors, will calibrate the models.

The calibrated models can then be used to make scientific sense of the large volume of citizens generated data.

* a spectroCam in our definition is a combination of an RGB camera for spatial pattern classification and generation of 2.5 D spatial models and a 2048 or 1024 channel spectrometer for the estimation of parameters of the biochemical processes in plants. It is a tool for spatial, temporal, spectral pattern recognition.

Nanno J. Mulder
M3X, prof. emeritus, RS data analysis.