Organisms, rivers, and coalgebras
University of Bayreuth
Norwegian Institute for Forest and Landscape
Last modified: March 24, 2006
There is a vast literature on modelling runoff from watersheds using either process-oriented or empirical approaches. Although watersheds are primary examples for terrestrial ecosystems,
none of these attempts include organisms in these systems explicitly.
However, we show that biological interactions in the interior of these systems imprint
structure on observed output, using an information-theoretic approach to quantify information content
and complexity of time series. These interactions are not coverable by conventional model classes.
Thus, we introduce a new modeling paradigm focusing on behavior rather than structure or processes,
using a coalgebraic representation of the dynamical system. The observed patterns may be described
as avoidance of certain behavioural types in this framework. In an attempt to reconstruct the observed information-complexity relationship, it is demonstrated that algebraic algorithms are not sufficient, whereas rule substitution in context-dependent grammars from symbolic dynamics and Markov chains are promising. As an outlook towards applications, we seek a classification of observed behavior within this framework (colorings) according to attributes from geology, climate, and land use.