HR: 08:00h
AN: H41A-01 INVITED [PDF]
TI: Process and Form in Holocene Landscapes: Modelling Human and Climatic Impacts
AU: * Wainwright, J
EM: john.wainwright@kcl.ac.uk
AF: Environmental Monitoring and Modelling Research Group, Department of Geography, King's College London,
Strand, London, WC2R 2LS
United Kingdom
AB:
Observations suggest that Mediterranean environments are significantly more sensitive to human impacts that to climate change
through the Holocene period. However, many explanations of landscape change rest on interpretations of climatic variability
that are difficult to sustain. Erosion and sedimentation data support an interpretation relating to local- (and later
regional-) scale human modifications of the landscape with major and often irreversible impacts on the soil resource. For
example, on the Causse de Gramat, SW France, erosion rates have varied significantly over distances of several kilometres
under very similar geological, soil and climate conditions over the last 6,000 years.
Despite the realization that human activity significantly modifies the erosional landscape in such settings, most work aiming
at understanding these impacts has been very simplistic. The difficulty of estimating time-varying human impacts has
commonly led to the use of relatively basic scenario-based models, particularly over the longer term. Scenario-based
approaches suffer from two major problems. They are typically static, so that there is no feedback between the impact and
its consequences, even though the latter might often lead to major behavioural modifications. Secondly, there is an element
of circularity in the arguments used to generate scenarios for understanding past landform change, in that changes are known
to have happened, so that scenarios big enough to produce them are often generated without considering the range of possible
alternatives.
The approach presented here draws on agent-based models to investigate human interactions with the landscape. Agent-based
models use a "bottom-up" approach to the development of these interactions by investing decision-making capacities at the
level of individuals, whose behaviour is constrained by the nature of the landscape itself and by interaction with other
individuals within it. The model presented consists of four elements. First, a cellular landscape-evolution model
characterizes erosion processes, soil development and sedimentation. Secondly, a plant-growth model allows the evolution of
vegetated landscapes according to major plant functional types. Thirdly, animal activity within the landscape is simulated
using an agent-based approach. Animals are simulated as agents, using cellular automata rules governed by underlying
energetics. Fourthly, there are the human agents, simulated either using cellular automata, or using neural networks.
Landscape evolution is simulated as the interaction of all four of these levels.
Results presented will illustrate a number of key factors. The stability of animal and human populations within these
landscapes is not a foregone conclusion, leading to quite different outcomes for similar initial conditions. The model can
illustrate the potential for self-organization and the development of quite different sets of behavioural patterns. This
approach is useful in the investigation of how well past human-landscape interactions are understood, and allows a contrast
with explanations simply based on climate.
DE: 1803 Anthropogenic effects
DE: 1809 Desertification
DE: 1815 Erosion and sedimentation
DE: 1824 Geomorphology (1625)
SC: Hydrology [H]
MN: 2003 Fall Meeting