Computer Simulation and Open Source Code

The data from the seismic mapping, sedaDNA and palaeoenvironmental analysis can then be used to build dynamic models of the changing geomorphology and ecology of Doggerland, from the opening of the Holocene around 10,000BC until its eventual total inundation around 5,500BC. In an ambitious departure from conventional approaches, the intention is to use complexity systems modelling approaches such as Agent-Based Modelling (ABM), for the modelling of ecological processes and simulating the dynamic interaction between the environment and the animals and plants which inhabit it.

This cannot be achieved by conventional modelling methods, which are both static and 'top-down'. The complex processes that emerge from the interaction between species and their effect on the environment require a modelling method which can simulate change through time and is able to simulate the dynamic interactions of a changing landscape and the flora and fauna which depend on it. ABM mimics nature's decentralised, 'bottom up' processes, taking individual organisms as its building blocks. It simulates individual plants and animals and the rules which govern their interactions with each other and the environment. ABM allows different hypotheses and 'what if?' scenarios to be explored through dynamic simulations, and processes such as plant and animal colonisation and the interactions between flora and fauna to be modelled in detail. With this we can simulate the effects of individual variation, spatial processes, variation in growth patterns, cumulative stress, and natural complexities with fidelity and levels of detail that is difficult or even impossible with conventional models.

While the principles of complex systems modelling are well established and the approach has been used by the group, the scale of the modelling which will be attempted for Doggerland has little precedent, and developing the appropriate methodologies and software tools will form a significant part of the work. This will take advantage of the group's piineering work on distributed simulation infrastructure to create a system whereby the models can be processed across networks of computers simultaneously.

Open source code prepared for the project and links to code repositories can be found here

Professor Eugene Ch'ng