The SimpopLocal model is an agent-based model using the NetLogo simulation platform.
This model was designed to study the emergence of a structured and hierarchical urban settlement system in the end Neolithic times by simulating the growth dynamics of a system of settlements under strong environmental constraints. The driving force behind the growth of settlements is demographic growth permitted by available resources but also increased through the societal innovation that improve labor productivity or enhance the amount of available and usable resources and thus the carrying capacity of each settlement. Nevertheless, this settlement system is vulnerable because susceptible to catastrophic hazards, such as climatic catastrophes (drought, famine, floods, etc.) or human catastrophes (invasions, war, etc.).
As in the other models of the Simpop family, the originality of the model is to consider settlements as entities at the micro-level and to model the emergence of a system of interdependent and hierarchical urban settlements at the macro-level. This emergence is produced by a differential in settlements growth rates which is not randomly distributed as in the Gibrat model but generated by the balance of exchange flows between settlements. These flows are represented as spatial interactions when settlements are exchanging information and artifacts stemming from societal (technological and organisational) innovations.
The SimpopLocal model is used as a benchmarking model for the conception of the Simprocess platform. Due to several none empirically determinable parameters and to several objectives in the behavior of the model, it is very difficult to manually calibrate the model. Hence, we are currently working on an automated protocol to realize a guided exploration using a fitness function in order to obtain a calibrated set of parameters. This guided exploration is using genetic algorithms as optimization method and a computer grid to reduce calculation time.
On this illustration, we can see the evolution and the progressive formation of a hierarchy in the settlement system during the simulation. In bleu, the hierarchy at the beginning of the simulation (homogenous distribution); in red the hierarchy at the end of the simulation (LogNormal distribution).
This illustration presents some results for an experiment on the impact of catastrophic hazard on the system’s hierarchical structure. The first type of graphics represents the evolution of the total population of the system during the simulation. The second type of graphics shows the linear evolution during time of each settlement according to its first ranking at the beginning of the simulation. The third type of graphics is the Rankclock visualization (tool developed at CASA) of the perturbation of the hierarchy due to catastrophic hazard.