Data models and scripting languages for field-agent based modelling

Schmitz1, O., Verstegen2, J.A., Karssenberg1, D.

1Faculty of Geosciences, Dept Physical Geography, Utrecht University, the Netherlands 2Faculty of Geosciences, Dept Human Geography and Spatial Planning, Utrecht University, the Netherlands

Workshop overview

We are organising this half-day workshop at the GIScience 2023 conference, https://giscience2023.github.io/ on September 12, 2023 in the morning session.

Preliminary schedule:

9:00 - 9:05 Welcome and introduction
9:05 - 9:30 Presentation: Geosimulation using fields and agents
9:30 - 10:30 Campo hands-on tutorial and exercises
10:30 - 10:45 Break
10:45 - 11:15 Roundtable
11:15 - 11:35 Discussion
11:35 - 12:25 LUE tutorial and exercises
12:25 - 12:30 Roundup

Software installation:

You will need Conda to create a Python environment with the software required for the workshop. See https://github.com/computationalgeography/giscience2023 for further information.

Topic

Many simulation models needed for instance for hydrological, ecological, or health studies require the combination of fields representing spatially distributed values (e.g. elevation, catchments, land use or contamination) and agents representing individual objects (e.g. trees, agricultural farms, or humans).

This workshop invites participants dealing with the challenge of combining fields and agents in their simulation models to give a lightning talk about their work. Furthermore, it provides hands-on exercises for Campo, a Python modelling environment providing operations accepting both fields and agents as arguments. Campo resembles and extends the map algebra approach to field-agent modelling and allows for the construction of static or dynamic models. Campo builds upon LUE, a conceptual and physical data model for storage and access of field or agent data.

Call for participation

When constructing simulation models of heterogeneous environmental systems it is required to incorporate phenomena that are represented as spatially and temporally continuous fields as well as phenomena that are modelled as spatially and temporally bounded agents. An example are humans (agents) moving through a city and thereby being exposed to air pollution (fields). Such phenomena require representation of multiple (potentially mobile or overlapping) objects that each exist in a subdomain of the space-time domain of interest. Storage and access of different phenomena requires an approach that integrates representation of fields and agents in a single data model.

In this workshop we share ideas on the design and implementation of our open-source LUE data model and Campo modelling environment for the development of field-agent based models. LUE contains an data model that explicitly stores and separates domain information, i.e. where phenomena exist in the space-time domain, and property information, i.e. what attribute value the phenomenon has at a particular space-time location, for a particular agent. The Campo module implemented in Python provides operations accepting both fields and agents as arguments. Domain specialists can therewith use a map algebra like approach to develop field-agent models.

The workshop will combine lightning talks of participants, presentations introducing the conceptual and physical data model, hands-on computer practicals using Jupyter notebooks, and discussions on the challenges of field-agent based modelling.

Expression of interest

Participants interested in giving a lightning talk are requested to send a short expression of interest by email to the workshop organizers.

Contact

For further information please contact o.schmitz@uu.nl or j.a.verstegen@uu.nl.