Design and demonstration of a data model to integrate agent-based and field-based modelling

Abstract

Dynamic environmental modelling of spatio-temporal systems often requires the representation of both fields and agents. Fields are continuous with values in the whole spatio-temporal domain of a model, while agents are bounded in space and often mobile. It is currently difficult for environmental modellers with limited software engineering background to construct such field-agent models, as modelling frameworks mostly do not support the integration of fields and agents. To overcome this issue, we describe a data model combining fields and agents in a single concept. This data model represents fields, agents and relations by grouping items sharing properties into a phenomenon. The concepts domain, property set and value handle spatio-temporal attribute representations. The data model is implemented in a software prototype that shows how data on fields and agents is stored and manipulated.

Integrating human behavior dynamics into drought risk assessment—A sociohydrologic, agent‐based approach

“…Another necessary choice requires balancing the trade‐offs between spatial resolution and domain size. Linking ABMs with existing physical models can also require extensive data processing (pre and post) and/or binding various languages such as R, FORTRAN, NetLogo, and MATLAB depending on the native formats and languages of the models selected (de Bakker, de Jong, Schmitz, & Karssenberg, ). Depending on the choice of hydrological model and agent‐based modeling language, two or more languages might need to be integrated, that cannot communicate with each other, hence, shell scripts may be needed.…”

“…Exposure describes the assets and activities located in hazard-prone areas (Birkmann et al, 2013;UNISDR, 2009a) and can be expressed as the number of people potentially impacted by water shortages (human exposure) (Kummu et al, 2016); the productive area prone to crop stress (agricultural exposure) (Murthy, Laxman, Sai, & Diwakar, 2014); or the ecosystems that could be harmed (environmental exposure) (Jalava, Kummu, Porkka, Siebert, & Varis, 2014). Due to droughts' large spatial extent, the exposure for a single event can be quite broad and time varying (CRED & UNISDR, 2018). This fact exponentiates when considering droughts can expose assets outside of a hazards explicit boundary to water deficiencies and economic strains.…”

A modelling framework for simulating large geographical systems of agents and fields

“…Whenever possible, time consuming conversions must be prevented, for example. Although in this chapter we focus on the physical data model and not on the representation of model state variables, our results align with the conceptual data model described in Bakker et al [8], which could be used as a basis for representing model state variables.…”

“…This grouping of spatio-temporal object information is similar to the conceptual data model presented in Bakker et al [8], whose design had a similar goal as our physical data model: to represent different kinds of state variables in a uniform manner in order to make it more convenient for environmental modellers to create models in which these different kinds of variables are manipulated. The conceptual data model is shown in Figure 3.11.…”

A Study on Data Processing Services for the Operation of Geo‐Analysis Models in the Open Web Environment

“…The use of geo‐analysis models enables the simulation and reproduction of geographic processes to make decisions and explore geographic rules. When considering complex geographic problems, typically, geo‐analysis models must be integrated to solve complex geographic problems due to the limited capacity of a single model (Overeem et al, ; Jones et al, ; Guzman et al, ; Belete, Voinov, & Morales, ; de Bakker et al, ; Buahin & Horsburgh, ; Marcot & Penman, ; Rossetto et al, ; Tian et al, ). After years of development, massive numbers of geo‐analysis models in multiple disciplines and domains have been produced all over the world (Goodchild, ; Goodall et al, ; Jagers, ; Voinov & Cerco, ; Granell, Schade, & Ostländer, ; Laniak et al, ; P. Yue et al, ).…”