Keywords
whole-farm modeling; bioeconomic modeling; model reuse; replication; validation; agent-based simulation; MPMAS
Start Date
6-7-2022 9:40 AM
End Date
6-7-2022 10:00 AM
Abstract
A well-established strand of agricultural economic agent-based modeling employs mathematical programming (MP) to simulate farming decisions (Berger Troost 2012; Kremmydas et al. 2018). More specifically, in the sense of the MoHuB framework’s conceptualization of human behavior (Schlüter et al. 2017), MP is used to implement the Selection component in these models. It can be combined with perception, knowledge, learning and value-oriented components to reflect different theories of decisionmaking and is often coupled to biophysical process models to evaluate human-environment feedback. We conceptualize different levels of component reuse in MP-based farm agent simulation (i.a. technical basis, coupling interfaces, components for between-decision dynamics and interactions, implementation patterns within farm MP problems, full parameterizations). We examine two decades of experience with component reuse within the MPMAS simulation framework (Berger 2001, Schreinemachers Berger 2011) encompassing both horizontal reuse across MPMAS model applications for different case studies or study areas and vertical reuse across analyses within the same application context but with different purposes and different degrees of integration and complexity. Examined applications range from individual farm planning (Mössinger et al. 2022), interactions between farmers and their producer organization (Latynskiy Berger 2017) to watershed-level simulations for policy analysis (Arnold et al. 2015). We analyze which model components could be frequently and successfully reused and which MPMAS model components had to be redesigned when being applied to different modeling contexts. We draw general lessons on reuse in models of farm land use and discuss recent major MPMAS revisions that facilitate the combination of (external) reusable building blocks with the well-tested core MPMAS functionality for dynamic MP-based simulation.
Horizontal and vertical model reuse in mathematical programmingbased farm agent simulation
A well-established strand of agricultural economic agent-based modeling employs mathematical programming (MP) to simulate farming decisions (Berger Troost 2012; Kremmydas et al. 2018). More specifically, in the sense of the MoHuB framework’s conceptualization of human behavior (Schlüter et al. 2017), MP is used to implement the Selection component in these models. It can be combined with perception, knowledge, learning and value-oriented components to reflect different theories of decisionmaking and is often coupled to biophysical process models to evaluate human-environment feedback. We conceptualize different levels of component reuse in MP-based farm agent simulation (i.a. technical basis, coupling interfaces, components for between-decision dynamics and interactions, implementation patterns within farm MP problems, full parameterizations). We examine two decades of experience with component reuse within the MPMAS simulation framework (Berger 2001, Schreinemachers Berger 2011) encompassing both horizontal reuse across MPMAS model applications for different case studies or study areas and vertical reuse across analyses within the same application context but with different purposes and different degrees of integration and complexity. Examined applications range from individual farm planning (Mössinger et al. 2022), interactions between farmers and their producer organization (Latynskiy Berger 2017) to watershed-level simulations for policy analysis (Arnold et al. 2015). We analyze which model components could be frequently and successfully reused and which MPMAS model components had to be redesigned when being applied to different modeling contexts. We draw general lessons on reuse in models of farm land use and discuss recent major MPMAS revisions that facilitate the combination of (external) reusable building blocks with the well-tested core MPMAS functionality for dynamic MP-based simulation.
Stream and Session
false