Reproducible Integrated Modeling: It Takes a Village…and a Mayor
This talk will provide lessons learned in reproducible integrated modeling distilled from the experiences of the Integrated Multisector, Multiscale Modeling (IM3) project funded by the U.S. Department of Energy, Office of Science, as part of research in MultiSector Dynamics, Earth and Environmental Systems Modeling Program. The IM3 project has been developing and integrating human and Earth system models at multiple scales and with increasing complexity for approximately nine years. The project’s overarching goal is to improve scientific understanding of the co-evolution of multisectoral (e.g., energy, water, land, urban) risks and vulnerabilities due to climate and non-climate stressors at scales ranging from local to national within the United States. Integrated modeling challenges identified in IM3 include the following: many best-in-class models used in the project were designed originally for standalone use by a specific community of users; differences in spatial, temporal, and process resolutions between models; a large, transdisciplinary team across eleven collaborating institutions; varying levels of software engineering expertise across collaborators; disparate experiences with open science; and communication gaps between modeling teams. While we do not claim to have all the answers, IM3’s approach to reproducible integrated modeling has been commended by reviewers from multiple scientific journals. We will describe the combination of hands-on management, software engineering support, data and software management tools, communication strategies, collaboration tools, and open science requirements that we have employed to meet these challenges.