Modelling for reproducibility, transparency, and continuous learning: Revisiting Jakeman’s 10 steps

Numerical modelling has become the standard tool to evaluate vast amounts of data, both for the public management of natural resources by government agencies and to understand environmental interactions within academia. While an ever-increasing amount of software tools is being developed and approaches are formulated how modellers should interact with stakeholders, the modelling community has not yet addressed criteria how these tools and approaches are applied costeffectively. Across the board, public agencies and academia continue to experience budget and time overruns that are systematic and need to be addressed if environmental modelling shall provide the benefits that modellers claim. This paper discusses public sector requirements such as reproducibility and transparency, lays out current approaches and practical challenges, and suggests a framework for organizing the modelling process that is derived both from public sector agencies and from integrated environmental modelling in academia. This paper identifies a multi-step layout of modelling studies (e.g. following Jakeman 2006) as core shortcoming, because project managers are mislead into translating these steps directly into project proposals, Gant charts, and budgeting. Authors always acknowledge that in practice, projects will have to revisit earlier steps in order to correct or refine assumptions, which basic project planning has not been accounting for adequately. With linear, multistep project planning approach to a task that is indeed iterative or circular, extra costs arise from access to knowledge, inadequate design of software tools and intellectual property rights to these tools that often don’t foster iterative work, and inadequate workflows. This paper suggest an alternative, circular framing to modelling projects and lays out requirements for the design of software tools, intellectual property rights, and the roles of knowledge and staff.