When to add a new process to a model – and when not: a marine biogeochemical perspective

Abstract

Models are critical tools for environmental science. They allow us to examine the limits of what we think we know and to project that knowledge into situations for which we have little or no data. They are by definition simplifications of reality. There are therefore inevitably times when it is necessary to consider adding a new process to a model that was previously omitted. Doing so may have consequences. It can increase model complexity, affect the time a model takes to run, impact the match between the model output and observations, and complicate comparison to previous studies using the model. How a decision is made on whether to add a process is no more objective than how a scientist might design a laboratory experiment. To illustrate this, we report on an event where a broad and diverse group of marine biogeochemists were invited to construct flowcharts to support making the decision of when to include a new process in a model. The flowcharts are used to illustrate both the complexity of factors that modellers must consider prior to making a decision on model development and the diversity of perspectives on how that decision should be reached. The purpose of this paper is not to provide a definitive protocol for making that decision. Instead, we argue that it is important to acknowledge that there is no objectively “best” approach and instead we discuss the flowcharts created as a means of encouraging modellers to think through why and how they are doing something. This may also hopefully guide observational scientists to understand why it may not always be appropriate to include a process they are studying in a model.