Revealing Dynamics: A Graphical Approach to Sensitivity Analysis in Complex Systems, Demonstrated with Compartmental Models

Abstract

Understanding the dynamics of model behavior in response to changes in input parameters is pivotal, especially within complex models featuring an array of input factors. Sensitivity Analysis (SA) serves as a fundamental methodology for elucidating and quantifying the uctuations in model behavior corresponding to variations in model input factors. In situations where models incorporate a wide range of input factors, identifying the most in uential variables is of utmost importance. Although the employment of graphical tools to encapsulate SA findings is gaining traction, it remains a relatively new and evolving approach. The advancement of graphical representations significantly enhances the understanding of SA outcomes. Within this work, an exploration into the efficacy of two modern graphical techniques, specifically star plots and dot charts, as tools for SA is undertaken. These visual aids provide a clear representation of key input factors, thereby accelerating the comprehension process. To showcase their utility in SA,
these techniques are applied to two distinct compartmental models elucidating the dynamics of the global carbon cycle.