Comparison of thermal-kinematic modeling approaches based on Taguchi method sensitivity analysis

Thermal-kinematic modeling of thermochronometric datasets is widely used to reconstruct exhumation histories. Yet, the sensitivity of model outputs to input parameters is rarely evaluated, even within a given modeling framework. Comparative sensitivity analysis across multiple tools is even less common, even though these models differ substantially in their treatment of heat transport, inversion structure, and kinetic behavior. In this study, we apply the Taguchi method, a statistical design of experiments, to assess parameter sensitivity in four thermal-kinematic modeling packages. We systematically vary thermal, kinetic, and optimization parameters across each tool and quantify their influence on predicted exhumation rates. Our results reveal substantial model-specific differences in sensitivity patterns. Two modeling platforms that use the same formulation, namely, Gaussian linear inversion of an age-elevation relationship, are most responsive to different things, with one most sensitive to thermal field parameters such as heat production and thermal diffusivity, and the other strongly influenced by inversion settings, including time step and prior exhumation rate. In age2exhume, activation energy (E_a) dominates, underscoring the role of kinetic parameters in diffusion-based models. These findings demonstrate that parameter sensitivity is not intrinsic to the thermochronometric system but is shaped by modeling assumptions. As tool selection influences both interpretation and uncertainty, there is a risk that model structure may overwhelm geological signals if not explicitly tested. We advocate for broader model intercomparisons and increased flexibility in parameter configuration to support more robust and transparent thermochronologic analysis.