Thermal history inversion from thermochronometric data and complementary information: New methods and recommended practices

Abstract

Thermal history inverse modeling has become one of the primary platforms for interpreting thermochronometric data. This paper introduces new features in the HeFTy software, and compares them with both earlier HeFTy versions and other programs. The approach for combining multiple goodness-of-fit tests into a single probability has been changed to Fisher's method, improving both statistical accuracy and ease of plain-language interpretation. The mechanism and guidelines for setting up Monte Carlo inverse modeling in HeFTy are reviewed, with particular attention to the role of allowing complex but geologically realistic time-temperature (t-T) paths. A new implementation of the controlled random search algorithm for paths with variable time spacing greatly accelerates finding solutions that fit the data while trying to also map the solution space as effectively as an unbiased Monte Carlo approach, so as to avoid providing an unrealistic impression of resolving power. A new time-depth (t-Z) modeling mode uses a 1-D thermal model to make a first-order conversion between depth and temperature, with depth change corresponding to erosion or deposition at the Earth surface. This mode approximates the thermal buffering effects of the crust, and allows temperature relationships between samples and the Earth surface to evolve in a physically appropriate fashion. These improvements enable multi-sample modeling along an elevation transect, including testing and constraining the timing of deformation, expressed as tilting, and topography development. In most cases, adding geological constraints during the modeling process is crucial for achieving meaningful results, as the resolving power of the thermochronometric data alone is usually limited, and the best-fitting results are not necessarily geologically realistic. A crucial step in interpreting any modeling result is to ascertain how the data and assumptions, including those embedded in the software, lead to the outcome obtained. These principles are illustrated using recent examples of modeling thermal histories associated with the Great Unconformity.