Basin-specific versus regional compaction models: quantifying uncertainty in exhumation estimates on the Australian North West Shelf

Exhumation strongly influences the structural, pressure and temperature evolution of sedimentary basins, and thus the formation and distribution of mineral and energy resources. It is commonly quantified using compaction-based methods that rely on sonic, porosity and thermal data to reconstruct uplift from maximum burial depths, typically via empirical relationships. However, these relationships are often calibrated for specific geological settings and then transferred elsewhere, and even region-specific models use parameters that vary within measurable ranges but are usually treated as exact. Data errors and unquantified parameter uncertainties can therefore propagate through the calculations, significantly compromising the reliability of exhumation estimates.

We previously developed a probabilistic compaction model for the Northern Carnarvon Basin (NCB) in the Australian North West Shelf (NWS) using sonic data from normally compacted, unexhumed intervals. Research shows that the dynamic evolution of the NWS basins has been shaped by multiple rifting and extensional phases and magmatic activities associated with Gondwana dispersal, and by later regional tilting linked to subduction along the northern margin. These complex histories imply significant basin-scale variability in subsidence and exhumation patterns, suggesting that NCB compaction behaviour may differ substantially from that in neighbouring basins such as Browse, Roebuck and Bonaparte. Here we extend probabilistic compaction analysis across these basins, deriving basin-specific shale compaction trends and comparing them to identify key similarities, differences and their geological controls. In doing so, we explicitly test whether a single “regional” compaction model is sufficient for exhumation analysis on the NWS, or whether basin-scale models are required.

Model robustness is evaluated using Markov chain Monte Carlo (MCMC) sampling, and uncertainty propagation is used to quantify the effect of parameter uncertainty on exhumation estimates. The NCB model shows strong similarity to Roebuck Basin trends but diverges from those of the Vulcan Sub-basin in the Bonaparte Basin. We attribute these differences to contrasting tectono-thermal histories, particularly the stronger influence of proximal subduction on the Bonaparte Basin. Our results indicate that while the NCB model may be cautiously transferable to the nearby Roebuck and parts of the Browse Basin, applying it to the distal Bonaparte Basin introduces substantial uncertainty. We demonstrate that, wherever data permit, basin-specific probabilistic compaction models are preferable to regional or global models for reliable exhumation analysis on complex passive margins such as the NWS.