Neogene Australian hydroclimate variability exceeds model predictions

Australia’s Neogene hydroclimate evolved in response to continental drift and major global climate reorganizations, yet the magnitude and spatial structure of these changes remain poorly constrained. Moreover, the ability of climate models to reproduce Australian hydroclimate variability during deep-time warm periods has rarely been evaluated against geological data. Here, we reconstruct hydroclimate evolution across the Northwest Shelf of Australia over the past 23 million years using a continent-scale compilation of downhole natural gamma radiation (NGR) records and directly compare these reconstructions with climate model simulations.

We integrate NGR measurements from 105 industrial and scientific boreholes into a regionally coherent stratigraphic framework using automated Dynamic Time Warping, with biostratigraphic age control providing temporal calibration. High-resolution (~100 kyr) time-slice reconstructions reveal pronounced spatiotemporal variability in terrigenous input, reflecting changes in precipitation and continental runoff.

The reconstructions indicate persistently humid conditions during the Early Miocene, followed by an abrupt transition to widespread aridity at ~18-17 Ma. This major hydroclimate shift is not reflected in HadCM3 simulations, which instead suggest wetter conditions than those inferred from the NGR reconstruction across subtropical Australia during this interval. A subsequent increase in hydroclimate variability at ~6.5 Ma, marked by elevated NGR values, aligns with enhanced modeled precipitation and is consistent with an intensified Leeuwin Current and southward migration of the Intertropical Convergence Zone, pointing to a transient return to wetter conditions. Lower NGR values during the Late Pliocene indicate the onset of a transitional phase preceding the establishment of fully arid conditions by ~2.4 Ma.

Together, these results demonstrate that the magnitude and spatial complexity of Neogene Australian hydroclimate variability inferred from geological records exceed those predicted by state-of-the-art climate models. The pronounced data-model mismatch in the Early and Middle Miocene highlights persistent challenges in simulating regional hydroclimate responses in warmer-than-present greenhouse climates. These findings underscore the importance of geological benchmarks for evaluating model performance and improving projections of future hydroclimate change.