Changes in the Occurrence and Distribution of Extreme Precipitation Events at the Paleocene-Eocene Thermal Maximum (PETM)

Abstract:

The study of the sensitivity of the hydrological cycle to episodes of global warmth in the geologic past is receiving increased attention, but knowledge of the occurrence of hydrological extremes remains limited. A range of geomorphological, microfossil and biomarker proxies indicate significant hydrological change accompanied the PETM hyperthermal at ~55.8 Ma, many of which have been interpreted to reflect changes to Mean Annual Precipitation (MAP) or runoff. Recently, changes in the occurrence of intense, episodic precipitation has been suggested at some sites, but it is currently unknown whether such regions were particularly susceptible to extremes, or whether proxies from other regions require further interpretation. In this work, we seek to understand whether a numerical climate model is capable of simulating changes in the frequency and global distribution of intense precipitation events by analysing GCM-simulated hourly precipitation rates. Our Eocene simulations are performed at x2 and x4 preindustrial CO₂ using a coupled atmosphere-ocean GCM, HadCM3L. Climatological differences between high- and low-CO₂ may be considered analogous to the PETM. We find that changes in storm characteristics and extremes are highly regionalised. In particular, our simulations support that extreme events occurred with a reduced return period at the PETM in tropical regions of Africa and South America, whilst in the mid-latitudes the importance of extremes varies significantly between sites in close proximity. We also identify regions where changes in extreme behaviour are decoupled from those of MAP, which may represent important proxy acquisition targets. Given that tropical precipitation distributions are highly sensitive to GCM parameterisation scheme and given biases in the representation of sub-daily precipitation within HadCM3L, there is a clear need for further modelling work to fully characterise the Eocene hydrological cycle. However, our results indicate that the interpretation of existing proxies must consider the influences of both changes in mean annual precipitation rate, but also the occurrence of intense, high impact events.