Using a climate emulator to investigate orbital forcing and climatic changes during the Middle to Late Pleistocene and the next 1 million years
Seminar Room 1, Geographical Sciences
A BRIDGE seminar by Natalie.
Nb. This event is only open to University of Bristol staff and students.
The simulation of long-term (103-106 years) past and future climate has a range of useful applications, but state-of-the-art, relatively complex climate models such as General Circulation Models (GCMs) require large amounts of computing power and are relatively slow to run, meaning that simulations of more than a few thousand years (kyr) are generally not feasible. Instead, lower complexity models are often used (e.g. Earth system Models of Intermediate Complexity), however these generally include a larger number of parameterizations and/or a lower spatial resolution.
We have therefore developed a statistical emulator which is calibrated on the climate output of a more complex climate model (in this case a GCM) (Lord et al. 2017). The emulator can be used to simulate the global climate occurring in response to any combination of atmospheric CO2 concentration, orbital configuration, and global ice volume (within the sampled parameter limits), making it useful for addressing a wide range of climate-related questions. Here, we use it to examine the role of orbital forcing on East Asian monsoons during the last 300 kyr (Middle to Late Pleistocene). To do this, we compare emulated precipitation over South-East (SE) China to oxygen isotope data (δ18O) measured from cave speleothems, which is assumed to be a proxy for the SE Asian monsoon. We find that emulated precipitation is able to reproduce the δ18O record well, and that the majority of the modelled response of precipitation is due to orbital forcing, with varying amounts of sub-orbital variability being provided by ice sheets and atmospheric CO2 concentration. We also apply the emulator to project the evolution of climate over the next 1 million years for a range of anthropogenic CO2 emissions scenarios. Along with orbital variations, the emulator is forced with atmospheric CO2 concentration data, produced using a response function capturing the long-term response of the global carbon cycle to different total CO2 emissions (Lord et al. 2016), and relative sea level, projected using a relatively simple insolation threshold model which simulates transitions between glacial and interglacial states (Paillard 1998; Archer + Ganopolski 2005). Our modelling results suggest that the next glacial inception may be delayed by 50 kyr or longer depending on total CO2 emissions, resulting in significant variations in projected climate changes over the next approximately 300 kyr between different CO2 scenarios, particularly in regions that are within or close to an expected ice sheet margin.