PLEASE NOTE: Application deadline date 08 Jan 2024. Applications are no longer being accepted for this project
Project Overview
Climate models predict a more El Niño-like climate under global warming, but observations during the past 60 years suggest changes of opposite sign. Reconciling this discrepancy, climate records will be extended to 200 years, combining knowledge of physical processes and state-of-the-art Artificial Intelligence -based approaches, with the ultimate goal of improving climate projections.
Project Description
El Niño represents the biggest fluctuation in Earth’s climate system. It is characterized by weakening of the east-west temperature gradient across the Equatorial Pacific, which weakens zonal atmospheric circulation, increases mean surface temperature, and alters global hydrological cycles. Therefore, understanding how this temperature gradient will change in a warming climate has crucial implications for accurate climate projections.
Climate scientists, however, do not know the answer. Whereas physics-based climate models predict a more El Niño-like climate with weakening gradient, observational estimates from 1960--2022 suggest an enhanced gradient.
Different explanations have been proposed to understand this model-data discrepancy, including model biases and natural climate fluctuations (Lee et al., 2022). In this project, we will take a different perspective, extending the time span of high-quality observational data to 200 years, to seek more robust comparisons and conclusions. Improving data has previously reconciled many model-data discrepancies, including patterns of early 20th-century warming (Chan et al., Nature, 2019) and the decadal variability of Atlantic hurricanes (Chan et al., Science Advances, 2021). In this project, we will use sea surface temperature (SST) and its physically coupled quantity, sea-level pressure (SLP), to reconstruct historical tropical climate, with three objectives,
1. Correcting 1850-onward SLP records in the International Comprehensive Ocean-Atmosphere Dataset. SST records have been corrected by Chan, and a set of tools are ready to use;
2. Developing and applying AI-based techniques to reconstruct dynamically consistent tropical SST and SLP evolution since 1850; and
3. Investigating long-term trends in tropical Pacific climate using the updated reconstruction, compared with model simulations.
1. Developing and applying AI algorithms.
2. Advanced statistical analyses, e.g., Bayesian inference and data assimilation.
3. Big data analysis of climate observations and simulations in Python (preferred) or MATLAB.
4. High performance computing platforms.
5. Online and one-on-one courses on climate physics and dynamics.
6. The history of climate monitoring through meeting with domain experts.
This mix of Machine Learning/AI, statistics, physics, and history training will uniquely equip the student to carry out work across a wide range of topics on climate science and beyond. Opportunities exist to attend international conferences and travel to collaborators overseas.
· Lee, S., L’Heureux, M., Wittenberg, A. T., Seager, R., O’Gorman, P. A., & Johnson, N. C. (2022). On the future zonal contrasts of equatorial Pacific climate: Perspectives from Observations, Simulations, and Theories. npj Climate and Atmospheric Science, 5(1), 82.
https://www.nature.com/articles/s41612-022-00301-2
· Chan, D., Kent, E. C., Berry, D. I., & Huybers, P. (2019). Correcting datasets leads to more homogeneous early-twentieth-century sea surface warming. Nature, 571(7765), 393-397.
https://www.nature.com/articles/s41586-019-1349-2
· Kadow, C., Hall, D. M., & Ulbrich, U. (2020). Artificial intelligence reconstructs missing climate information. Nature Geoscience, 13(6), 408-413.
