Into the Abyss: Combining Deep Ocean Observations and ML to Reduce Sea Level Rise Uncertainty
Presented by Dr. Ratnaksha Lele
Wednesday, September 4, 2024 · 12 - 1 PM
Join us virtually on Wednesday, September 4 at 12p for a research talk.
Guest Speaker
Ratnaksha Lele, PhD, postdoctoral scientist at Scripps Institution of Oceanography, UC San Diego
Title
Into the Abyss: Combining Deep Ocean Observations and ML to Reduce Sea Level Rise Uncertainty
Abstract
Quantifying future estimates of sea level rise is a key focus in understanding climate impacts.Although contemporary global mean sea level changes are well observed and resolved from robust and sustained global satellite and in-situ measurements, there are substantial gaps in our understanding of regional and local sea level variability and projection of future sea level rise (SLR). This is in-part attributable to the complexities in the interaction of different components of SLR on local scales, as well as a dearth of observations that can resolve the variability in these components at a high spatiotemporal resolution. The world’s deep and abyssal oceans are one such challenging frontier. In this talk- I will present insights into deep ocean warming, ocean heat content change and quantify steric sea level rise estimates using high resolution data collected over a decade from a suite of autonomous robotic floats called Deep Argo in the Southwest Pacific Basin. Data collected from Deep Argo shed light not only on spatiotemporal variability in ocean heat content in the deep ocean, but also crucially quantify impacts of the warming on SLR trends and uncertainty. Next, I will present some new work on the applicability of predictive machine learning (ML) techniques to these novel data collected in the data sparse regions of the deep ocean, in order to improve spatiotemporal coverage of physical signals of ocean warming, ocean heat content and SLR variability globally. Incorporation of ML in this framework has the potential to improve our understanding of global SLR changes, as well as uncertainty in local SLR projections for coastal communities in future climate scenarios.
Bio
Ratnaksha Lele is a postdoctoral scientist at Scripps Institution of Oceanography, UC San Diego. As a physical oceanographer his work involves the synthesis and interpretation of observational in-situ and remote sensing data to understand large-scale dynamics, circulation and spatiotemporal variability in the deep and abyssal ocean, and fundamentally its role in the global climate system. His recent work involves using data from novel autonomous profiling floats called Deep Argo, which measure fundamental oceanographic properties in near real-time from the surface ocean to up to 20000ft deep, to study changes in the ocean’s heat content and improve estimates of local and regional sea level rise. He also has experience working as a data scientist in industry at Jupiter Intelligence and Corteva Agriscience where he worked on designing and implementing machine learning methods at scale, and applied it to improve estimates of flood risk in future climate scenarios and probabilistic corn yield prediction across the North American corn-belt. Ratnaksha enjoys thinking about applied science problems at the nexus of AI, Climate and Society and is keen to pursue R&D activities in that realm.