Abhiraami Navaneethanathan(She/Her)University of Exeter
Abhiraami Navaneethanathan is a PhD student at the University of Exeter as part of the UKRI Centre for Doctoral Training in Environmental Intelligence, and her research is in partnership with the National Oceanography Centre. Her research focuses on better understanding the processes that influence the ocean biological carbon pump by applying data-driven methods to in situ particulate organic carbon (POC) flux observations from a variety of instruments. She is interested in the use of data fusion methods to leverage heterogeneous data sources that can produce more accurate estimates and reveal insights into the underlying processes within complex environmental systems. She holds an MSci from Imperial College London in Theoretical Physics. |
 |
TalksEstimating global ocean POC fluxes through machine learning and data fusion on heterogeneous and sparse in-situ observations24-Apr-24The ocean biological carbon pump, a significant set of processes in the global carbon cycle, drives the sinking of particulate organic carbon (POC) towards the deep ocean. Models trained to predict global POC fluxes can advance our understanding of how environmental factors influence organic ocean carbon transport, in addition to helping quantify how much carbon is sequestered in the ocean and how nutrients are distributed to different marine ecosystems. POC fluxes can be derived from observations taken by a variety of in-situ instruments such as sediment traps, 234-Thorium tracers and Underwater Vision Profilers. However, the manual and time-consuming nature of data collection poses challenges regarding spatial data sparsity on a global scale, resulting in large estimate uncertainties in under-sampled regions. This research takes an observation-driven approach with machine learning and statistical models trained to estimate POC fluxes globally using in-situ observations and well-sampled environmental driver datasets as predictors, such as temperature and nutrient concentrations. This approach can both fill observational gaps spatiotemporally and reveal the importance of each environmental predictor for estimating POC fluxes. The models built include random forests and neural networks, where their global POC flux estimates, feature importance and model performances are studied and compared. Additionally, this research explores the use of data fusion methods to combine all three heterogeneous in-situ POC flux data sources to achieve improved accuracy and better-informed inferences about organic carbon transport than what is possible using a single data source. By treating the heterogeneous data sources differently, accounting for their uncertainties, and introducing domain knowledge, the proposed Bayesian hierarchical data fusion model can not only harness the information from all three data sources, but also gives insights into their key differences. |