Seminars

Abstract: Environmental monitoring – traditionally relied on collecting point samples – is undergoing transformational changes with new technologies such as remote sensing, low-cost in situ sensors, and wireless data transmission. In parallel, simulation capabilities are advancing rapidly, predicting complex systems and quantifying the uncertainty in predictions. However, there are still significant challenges in integrating these multi-type multiscale datasets with model simulations. In particular, calibrating simulation results is often difficult to match all the time series at multiple locations due to natural heterogeneities and/or fine-scale processes not captured in conceptual models.

This talk presents two physics-informed machine learning (ML) applications for integrating multiple data streams and hydrological simulations. Rather than improving long-term predictability, the purpose is to support day-to-day operations and environmental management, and to provide decision-relevant metrics. The first application is real-time groundwater monitoring strategies with low-cost in situ sensors. Specifically, we aim to improve contaminant plume mapping and to provide real-time monitoring of plume stability by integrating physics – such as source locations, flow direction, and contaminant mobility – into the spatiotemporal interpolation of contaminant concentrations. We use a Fourier Neural Operator (FNO)-based emulator within a Bayesian hierarchical approach coupled with Gaussian process models. The second application is real-time near-surface soil moisture monitoring and forecasting based on a wireless sensing network for agriculture and urban stormwater management. We integrate real-time data streams with hydrological simulations using ensamble Kalman filters. Through these applications, we aim to accelerate the adoption of ML/AI techniques in real-world environmental applications and decision-making.

Bio: Haruko Wainwright is an assistant professor of Nuclear Science and Engineering, and Civil and Environmental Engineering at the Massachusetts Institute of Technology. She received her BEng in Engineering Physics from Kyoto University, Japan in 2003; her MS in nuclear engineering in 2006, her MA in statistics in 2010 and PhD in nuclear engineering in 2010 from the University of California, Berkeley. Before joining MIT, she was a Staff Scientist in the Earth and Environmental Sciences Area at Lawrence Berkeley National Laboratory, and an adjunct professor in Nuclear Engineering at the University of California, Berkeley. Her research focuses on environmental informatics, aiming to improve understanding and predictions in Earth and environmental systems through mechanistic modeling and machine learning.