Search for: All records

Compound flooding, driven by nonlinear interactions between multiple hydrometeorological factors, poses a significant challenge to hazard prevention. Existing forecasting approaches, whether physics-based or data-driven, often emphasize temporal patterns while underexploring how multiple interacting factors jointly shape flood dynamics. To address this problem, we conduct a large-scale data-driven analysis of compound flooding in South Florida, a typical area for compound flooding, by integrating tidal conditions, rainfall, groundwater stage, and human water management activities. Our analysis reveals three key findings: (i) models that capture temporal dynamics alone fail to represent multi-factor interactions during compound events; (ii) subsurface saturation, as reflected by groundwater levels, emerges as a dominant predictor of flood severity, often outweighing immediate rainfall intensity in this porous coastal region; and (iii) the spatial state of surrounding monitoring stations within a finite effective radius provides critical causal context for flooding, while extending temporal history yields diminishing returns during extreme events. These findings suggest that compound flooding is governed more by spatially coupled system states than by long-term temporal dependencies, challenging rain-centric and sequence-dominated forecasting paradigms. By framing data-driven models as tools for scientific inquiry rather than prediction alone, this study offers new insights into the mechanisms of compound flooding and informs the design of more physically grounded early-warning systems for coastal environments. 

Abstract Recent work has shown that the weakening of the Atlantic meridional overturning circulation (AMOC) under anthropogenic climate change is sensitive not only to the amount of atmospheric CO₂increase but also to the rate of CO₂increase. In this work, we explicitly test whether a previously proposed AMOC–sea ice feedback controls the AMOC’s sensitivity to the rate of CO₂increase using mechanism-denial experiments in a global climate model. We find that when the AMOC–sea ice feedback is suppressed, the AMOC is nearly insensitive to the rate of CO₂increase, supporting the hypothesis that this feedback is the primary reason for the AMOC’s sensitivity to the rate of CO₂increase. Additionally, we show that the reduction of sea ice meltwater in the North Atlantic that is caused by the retreat of Arctic sea ice in a warming climate constitutes a large strengthening forcing on the AMOC, reducing the total amount of AMOC weakening by nearly half. The strong influence of sea ice meltwater on the AMOC—both in its role as a forcing and a feedback—occurs through the ability of the meltwater to modulate how much surface water mass transformation occurs via ocean heat loss to the atmosphere over the deep convection zones in the North Atlantic. This work demonstrates that Arctic sea ice–AMOC interactions are a first-order control on future AMOC weakening and highlights the importance of constraining present-day sea ice meltwater fluxes to assess their potential counteracting influence on the weakening of Earth’s AMOC. Significance StatementIn this work, we aim to understand how Arctic sea ice influences the weakening of the Atlantic meridional overturning circulation (AMOC) under future anthropogenic greenhouse gas emissions. We find that when Arctic sea ice and the AMOC are allowed to mutually influence each other in a positive feedback cycle, the amount of AMOC weakening is highly sensitive to the rate of CO₂change (even when the CO₂concentration is the same). We also show that the retreat of sea ice under global warming can counteract some of the AMOC weakening induced by CO₂increases. This work highlights how understanding the fundamental mechanisms of AMOC–sea ice interactions is critical for constraining the future weakening of Earth’s AMOC. 

Accurate chemical reaction prediction is essential for drug discovery and synthetic planning. However, this task becomes particularly challenging in low-data scenarios, where novel reaction types lack sufficient training examples. To address this challenge, we propose FewRxn, a novel model-agnostic few-shot reaction prediction framework that enables rapid adaptation to unseen reaction types using only a few training samples. FewRxn integrates several key innovations, including segmentation masks for enhanced reactant representation, fingerprint embeddings for richer molecular context, and task-aware meta-learning for effective knowledge transfer. Through extensive evaluations, FewRxn achieves state-of-the-art accuracy in few-shot settings, significantly outperforming traditional fine-tuning methods. Additionally, our work provides insights into the impact of molecular representations on reaction knowledge transfer, demonstrating that knowledge captured under molecular graph-based formulation consistently outperforms those learned in forms of SMILES generation in few-shot learning.