H069
Frontier AI Models Transforming Water Science
Session ID#: 249965
Session Description:
Frontier artificial intelligence (AI) models, including transformers, diffusion models, and long short-term memory (LSTM), are advancing our ability to model, interpret, and predict complex hydrological processes. This session highlights cutting-edge applications of AI in water science, with a focus on hydrologic modeling, uncertainty quantification, model interpretability, and the integration of physical knowledge into data-driven frameworks. We welcome submissions on topics including, but not limited to: (1) Applications of AI models, such as transformers, LSTM, diffusion models, and large-scale pretrained foundation models for improving model accuracy and transferability; (2) Interpretable AI approaches for enhancing transparency in water modeling; (3) Uncertainty quantification in water science applications; (4) Use of large language models (LLMs) for information extraction or decision support in water resource management; (5) Integration of AI with multi-source data, such as remote sensing; and (6) Hybrid models that integrate physics‑based hydrological approaches with deep learning techniques.
Index Terms:
1805 Computational hydrology [HYDROLOGY]
1839 Hydrologic scaling [HYDROLOGY]
1847 Modeling [HYDROLOGY]
1873 Uncertainty assessment [HYDROLOGY]
Primary Convener: Jiangtao Liu, Pennsylvania State University, Department of Civil and Environmental Engineering, University Park, United States
Conveners: Chaopeng Shen, Pennsylvania State University Main Campus, Department of Civil and Environmental Engineering, University Park, United States, Dan Lu, Oak Ridge National Laboratory, Oak Ridge, United States and Yalan Song, Pennsylvania State University Main Campus, Department of Civil and Environmental Engineering, University Park, PA, United States
Student/Early Career Convener: Yuan Yang, University of California San Diego, Scripps Institution of Oceanography, Center for Western Weather and Water Extremes, La Jolla, United States
Improved Multi-Domain Environmental Forecasting Through Landscape-Based AI Modeling (1954827)
Nicholas Rigdeway Kraabel, Pennsylvania State University Main Campus, Civil and Environmental Engineering Department, University Park, United States, Jiangtao Liu, Pennsylvania State University, Department of Civil and Environmental Engineering, University Park, United States, Yuchen Bian, Amazon Search, Palo Alto, United States, Daniel Kifer, Pennsylvania State University, Department of Computer Science & Engineering, University Park, United States, Chaopeng Shen, Pennsylvania State University Main Campus, Department of Civil and Environmental Engineering, University Park, United States and Chaopeng Shen
AI-Augmented Floodplain Management: Integrating Conversational Agents, Adaptive Training, and Multi-Agent Decision Support for Resilient Water Resource Planning (1955137)
Ibrahim Demir1, Yusuf Sermet1, Ramteja Sajja1 and Vinay Pursnani2, (1)Tulane University, New Orleans, United States, (2)University of Iowa, Iowa City, United States
Scalable and Accurate Global River Streamflow Forecasting with Transformers (1959748)
Jinyang Li1,2, Hayato Ushijima Mwesigwa1, Kuo-lin Hsu2 and Soroosh Sorooshian2, (1)Fujitsu Research of America, San Jose, United States, (2)University of California, Irvine, Department of Civil and Environmental Engineering, Irvine, United States
Design of An LLM-based Agentic Framework for National Water Model Data Access, Insight Generation, and Decision Support (1968788)
Sujan Chandra Mondol, Brigham Young University, Civil and Construction Engineering, Provo, United States and Daniel P Ames, Brigham Young University, Civil and Environmental Engineering, Provo, United States
Beyond Lumped Inputs: Transformer Models and Spatially-Explicit Data Integration for Streamflow Forecasting in Arizona (1968999)
Kshitij Dahal1, Saurav Kumar2 and Laxman Bokati2, (1)Arizona State University, School of Sustainable Engineering and the Built Environment, Tempe, AZ, United States, (2)Arizona State University, School of Sustainable Engineering and the Built Environment, Tempe, United States
From Text to Insight: Enhancing LLM Accuracy in Hydrology with Hybrid Retrieval and Knowledge Graphs (1975119)
Aditya Sharma, University of Southern Mississippi, Stennis Space Center, MS, United States, Md Ubayeid Ullah, University of Alabama, Tuscaloosa, United States, Jiaqi Gong, University of Alabama, Department of Computer Science, Tuscaloosa, United States, Steven J Burian, Cooperative Institute for Research to Operations in Hydrology (CIROH), Alabama Water Institute, The University of Alabama, Tuscaloosa, AL, Alabama Water Institute, Tuscaloosa, United States and Travis Loof, The University of Alabama, Tuscaloosa, United States
Dam level forecasting for water supply management via advanced time-series analysis: Colombia’s Andean region case study (1980361)
Antonio Preziosi-Ribero, Universidad Distrital Francisco José de Caldas, Bogota, Colombia, Maria Cristina Arenas, Pontificia Universidad Javeriana, Territory and Ecology, Bogota, Colombia and Paula Catalina Bogotá Parra, Universidad Santo Tomás, Civil engineering, Bogotá, Colombia
Applying LSTM Networks to Estimate Hydrographs in Ungauged Intermittent Streams in the Pacific Northwest (1990405)
Hannah Steele, Oregon State University, Corvallis, OR, United States, Melanie Davis, U.S. Geological Survey, Oregon Cooperative Fish and Wildlife Research Unit, Corvallis, United States and Mark S Raleigh, Oregon State University, College of Earth, Ocean, and Atmospheric Sciences, Corvallis, OR, United States
The Role of Uncertainty Quantification Algorithms in Deep Neural Network Models for Flood Prediction (1995199)
Mostafa Saberian, Clemson University, Department of Civil Engineering, Clemson, United States, Vidya Samadi, Clemson University, Department of Agricultural Sciences, Clemson, SC, United States, Thorsten Wagener, University of Potsdam, Institute of Environmental Science and Geography, Potsdam, Germany and Ioana Popescu, Delft University of Technology, Civil Engineering and Geosciences, Delft, Netherlands
A Deep State Space Model for Rainfall-Runoff Simulations (1875887)
Yihan Wang1, Lujun Zhang1, Annan Yu2, N. Benjamin Erichson3 and Tiantian Yang1, (1)University of Oklahoma Norman Campus, School of Civil Engineering and Environmental Science, Norman, United States, (2)Cornell University, Center for Applied Mathematics, Ithaca, NY, United States, (3)Lawrence Berkeley National Laboratory, Berkeley, United States
AGFormer: Adaptive Spatiotemporal Graph Learning for Enhanced Multi-Reservoir Inflow Forecasting (1894120)
Ming Fan1, Pengfei Hu2, Dan Lu2, Mr. Hyungyu Kang, PhD2 and Wei Zhang2, (1)Oak Ridge National Laboratory, Oak Ridge, TN, United States, (2)Oak Ridge National Laboratory, Oak Ridge, United States
Cloud Dynamics to Soil Moisture: AI-Enabled Shortcuts in Hydrologic Prediction (Invited) (1866228)
Jonathan M Frame, University of Alabama, Department of Geological Sciences, Tuscaloosa, AL, United States and Savalan Naser Neisary, The University of Alabama, Department of Civil, Construction, and Environmental Engineering, Tuscaloosa, United States
Deep Learning and Reinforcement Learning for Reservoir Storage Forecasting Enhanced by SWOT: A Comparison with NWM–ESP (1888672)
Xiaoyu Ma1, Athena Liu2 and Dennis P. Lettenmaier1, (1)University of California Los Angeles, Department of Geography, Los Angeles, CA, United States, (2)Duke University, Interdisciplinary Data Science, Durham, United States
Enhancing Interpretability in Hydrological Modeling: Explainable AI for Pan Evaporation Prediction in Kerala using Ensemble Machine Learning (1861467)
Sherin Babu, Assumption College Autonomous, Changanassery, Computer Science, Kottayam, Kerala, India and Binu Thomas, Marian College Autonomous, Kuttikkanam, Computer Applications, Idukki, Kerala, India
Exploring the Potential and Limitations of Machine Learning for Forecasting Groundwater Recharge from Meteorological Time Series (1900224)
Florian Lam1, Simon Damm2, Asja Fischer2 and Thomas Heinze1, (1)Ruhr-University Bochum, Hydrogeology and Environmental Geology, Bochum, Germany, (2)Ruhr-University Bochum, Machine Learning, Bochum, Germany
Inclusion of Blockage Effects for Scour Depth and Width Prediction using Machine Learning (1947922)
Priscilla Williams, University of Iowa, Department of Civil and Environmental Engineering, Iowa City, United States, Mia Marrocco, University of Windsor, Civil and Environmental Engineering, Windsor, ON, Canada and Ram Balachandar, University of Windsor, Windsor, ON, Canada
Integrated Probabilistic Forecasting Framework for Long-Term Reservoir Outflow through Dynamic Coupling of Meteorological–Hydrological–Engineering Processes (1864378)
Jiaying Tan, Bin Xu, Jian Zhu, Ping-an Zhong, Ran Mo, Jiangyuan Li, Yuanheng Dong, Xinman Qin, Jiayi Jiang, Huili Wang and Lingwei Zhu, Hohai University, Nanjing, China
Learning Data-Driven Hydrological Signatures from Daily Runoff with Contrastive Learning (1938172)
Yang Yang1,2, Yuan-Heng Wang3, Hoshin Gupta4, Shuxin Luo5,6 and Ting Fong May Chui2, (1)University of Massachusetts Boston, School for the Environment, Boston, United States, (2)The University of Hong Kong, Department of Civil Engineering, Hong Kong SAR, China, (3)Lawrence Berkeley National Laboratory, Earth and Environmental Sciences Area, Berkeley, United States, (4)Hydrology and Atmospheric Sciences, The University of Arizona, Tucson, United States, (5)Southern University of Science and Technology, School of Environmental Science and Engineering, Shenzhen, China, (6)Eastern Institute of Technology, Zhejiang Key Laboratory of Industrial Intelligence and Digital Twin, Ningbo, China
One to Three-Day Ahead Streamflow Forecasting Using Multi-Head Attention with LSTM (1936639)
Hiren Solanki, Indian Institute of Technology Gandhinagar, Earth Sciences, Ahmedabad, India, A S Aravinthakshan, Manipal Institute of Technology, Computer Science Engineering, Manipal, India, Sayuj Gupta, Indian Institute of Technology Jammu, Computer Science Engineering, Jammu, India and Vimal Mishra, Indian Institute of Technology Gandhinagar, Department of Civil Engineering, Gandhinagar, India
Overcoming a hurdle in probabilistic deep learning for hydrology (Invited) (1869466)
John Quilty, University of Waterloo, Civil and Environmental Engineering, Waterloo, ON, Canada and Mohammad Sina Jahangir, McGill University, Bioresource Engineering, Montreal, QC, Canada
Probabilistic Diffusion Models Advance Extreme Flood Forecasting (1891825)
Zhigang Ou1, Congyi Nai2, Baoxiang Pan3, Yi Zheng1, Chaopeng Shen4, Peishi Jiang5, Xingcai Liu6, Qiuhong Tang7, Wenqing Li8 and Ming Pan9, (1)Southern University of Science and Technology, School of Environmental Science and Engineering, Shenzhen, China, (2)Institute of Atmospheric Physics, Chinese Academy of Science, Beijing, China, (3)Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China, (4)Pennsylvania State University Main Campus, Department of Civil and Environmental Engineering, University Park, United States, (5)Pacific Northwest National Laboratory, Richland, United States, (6)Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China, (7)Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing, China, (8)China Institute of Water Resources and Hydropower Research, State Key Laboratory of Stimulation and Regulation of Water Cycles in River Basin, Beijing, China, (9)Center for Western Weather and Water Extremes (CW3E), Scripps Institution of Oceanography, University of California San Diego, La Jolla, United States
