Embedded Files
Machine Learning Webinar
Machine Learning Webinar
In the first part of the webinar, we will demonstrate how machine learning (ML) can be used to predict GRACE satellite-derived terrestrial water storage (TWS) anomalies using monthly evapotranspiration, precipitation, and soil moisture data. The session will cover data preparation techniques, with a focus on capturing spatial-temporal hydrological patterns. We will walk through the application of various ML models, such as random forest, support vector machine, gradient boosting and other, highlighting their strengths and limitations. Participants will learn how to select key predictive features, evaluate model performance using cross-validation, and apply these methods for groundwater and hydrological assessments.
The second part will be devoted to groundwater level modeling using ML. A short introduction will be presented, as well as case studies and experience from The Groundwater Modeling Challenge (https://github.com/gwmodeling/challenge). Advantages and limitations of applying ML for groundwater level prediction will also be discussed.
By the end of the webinar, attendees will gain practical insights into AI-driven hydrology and its applications in water resource management.
Machine Learning to Predict GRACE satellite-derived terrestrial water storage (TWS)
Machine Learning to Predict GRACE satellite-derived terrestrial water storage (TWS)
Poland Case study files
Vytautas Samalavičius (vytautas.samalavicius@chgf.vu.lt) is an assistant professor in the Department of Hydrogeology and Engineering Geology at Vilnius University’s Institute of Geosciences. His research focuses on groundwater studies, with a strong emphasis on applying machine learning to geological challenges, e.g. AI-driven methods for predicting groundwater isotopic composition and improving filtration coefficient calculations in soil studies.
As a principal investigator in the international EAGER IMPRESS-U project, Vytautas is currently developing machine learning algorithms for groundwater resilience assessment across Lithuania, Poland, Ukraine, Latvia, Estonia, and the U.S. His work integrates artificial intelligence with hydrodynamic modeling to enhance the understanding of subsurface water systems.
Beyond research, Vytautas is dedicated to education, incorporating the latest scientific advancements into his teaching while mentoring students in AI applications for hydrogeology. He also contributes to public science outreach, coordinating national competitions and delivering educational lectures on groundwater sciences and environmental sustainability. https://www.researchgate.net/profile/Vytautas-Samalavicius/research
Groundwater level modeling using machine learning (ML)
Groundwater level modeling using machine learning (ML)
Jānis Bikše (janis.bikse@lu.lv) is a researcher and PhD candidate at the University of Latvia specializing in hydrogeology and environmental sciences. His work focuses on groundwater dynamics, resilience, and sustainable water management. He applies statistical and machine learning methods in R to analyze hydrological data and has experience with GIS-based spatial analysis and modeling.
Page updated
Google Sites
Report abuse