Machine Learning for Hydrological and Earth Systems
Dr. Shijie Jiang
Mission
In the face of increasing environmental and climatic pressures, understanding the drivers and responses of coupled Earth system processes is essential. The Machine Learning for Hydrological and Earth Systems (ML4HES) group, also part of the ELLIS Unit Jena, combines data-driven methods with physical and ecological knowledge to study water, energy, and carbon dynamics across hydrological, ecological, and climate systems. We aim to identify key drivers, interpret system interactions and feedbacks, and improve the representation of coupled dynamics to support process understanding under environmental change.
Our approach
To support system-level analysis of coupled processes, we integrate remote sensing and in-situ observations across spatial and temporal scales to extract meaningful signals. Leveraging recent advances in machine learning, we use knowledge-integrated, data-driven models, particularly hybrid and explainable approaches, to connect observations with underlying processes, identify drivers, interpret responses, and inform process-based modeling.
Focus areas
Land-atmosphere interactions
We investigate how vegetation redistributes water and energy across landscapes and into the atmosphere. Our work examines the biophysical processes that mediate land–atmosphere coupling, including both local and non-local effects, to enhance our understanding of hydroclimatic variability across various scales.
Soil-plant-atmosphere continuum
We study how vegetation interacts with soil and atmospheric conditions to regulate water movement through the continuum from root to canopy. Our research aims to characterize water fluxes and states and to understand plant water-use strategies under environmental stress.
Watershed hydro-biogeochemistry
We study hydrological extremes and riverine carbon dynamics in response to climate and landscape variability, including how extreme events emerge and how water flow regimes and land surface processes shape carbon fluxes and biogeochemical conditions. Our work analyzes patterns across regions and scales to understand potential mechanisms.
News
November 2025
[People] We are pleased to welcome Xiaoze Chen, who has started a 1.5-year research visit in our group. His work focuses on hybrid flood modelling. [Events] Shijie Jiang and Georgios Blougouras gave a keynote presentation at GIS Day 2025 "GeoAI for environment" at Friedrich Schiller University Jena.
October 2025
[Publication] We are pleased to share that our new study led by Shengyue Chen has been published in Geophysical Research Letters. Using an interpretable machine learning approach, we examine broad patterns in how terrestrial and riverine productivity vary across different environmental conditions. Diverse Environmental Factors Shape Patterns of Terrestrial and Riverine Productivity Decoupling
September 2025
[Event] Our group contributed to organizing the ELLIS Summer School “AI for Earth and Climate Sciences“, held in Jena from September 1–5, 2025. The summer school brought together PhD and MSc students as well as postdocs interested in the interface between machine learning and the geosciences.
July 2025
[Event] Shijie Jiang attended the ISMC–GEWEX SoilWat Meeting at the University of Reading. He gave a talk and chaired the session The Soil–Plant–Hydraulic–Energetic Continuum.
GENAI-X addresses a fundamental AI challenge: achieving robust model generalizability in non-stationary environmental systems, where conditions vary across space and evolve unpredictably. Focusing on hydro-climatic extremes (floods, landslides, droughts and late-frost events) and their impacts, we advance AI methods that adapt to shifting data patterns and uncertainties.
The project funded by Carl Zeiss Stiftung aims for excellent research in AI and will be located at the interface between AI and environmental research. The goal is to integrate expertise into AI approaches and to draw new insights from this, e.g., for climate extremes and their effects ecosystem functions or services.
Publications
2025
Wang, Lijun, L.; Shi, L.; Reimers, C.; Wang, Y.; He, L.; Wang, Y.; Reichstein, M.; Jiang, S.: A self-supervised deep learning model for enhanced generalization in soil moisture prediction. Journal of Hydrology 662 (Part B), 133974 (2025)
Slater, L.; Blougouras, G.; Deng, L.; Deng, Q.; Ford, E.; Hoek van Dijke, A. J.; Huang, F.; Jiang, S.; Liu, Y.; Moulds, S.et al.; Schepen, A.; Yin, J.; Zhang, B.: Challenges and opportunities of ML and explainable AI in large-sample hydrology. Philosophical Transactions of the Royal Society of London - Series A: Mathematical Physical and Engineering Sciences (383), 20240287 (2025)
Agathangelidis, I.; Blougouras, G.; Cartalis, C.; Polydoros, A.; Tzanis, C. G.; Philippopoulos, K.: Global climatology of the daytime surface cooling of urban parks using satellite observations. Geophysical Research Letters 52 (2), e2024GL112887 (2025)
Shi, Z.; Zhao, W.; Xiong, Y.; Lian, X.; Fang, J.; Jiang, S.; Yan, C.; Winkler, A.; Qiu, G. Y.: A theoretical framework for critical canopy temperature and its potential to improve remote-sensing-based estimation of evapotranspiration. Environmental Research Letters 21 (2), 024011 (2025)
