Projects
We're currently working on water quality ensemble modeling and Knowledge-Guided Machine Learning. See more details below!
Uncertainty in Lake Ecosystem Ensemble Modeling
Various one-dimensional vertical, process-based aquatic ecosystem models exist that couple hydrodynamic with water quality models to replicate ecosystem dynamics. Although computationally similar, these models differ regarding their specific assumptions and mathematical formulations. In the last decades, several studies have pointed out that a systematic comparison of this plethora of aquatic ecosystem models is missing, as well as an evaluation of their accuracy and uncertainty. Building on the estabslished software framework for hydrodynamic ensemble modeling of lakes in R – the LakeEnsemblR package – we are working on an updated ensemble framework to run aquatic ecosystem models, assess their uncertainty on multiple ecosystem metrics, and improve overall ensemble model confidence.
​
This project is supported by an AUFF Recruiting Grant (2024-2026).
Curious? Contact Tuba.
Integrating AI into Aquatic Ecosystem Models to Decode Ecological Complexity
One approach is to apply the emerging scientific paradigm of “Knowledge-Guided Machine Learning” (KGML), which focuses on creating hybrid models that combine process-based principles (scientific theory) with data-driven deep learning models. Modular compositional learning (MCL) is a novel design choice for the development of hybrid KGML models in which the overall model is decomposed into modular sub-components that can be process-based models and/or deep learning models. Here, first results in Ladwig et al. (2024) highlighted how MCL can improve the performance of a 1D hydrodynamic model in replicating water temperature dynamics.
​
​
This project aims to discover the key to unlocking ecosystem complexity by integrating AI with ecological knowledge to shed light into the dynamics of crucial aquatic processes: deep-water mixing and algal growth dynamics. Both processes are linked, nonlinear and challenging to predict using AEMs. Machine learning (ML) for scientific discovery will be used on observed data to identify governing equations, which will be further tested for accuracy in state-of-the-art AEMs. To improve projections, we will develop hybrid MCL models incorporating ML into AEMs. Hybridizing models applies to many scientific disciplines, and by developing a trustworthy methodology, this project will guide future applications beyond freshwater ecology. ​​
​
This project is funded through a Villum Young Investigator Grant by Villum Fonden (2025-2030).
Lake water management needs reliable projections by aquatic ecosystem models (AEM); however, these models underperform for non-linear and complex processes, i.e., deep-water mixing as well as algae dynamics. Technological advancements in sensor technology and data collection have significantly increased the volumes of data available for limnologists. At the same time, deep learning algorithms have revolutionized the sciences by their ability to extract information from data of complex systems. This sets up our conundrum: how can we, in a sound way, infer information from environmental data to update our scientific understanding and improve our model projections?