Bayerisches Landesamt für Umwelt

Coupled flow and heat transport models are essential for understanding subsurface processes and for assessing the long-term sustainability of deep geothermal systems. Due to its favourable geothermal conditions, the South German Molasse Basin, characterised by the fractured and karstic carbonate Upper Jurassic Malm aquifer, is Germany’s most productive geothermal region. Extensive development has made numerical reservoir modelling mandatory for permits and critical for assessing project viability and impacts on nearby plants, supporting informed decisions by both operators and authorities. This study presents an analytical review of geothermal reservoir modelling methodologies applied at 22 existing geothermal plants in the South German Molasse Basin. We examine how key geological features, including stratigraphy, faults, karst horizons, and lateral facies variations, are represented with different conceptual and numerical approaches, from explicit structural integration to effective or homogeneous parametrisation. We systematically evaluate the strengths and limitations of prevailing modelling approaches, benchmark them against state-of-the-art methods, and identify key methodological gaps. We further analyse how subsurface data (e.g., pressure, temperature, porosity, permeability, and inflow zones) are incorporated into models and classify parameters by their level of constraint (measured, derived, calibrated, or assumed), enabling consistent cross-comparison. We identify a wide range of modelling approaches, largely due to geological heterogeneity within the reservoir. Calibration practices also vary significantly, with most studies focused on hydraulic calibration with pressure data, while thermal calibration based on temperature measurements remains rare. We also discuss methodological limitations, including the absence of uncertainty analysis of model outcomes and the limited use of operational data for model validation. These factors influence model predictions and have implications for the long-term sustainable management of geothermal resources. By synthesising reservoir modelling practices and contextualising them within state-of-the-art approaches from other sedimentary reservoirs, this review provides a reference framework to support more consistent, transparent, and robust geothermal reservoir modelling and to facilitate knowledge transfer across sedimentary systems.

Faults are essential to reconstruct geologic histories and their presence (or absence) in the subsurface plays a significant role for many societal issues, such as geothermal energy, carbon, hydrogen and hydrocarbon storage or nuclear waste deposition. In the area of the Permo-Mesozoic Franconian Platform (Northern Bavaria, SE Germany) the existence and properties of faults in the subsurface are poorly known due to limited exposure and lack of geophysical data coverage. To address this issue, we have constructed a new fault network based on stratigraphic offsets identified in a network of 18 cross sections. In this contribution, we present our workflow which allows the effective construction of an internally consistent cross-section network covering a study area of >30,000 km 2 , while representing the geology in high detail. Data input is provided by >3,500 (mostly shallow) wells and >250 high-resolution geological maps, complemented by geophysical surveys and field data. We demonstrate how fault segments and their attributes (e.g., fault type, dip direction, offset, certainty of the interpretation, etc.) can be inferred from the profiles, resulting iteratively in consistent profile and fault networks. While we highlight the non-uniqueness of the interpretation and its restriction to vertical offsets, we demonstrate advantages of the fault network, inferred through this pseudo-3D approach, compared to previously available datasets. We discuss the role of newly identified fault populations for Cretaceous inversion tectonics and Cenozoic rifting. Through the incorporation of basic measures of uncertainty and scale-dependency in the resulting fault model, quantification of the fault network can be achieved. This enables the comparison with independent datasets and different model realizations. • We present a new method for the construction of a 3d consistent fault network for study areas based on geological data. • We show how important it is to include scale dependencies in fault networks in order to achieve reliable results. • Through the incorporation of basic measures of uncertainty and scale-dependency in the resulting fault model, quantification of the fault network can be achieved.

In recent years, several regions in Germany experienced devastating floods caused by heavy precipitation events, often associated with severe convective storms. To improve the prediction of such events and corresponding warning strategies, Deutscher Wetterdienst (DWD) is intensifying its collaboration with Germany’s flood forecasting authorities. DWD has significantly enhanced and diversified its forecasting strategies through a range of new model systems. With a focus on seamless and probabilistic prediction in combination with more frequent initializations, DWD’s novel Seamless Integrated Forecasting System (SINFONY) marks a major step forward in predicting severe summertime convective events and heavy precipitation within a lead time of minutes to approximately 12 hours. Additional advancements include the development of 500 m high-resolution NWP with ICON as well as the establishment of an AI center and the integration of AI-based forecasting methods– all together paving the way for next-generation weather forecast systems. Beyond the improvement of forecasting techniques, DWD is placing strong emphasis on collaboration and communication with regional flood forecasting centers, including coordinated outreach to local disaster management authorities. To that end, the joint “Co-Design Project” was launched in 2023 together with regional German Flood Forecasting Centers, aiming to strengthen the hydrometeorological value chain. It is part of DWD’s new binational research initiative “Italia–Deutschland science-4-services network in weather and climate (IDEA-S4S)”. The project focuses on identifying user needs, creating a shared knowledge base for forecast evaluation, developing tailored forecast and warning tools, and supporting decision-making in the face of diverse and uncertain weather predictions. Its four main activities include: a user-oriented evaluation of DWD’s precipitation forecasts the establishment of standardized hydrological verification and thereupon analysis of (new) DWD forecasts within operational flood forecasting models the tailoring of DWD’s new warning system to meet requirements of flood forecasting centers a serious game and E-Learning initiative to improve the communication along the entire warning chain of rainfall and flood forecasts for better decision-making. This contribution provides an overview of the “Co-Design Project”. We will share first results and progress, and look forward to exchanging experiences with other initiatives at the operational intersection of meteorology and hydrology.

Abstract Suction buckets are a promising concept for the foundations of offshore wind turbines. During the installation process of a suction bucket, localized fluidization of the granular soil, so-called piping erosion, may lead to installation failure. A 3D fluid-solid coupled micromechanical simulation is presented to study the occurrence of piping. An Euler-Lagrangian coupling employs momentum exchange between the fluid phase and the geometrically resolved particles. We investigate the behavior of the soil for three cases with varying prescribed suction velocities. We observe piping in the case with the highest suction velocity by analyzing the deformation of the granular fabric and monitoring the differential pressure. The grains under the bucket wall-tip show the highest hydraulic gradients and forces at the onset of piping. This approach permits a detailed analysis of piping phenomena and brings novel insights on the triggering conditions for piping failure of suction-aided foundations.

Abstract Large ocean worlds in the solar system and in other planetary systems likely contain significant fraction of volatile and organic compounds, which condition their habitability potential. Some of these compounds are known to combine with water molecules to form clathrate hydrates under pressure. However, the clathration process in complex compositions and under high pressures representative of these ocean worlds still remain unexplored. Herein we report on the first experimental results on the characterization of the pressure-induced transitional pathway of a mixed clathrate hydrate. We studied crystalline compounds forming in two systems: binary tetrahydrofuran (THF)-H2O that served as a reference system, and ternary THF-methanol (MeOH)-H2O in order to address the role of methanol at high-pressure. We show that high pressure facilitates the incorporation of methanol in clathrate structure, with formation of two mixed clathrate hydrates, THF•xMeOH•17H2O (sII) and THF•0.75MeOH•10H2O. The preferential trapping of methanol at high-pressure offers an explanation for the non-detection of methanol on surfaces of large ocean worlds, while more common on small icy worlds. Moreover, the hydrogenolysis of enclathrated primordial methanol into methane presents a novel hypothesis for the deep-seated endogenic source of methane in Titan and potentially for other ocean worlds.

Abstract. In the recent years, the German federal state of Brandenburg has been particularly impacted by soil moisture droughts. To support the timely and informed management of such water-related risks, we introduce a novel soil moisture and drought monitoring network based on cosmic-ray neutron sensing technology. This initiative is driven by a joint collaboration of research institutions and federal state agencies, and it is the first of its kind in Germany to have started operation. By mid 2024, eight sites were instrumented across Brandenburg; four more are planned for 2025. The data will be openly accessible to foster applications and collaboration right from the start.

This paper addresses the question how catchment scale water and energy balance have responded to climatic and land-surface changes over the last 70 years in the federal state of Saxony, in eastern Germany. Therefore observational data of hydrological and meteorological monitoring sites from 1951-2020 across 71 catchments are examined in a relative water and energy partitioning framework to put the recent drought induced changes in a historical perspective. A comprehensive visualization method is used to analyze the observed time series. The study focuses on changes in decadal time scale and finds the largest decline in observed runoff in the last decade . The observed decline can be explained by the significant increase in aridity, caused by the reduction of annual mean rainfall and an increase in potential evaporation at the same time. In a few mainly forested head water catchments the observed decline in runoff was even stronger than predicted by climate conditions alone. These catchments are still on the recovery from past widespread forest damages in the 1970-80s resulting in a continuous increase of actual evapotranspiration due to forest regrowth. In contrary runoff stayed almost constant in other catchments despite an increase in aridity. These catchments showed declines in actual evapotranspiration which could be signatures of either contributing groundwater at longer time scales or drought induced vegetation damages. These results highlight that water budgets in Saxony are in an unstable, non-stationary regime due to significant climatic changes and regional impacts of land-surface changes such as forest health. The recent decreases in the mean annual runoff are substantial and must be taken into account by the authorities for fresh water management.

ZusammenfassungDie amtliche Statistik zeichnet sich durch ihren gesetzlich auferlegten Fokus auf die Qualität ihrer Veröffentlichungen aus. Dabei folgt sie den europäischen Qualitätsrahmenwerken, die auf nationaler Ebene in Form von Qualitätshandbüchern konkretisiert und operationalisiert werden, sich jedoch bis dato hinsichtlich Ausgestaltung und Interpretation an den Anforderungen der „klassischen“ Statistikproduktion orientieren. Der zunehmende Einsatz maschineller Lernverfahren (ML) in der amtlichen Statistik muss daher zur Erfüllung des Qualitätsanspruchs durch ein spezifisches, darauf zugeschnittenes Qualitätsrahmenwerk begleitet werden. Das vorliegende Papier leistet einen Beitrag zur Erarbeitung eines solchen Qualitätsrahmenwerks für den Einsatz von ML in der amtlichen Statistik, indem es (1) durch den Vergleich mit bestehenden Qualitätsgrundsätzen des Verhaltenskodex für Europäische Statistiken relevante Qualitätsdimensionen für ML identifiziert und (2) diese unter Berücksichtigung der besonderen methodischen Gegebenheiten von ML ausarbeitet. Dabei (2a) ergänzt es bestehende Vorschläge durch den Aspekt der Robustheit, (2b) stellt Bezug zu den Querschnittsthemen Machine Learning Operations (MLOps) und Fairness her und (2c) schlägt vor, wie die Qualitätssicherung der einzelnen Dimensionen in der Praxis der amtlichen Statistik ausgestaltet werden kann. Diese Arbeit liefert die konzeptionelle Grundlage, um Qualitätsindikatoren für ML-Verfahren formell in die Instrumente des Qualitätsmanagements im Statistischen Verbund zu überführen und damit langfristig den hohen Qualitätsstandard amtlicher Statistik auch bei Nutzung neuer Verfahren zu sichern.