Highlight articles

Representation learning with unconditional denoising diffusion models for dynamical systems

We train neural networks as denoising diffusion models for state generation in the Lorenz 1963 system and demonstrate that they learn an internal representation of the system. We make use of this learned representation and the pre-trained model in two downstream tasks: surrogate modelling and ensemble generation. For both tasks, the diffusion model can outperform other more common approaches. Thus, we see a potential of representation learning with diffusion models for dynamical systems.

Mesoscale permeability variations estimated from natural airflows in the decorated Cosquer Cave (southeastern France)

Conservation of decorated caves is highly dependent on airflows and is correlated with rock formation permeability. We present the first conceptual model of flows around the Paleolithic decorated Cosquer coastal cave (southeastern France), quantify air permeability, and show how its variation affects water levels inside the cave. This study highlights that airflows may change in karst unsaturated zones in response to changes in the water cycle and may thus be affected by climate change.

Precursors and pathways: dynamically informed extreme event forecasting demonstrated on the historic Emilia-Romagna 2023 flood

Natural Hazards and Earth System Sciences
DOI 10.5194/nhess-24-2995-2024
6 September 2024

Extreme rainfall is the leading weather-related source of damages in Europe, but it is still difficult to predict on long timescales. A recent example of this was the devastating floods in the Italian region of Emiglia Romagna in May 2023. We present perspectives based on large-scale dynamical information that allows us to better understand and predict such events.

Ice viscosity governs hydraulic fracture that causes rapid drainage of supraglacial lakes

Due to surface melting, meltwater lakes seasonally form on the surface of glaciers. These lakes drive hydrofractures that rapidly transfer water to the base of ice sheets. This paper presents a computational method to capture the complicated hydrofracturing process. Our work reveals that viscous ice rheology has a great influence on the short-term propagation of fractures, enabling fast lake drainage, whereas thermal effects (frictional heating, conduction, and freezing) have little influence.

Optimizing the iodide-adduct chemical ionization mass spectrometry (CIMS) quantitative method for toluene oxidation intermediates: experimental insights into functional-group differences

We introduce detailed and improved quantitation and semi-quantitation methods of iodide-adduct time-of-flight chemical ionization mass spectrometry (I-CIMS) to measure toluene oxidation intermediates. We assess the experimental sensitivity of various functional group species and their binding energy with iodide ions in I-CIMS. A novel classification approach was introduced to significantly enhance the accuracy of semi-quantitative methods (improving R2 values from 0.52 to beyond 0.88).

How is particulate organic carbon transported through the river-fed submarine Congo Canyon to the deep sea?

The land-to-ocean flux of particulate organic carbon (POC) is difficult to measure, inhibiting accurate modeling of the global carbon cycle. Here, we quantify the POC flux between one of the largest rivers on Earth (Congo) and the ocean. POC in the form of vegetation and soil is transported by episodic submarine avalanches in a 1000 km long canyon at up to 5 km water depth. The POC flux induced by avalanches is at least 3 times greater than that induced by the background flow related to tides.

The crucial representation of deep convection for the cyclogenesis of Medicane Ianos

Cyclone Ianos of September 2020 was a high-impact but poorly predicted medicane (Mediterranean hurricane). A community effort of numerical modelling provides robust results to improve prediction. It is found that the representation of local thunderstorms controlled the interaction of Ianos with a jet stream at larger scales and its subsequent evolution. The results help us understand the peculiar dynamics of medicanes and provide guidance for the next generation of weather and climate models.

Merging modelled and reported flood impacts in Europe in a combined flood event catalogue for 1950–2020

Long-term trends in flood losses are regulated by multiple factors, including climate variation, population and economic growth, land-use transitions, reservoir construction, and flood risk reduction measures. Here, we reconstruct the factual circumstances in which almost 15 000 potential riverine, coastal and compound floods in Europe occurred between 1950 and 2020. About 10 % of those events are reported to have caused significant socioeconomic impacts.

Late Quaternary glacial maxima in southern Patagonia: insights from the Lago Argentino glacier lobe

Investigating past glaciated regions is crucial for understanding how ice sheets responded to climate forcings and how they might respond in the future. We use two independent dating techniques to document the timing and extent of the Lago Argentino glacier lobe, a former lobe of the Patagonian Ice Sheet, during the late Quaternary. Our findings highlight feedbacks in the Earth’s system responsible for modulating glacier growth in the Southern Hemisphere prior to the global Last Glacial Maximum.

Cautionary remarks on the planetary boundary visualisation

Our paper examines the visual representation of the planetary boundary concept, which helps convey Earth’s capacity to sustain human life. We identify three issues: exaggerated impact sizes, confusing color patterns, and inaccessibility for colour-vision deficiency. These flaws can lead to overstating risks. We suggest improving these visual elements for more accurate and accessible information for decision-makers.

News & press Highlight articles

Highlight articles

Information for

External links