Highlight articles

Moving beyond post hoc explainable artificial intelligence: a perspective paper on lessons learned from dynamical climate modeling

We draw from traditional climate modeling practices to make recommendations for machine-learning (ML)-driven climate science. Our intended audience is climate modelers who are relatively new to ML. We show how component-level understanding – obtained when scientists can link model behavior to parts within the overall model – should guide the development and evaluation of ML models. Better understanding yields a stronger basis for trust in the models. We highlight several examples to demonstrate.

Causes of the exceptionally high number of fatalities in the Ahr valley, Germany, during the 2021 flood

Natural Hazards and Earth System Sciences
DOI 10.5194/nhess-25-581-2025
12 February 2025

In July 2021, flooding killed 190 people in Germany, 134 of them in the Ahr valley, making it the deadliest flood in recent German history. The flash flood was extreme in terms of water levels, flow velocities and flood extent, and early warning and evacuation were inadequate. Many died on the ground floor or in the street, with older and impaired individuals especially vulnerable. Clear warnings should urge people to seek safety rather than save belongings, and timely evacuations are essential.

Turbulent heat flux dynamics along the Dotson and Getz ice-shelf fronts (Amundsen Sea, Antarctica)

Few observations exist in the Amundsen Sea. Consequently, studies rely on reanalysis (e.g., ERA5) to investigate how the atmosphere affects ocean variability (e.g., sea-ice formation and melt). We use data collected along ice shelves to show that cold, dry air blowing from Antarctica triggers large ocean heat loss, which is underestimated by ERA5. We then use an ocean model to show that this bias has an important impact on the ocean, with implications for sea-ice forecasts.

Internal tides off the Amazon shelf in the western tropical Atlantic: analysis of SWOT Cal/Val mission data

This study focuses on the internal tides (ITs) off the Amazon shelf in the tropical Atlantic. It is based on 2 km horizontally gridded observations along the swaths of SWOT (Surface Water and Ocean Topography) track 20 during the calibration/validation phase (Cal/Val, 1 d orbit) from late March to early July 2023. We evaluate the amplitude of M2, N2, and S2 frequencies and use the M2 atlas as an internal tide correction model for SWOT observations. Internal tide amplitudes (models or atlases) are first derived by harmonic analysis of the SWOT sea level anomaly (SLA). The estimation is improved by performing a principal component analysis before the harmonic analysis. The results compare very well with the high-resolution empirical tide (HRET) internal tide model, the reference product for internal tide corrections in altimetry observations. The coherent mode 1 and mode 2 M2 can be distinguished in the internal tide model derived from SWOT, while the higher modes with their strong SLA signature are seen mostly in the incoherent part. In comparison to HRET, the correction of SWOT observations with SWOT-based atlases may be more relevant for this track.

The alongshore tilt of mean dynamic topography and its implications for model validation and ocean monitoring

Mean dynamic topography (MDT) describes variations in the mean sea surface height above a reference surface called a geoid. We show that MDT predicted by a regional ocean model, including a significant tilt of several centimeters along the coast of Nova Scotia, is in good agreement with estimates based on sea level observations. We demonstrate that this alongshore tilt of MDT can provide a direct estimate of the average alongshore current and also of the area-integrated nearshore circulation.

Assessment framework to predict sensitivity of marine calcifiers to ocean alkalinity enhancement – identification of biological thresholds and importance of precautionary principle

The environmental impacts of ocean alkalinity enhancement (OAE) are unknown. Our synthesis, based on 68 collected studies with 84 unique species, shows that 35 % of species respond positively, 26 % respond negatively, and 39 % show a neutral response to alkalinity addition. Biological thresholds were found from 50 to 500 µmol kg−1 NaOH addition. A precautionary approach is warranted to avoid potential risks, while current regulatory framework needs improvements to assure safe biological limits.

Blending 2D topography images from the Surface Water and Ocean Topography (SWOT) mission into the altimeter constellation with the Level-3 multi-mission Data Unification and Altimeter Combination System (DUACS)

The Surface Water and Ocean Topography (SWOT) mission delivers unprecedented swath-altimetry products. In this paper, we describe how we extended the Level-3 algorithms to handle SWOT’s unique swath-altimeter data. We also illustrate and discuss the benefits, relevance, and limitations of Level-3 swath-altimeter products for various research domains.

Modelling current and future forest fire susceptibility in north-eastern Germany

Natural Hazards and Earth System Sciences
DOI 10.5194/nhess-25-383-2025
30 January 2025

In this study we applied a random forest machine learning algorithm to model current and future forest fire susceptibility (FFS) in north-eastern Germany using anthropogenic, climatic, topographic, soil, and vegetation variables. Model accuracy ranged between 69 % and 71 %, showing moderately high model reliability for predicting FFS. The model results underline the importance of anthropogenic and vegetation parameters. This study will support regional forest fire prevention and management.

Present-day mass loss rates are a precursor for West Antarctic Ice Sheet collapse

In this study, we present an improved way of representing ice thickness change rates in an ice sheet model. We apply this method using two ice sheet models of the Antarctic Ice Sheet. We found that the two largest outlet glaciers on the Antarctic Ice Sheet, Thwaites Glacier and Pine Island Glacier, will collapse without further warming on a timescale of centuries. This would cause a sea level rise of about 1.2 m globally.

Opinion: Understanding the impacts of agriculture and food systems on atmospheric chemistry is instrumental to achieving multiple Sustainable Development Goals

We discuss our current understanding of and knowledge gaps in how agriculture and food systems affect air quality and how agricultural emissions can be mitigated. We argue that scientists need to address these gaps, especially as the importance of fossil fuel emissions is fading. This will help guide food-system transformation in economically viable, socially inclusive, and environmentally responsible ways and is essential to help society achieve sustainable development.

Publications Highlight articles

Highlight articles

Information for

External links