Events

A suite of field experiments illuminate the intimate involvement of moisture in the progressive physical disintegration (cracking) process of surface rocks, even in extremely dry deserts, and in the formation of asymmetrical hyperarid landscapes on Earth and Mars.

Long-term high-resolution orbital imaging at Mars has led to extraordinary advances in understanding martian ice and its connection to climate.  Icy seasonal phenomena such as flows in gullies, avalanches, and exotic defrosting patterns characterize the present climate. Interannual variability over a martian decade helps us deduce climatic averages and current trends. Observations of polar ice layers have characterized periodicities related to orbital change over longer timescales up to millions of years.

Here, I’ll describe the HiRISE camera and its continued mission to describe a dynamic Mars over 20 years of observations, with a special focus on north polar avalanches.  HiRISE has uniquely high resolution and benefits from high signal-to-noise (even at the poles); a near-polar orbit that allows imaging of almost any location within two weeks; color bands that are sensitive to ice; and sufficient imaging stability to construct high-quality meter-scale DTMs. The scientific impact of HiRISE owes much to rapid data releases and community targeting via our online tool HiWISH, ensuring acquisition and analysis of data relevant to today’s scientific questions. 

Atmospheric mineral dust is a well-established source of nutrients to marine ecosystems, yet its contribution to terrestrial plant nutrition has long been underestimated, largely due to the assumption that nutrient acquisition occurs predominantly through root uptake from soils. Here, we present evidence from controlled greenhouse experiments under ambient and elevated CO₂, laboratory simulations of leaf microenvironments, isotopic and geochemical tracing, and field fertilization experiments conducted in both a Mediterranean ecosystem and a tropical forest in Puerto Rico, demonstrating that plants can directly acquire nutrients through their leaf surfaces following atmospheric dust deposition. Using rare earth elements and Nd isotopes, we distinguish nutrients derived from soils from those delivered by deposited atmospheric particles. Laboratory simulations show that mildly acidic leaf surfaces, together with organic acids secreted by leaves, enhance mineral dissolution and facilitate foliar uptake of dust-borne nutrients. In a pioneering Mediterranean field experiment explicitly designed to isolate foliar uptake, we quantified the bioavailable fraction of key nutrients supplied by dust, including P, Fe, Mn, and Cu, and observed clear enrichment of multiple micronutrients in leaf tissues following dust application. These field-based measurements enabled the construction of a global geospatial framework integrating dust deposition with soil nutrient fluxes, indicating that dust-derived inputs can constitute a meaningful fraction of total nutrient supply across large regions, and that during dust events, short-term foliar inputs can rival or exceed soil-derived fluxes. Complementary field observations in a tropical forest in Puerto Rico further reveal foliar nutrient responses consistent with direct dust uptake. Building on these results, we outline a pathway for incorporating foliar dust uptake into Earth system representations of terrestrial nutrient cycling by explicitly accounting for atmospheric nutrient inputs at the canopy level and their interaction with soil-derived fluxes. Together, these findings identify foliar dust uptake as an overlooked but consequential nutrient acquisition pathway and highlight its relevance in highly weathered, nutrient-limited tropical forests, where atmospheric inputs may play a critical role in regulating nutrient availability and carbon–nutrient interactions.

Earth system data is rapidly increasing in volume as new observation systems generate data at rates of terabytes/day and modelling systems continue to

increase their resolution and the number of ensemble members. Coping with such amounts of data presents

substantial challenges to Earth system researchers who often find it difficult to identify suitable tools and concepts to efficiently process such data to the extent that is necessary to obtain statistically meaningful results. Conversely, computer scientists are generally more familiar with the technical aspects of data handling, but they have difficulties to understand the domain-specific aspects of Earth system data with respect to Earth’s geometry or important data and

metadata properties that cannot be neglected. This seminar builds on a one semester university course (slides and videos are available at https://

b2drop.eudat.eu/s/iwYob4QonXHjEPH ) and will discuss selected aspects in an interactive fashion,

including best practices for handling massive amounts

of (Earth system) data, the role of metadata and quality control and more.

Recent years have seen a revolution in weather

prediction. Since 2023, data-driven weather

models are outperforming even the best

numerical models that have been developed

over several decades around the world. AI

weather predictions are a lot faster and cheaper

and they often detect future weather patterns

earlier than their numerical counterparts. The

models also appear robust and can be used to

extrapolate to unseen climate conditions, at

least within a limited range. However, several

challenges remain, in particular when it comes

to long-term climate projections and multiscale

feedbacks within the Earth system. The

talk will review the recent achievements of AI

in weather and climate modelling and discuss

the current limitations.

17Oexcess is the deviation of d17O from the generally accepted 17O-18O mass dependent reference line. In rainfall, 17Oexcess depends mainly on relative humidity at the moisture source region, with lower relative humidity corresponding to higher 17Oexcess. In some cases, however, rainfall 17Oexcess is influenced by atmospheric processes like partial re-evaporation of the raindrops or moisture recycling. We examine how does 17Oexcess in CaCO3 record 17Oexcess of its parent water and apply it to paleo hydrology in Soreq Cave (Israel) and in Devils Hole (Nevada, USA).

In Soreq Cave, 17Oexcess of 50 per meg was obtained in the weighted mean modern rainfall, consistent with the low relative humidity at the moisture source region of the Eastern Mediterranean Sea. 17Oexcess of paleo water were reconstructed from Soreq Cave speleothems, at an age range of 0 - 160 ka. In most of the record values are similar to that in modern cave water, but a few events suggest higher relative humidity, consistent with a more marine storm trajectory. The values at the Last Glacial Maximum suggest low relative humidity and likely indicate the penetration of very cold air.

In Devils Hole, 17Oexcess in modern and interglacial reconstructed water is higher than expected by relative humidity, suggesting significant moisture recycling in this continental site. In glacial periods, however, 17Oexcess suggest much less evaporation of water from land surfaces.

After an earthquake occurs, slip models of the event may be estimated from geodetic observations. This process generally requires static coseismic Green's functions, which must be calculated via a forward model which includes an approximation of the material properties, topography, and fault geometry in the region of interest. Until recently, the lack of seafloor geodetic instrumentation and the use of unrealistically simple forward models have resulted in poor resolution of near-trench slip in subduction zone settings.  In this talk, I will present an investigation into the effects of 3D structure, particularly topography, on forward models of earthquake deformation and on earthquake static slip estimates. I will show that models which neglect 3D structure yield inaccurate estimates of near-trench slip, particularly when seafloor geodetic data are utilized in the inversion.

Climate change is a global phenomenon, yet its fingerprints vary

substantially across regions. This talk highlights a range of these

regional patterns using observational records and climate model

simulations, analyzed with machine learning and complementary

statistical tools.

The first part of the talk examines the magnitude of climate

change across temporal and spatial scales, showing how longterm

warming reshapes seasonal and diurnal temperature cycles

in different regions.

The second part examines how quickly climate mitigation signals

can be detected against regional climate variability, highlighting

where the effects of emission reductions are likely to emerge

sooner or later across the globe.

The final part of the talk addresses the question of climate

change acceleration. Despite rapidly increasing greenhouse gas

emissions, recent studies suggest that the global mean warming

rate remains linear. We revisit this issue by shifting the focus

from global averages to regional scales, where we detect

significant acceleration in warming across a substantial fraction

of the world.

The Dead Sea region has seen a rapid increase in sinkhole formation, posing serious environmental and infrastructure risks. The Geological Survey of Israel monitors sinkhole-related land subsidence along the western shore using InSAR, but current detection relies on manual interpretation of interferometric phase data, which is time-consuming and error-prone.

In this talk, I present an AI-based Deep Learning framework for automated detection of sinkhole-related subsidence from InSAR data. The model learns interferometric phase deformation patterns, rather than visual features, and is trained using expert-labeled subsidence maps from years of operational monitoring. I demonstrate the model’s ability to generalize across spatial and temporal settings using multiple evaluation schemes and object-level performance metrics. Results show effective detection of subsidence areas, promising generalization to unseen regions, and the ability to reconstruct large-scale subsidence trends from patch-level predictions.

Many fossil materials have embedded signals that enable aspects of the past to be reconstructed. These signals however can be altered or lost due to processes that take place once the fossil material is buried (diagenesis). Thus extracting reliable signals can be a major challenge. Here I present a new approach to better understanding diagenesis that I apply to bone.

Satellite instruments, such as TROPOMI, are routinely

used to quantify tropospheric nitrogen dioxide (NO2)

based on its narrowband light absorption in the UV/

visible spectral range. The key limitation of such

retrievals is that they can only return the „vertical

column density“ (VCD), defined as the integral of the

NO2 concentration profile. The profile itself, which

describes the vertical distribution of NO2, remains

unknown.

This presentation showcases „NitroNet“, the first NO2

profile retrieval for TROPOMI. NitroNet is a neural

network, which was trained on synthetic NO2 profiles

from the regional chemistry and transport model WRFChem,

operated on a European domain for the month of

May 2019. The neural network receives NO2 VCDs from

TROPOMI alongside ancillary variables (meteorology,

emission data, etc.) as input, from which it estimates NO2

concentration profiles.

The talk covers:

• an introduction to satellite remote sensing of NO2.

• the theoretical underpinnings of NitroNet, how the

model was trained, and how it was validated.

• practical new applications that NitroNet enables.

The soils of the Golan Heights plateau, in northern Israel, are underlaid by basaltic rocks ranging in age from ~5.5 to 0.1 Ma. Volcanism in this region is associated with the development of the Red Sea rift, and in accordance with the propagation of the rift, the age of the volcanic units displays a general northward decrease. Topographic position, field evidence and morphology, indicate that nearly all of the soils were formed in situ by weathering of the basaltic bedrock and it has been generally assumed the soils form a chronosequence. While the soils are predominantly of basaltic origin, the contribution of allochthonous aeolian sediments to the soils have long been recognized, mainly through the presence of quartz grains, typical to the regional dust.

Based on geochemical mass balance calculations, we found that not only are the soil ages decoupled from the ages of the underlying basalts, they represent up to a few thousand years of soil production, at most (Zaarur et al., 2024). This time frame is orders of magnitude shorter than the basalt age, challenging the prevalent assumption that the soils form a chronosequence. In addition, new OSL measurements provide independent soil ages, based on dust burial in the soils.

OSL measurements were conducted on soils collected from sites from the southern and central Golan, and include samples of mature Vertisols covering the oldest Pliocene basalts on the southern plateau, and soils collected from deep crevasses in the basalts. The ages of all measured samples range between ~0.4 and ~7 ka. Older particles are restricted to deep and protected microenvironments. These results strengthen the chemical modeling findings, that suggested that the soils represent up to a few thousand years of soils accumulation. Furthermore, these results point to a massive soil-loss event that pre-dates the accumulation and development of these soils.

In addition, our findings strongly suggest that erosion is a significant factor controlling soil formation and accumulation on the plateau, despite the generally flat morphology of the Golan Heights. The erosion is likely associated with tectonic activity along the Dead Sea transform, with the development of the Kinarot and Hula valleys, and with the consequential development of drainage systems of various sizes on the plateau.

Rivers play a central role in shaping the Earth's surface and ecosystems through physical, chemical, and biological interactions. The intensity, time, and location of these interactions change as rivers continuously migrate across the landscape. In recent decades, human activity and climate change have altered river hydrology and sediment fluxes, leading to changes in river positions. Climate warming, increasing flood extremes, and human-induced land use changes have slowed river migration rates in some river reaches while accelerating them in others. However, a comprehensive, spatially continuous, large-scale perspective on and understanding of these recent changes in the rate of river position shifts is lacking.

To address this knowledge gap, we created a continuous global dataset of yearly river positions and migration rates over the past four decades. The continuous annual river positions were detected using Landsat-derived surface-water datasets and processed in Google Earth Engine, a cloud-based parallel-computation platform. The resulting river extents and centerlines reflect their yearly permanent positions, corresponding to the river locations during base flow. This approach improves the representation of position changes derived from geomorphological rather than hydrological processes. To analyze river position changes across different patterns and complexities at large scales, we developed and applied a global reach-based quantification method for river mobility rates.

Results show that while some alluvial rivers maintain a stable annual pace of mobility, others exhibit trends in migration rates. For instance, the Amazon Basin, which has experienced significant deforestation and hydrological modifications, has shown increased rates of river position change in recent decades, impacting floodplain forests and communities. In this talk, we will discuss the advantages, limitations, and applications of the detected yearly river positions and mobility rates, offer insights into the forcings driving changes in river positions and their environmental outcomes, and highlight current and future impacts on one of Earth’s most vulnerable hydrologic systems.

A prominent mode of large earthquakes is self-healing slip pulses, which feature a finite slipping length. The latter implies that the rise time of the displacement waveforms of such earthquakes is significantly shorter than the source duration, in contrast to expanding crack-like earthquakes. The slipping length emerges from an interplay between leading-edge contact breakage and trailing-edge re-strengthening (healing), which is intrinsically related to the generic frictional rate and state dependence of faults. Our understanding of slip pulse earthquakes lags behind that of their crack-like counterparts. We show that steady-state slip pulses are intrinsically unstable, yet that the spatiotemporal dynamics of unsteady slip pulses are surprisingly and fundamentally related to the corresponding steady-state family of solutions, leading to a reduced-dimensionality description. We further show that the development of instability of growing pulses is slow, explaining their emergence in natural and manmade frictional systems. The theory culminates in an equation of motion for unsteady slip pulses, and is discussed in relation to large-scale numerical simulations, laboratory earthquakes and geophysical observations.

While AI has been disrupting conventional weather

forecasting, we are only beginning to witness the

impact of AI on long-term climate simulations. The

fidelity and reliability of climate models have been

limited by computing capabilities. These limitations

lead to inaccurate representations of key processes

such as convection, cloud, or mixing or restrict the

ensemble size of climate predictions. Therefore, these

issues are a significant hurdle in enhancing climate

simulations and their predictions.

Here, I will discuss a new generation of climate

models with AI representations of unresolved ocean

physics, learned from high-fidelity simulations, and

their impact on reducing biases in climate

simulations. The simulations are performed with

operational ocean model components. I will further

demonstrate the potential of AI to accelerate climate

predictions and increase their reliability through the

generation of fully AI-driven emulators, which can

reproduce decades of climate model output in seconds

with high accuracy

Satellite instruments, such as TROPOMI, are routinely

used to quantify tropospheric nitrogen dioxide (NO2)

based on its narrowband light absorption in the UV/

visible spectral range. The key limitation of such

retrievals is that they can only return the „vertical

column density“ (VCD), defined as the integral of the

NO2 concentration profile. The profile itself, which

describes the vertical distribution of NO2, remains

unknown.

This presentation showcases „NitroNet“, the first NO2

profile retrieval for TROPOMI. NitroNet is a neural

network, which was trained on synthetic NO2 profiles

from the regional chemistry and transport model WRFChem,

operated on a European domain for the month of

May 2019. The neural network receives NO2 VCDs from

TROPOMI alongside ancillary variables (meteorology,

emission data, etc.) as input, from which it estimates NO2

concentration profiles.

The talk covers:

• an introduction to satellite remote sensing of NO2.

• the theoretical underpinnings of NitroNet, how the

model was trained, and how it was validated.

• practical new applications that NitroNet enables.

Forecasting extreme weather relies on the intrinsic predictability of the atmospheric flow, the model resolution needed to represent key processes, and the quality of the initial conditions used to initiate forecasts. In the seminar, I shall present a unified multiscale perspective showing how recent work from my group links these elements into a coherent framework for understanding predictability in the Mediterranean region.

We shall begin at the large scale, where dynamical-systems diagnostics show that Atlantic–European weather regimes are dynamically grounded states with characteristic stability and persistence. These regimes shape the background flow in which Mediterranean extremes develop, thereby defining the intrinsic limits and opportunities for extended-range predictability.‎1 This large-scale structure naturally informs how specific high-impact systems evolve.

At the synoptic scale, a newly developed Lagrangian framework allows us to analyze Mediterranean cyclones within their full potential-vorticity (PV) architecture. The same dynamical features that govern regime persistence help explain why some cyclones maintain long predictability horizons while others amplify uncertainty rapidly, depending on their depth, PV structure, and regional context.‎2 This insight flows directly into our analysis of compound “wet” and “windy’’ extremes, which preferentially arise during particularly persistent atmospheric configurations, effectively the multivariate expression of the dynamical behaviour captured at both the regime and cyclone scales.‎3

At the mesoscale, realizing this dynamical predictability in practice requires sufficient model resolution. High-resolution simulations are essential for capturing sea-breeze interactions, mountain–valley circulations, and other thermally driven flows that modulate extremes in the Eastern Mediterranean, features that coarse reanalysis systematically underestimate.‎4

Finally, we turn to improving the accuracy of initial conditions. Here, Syncope, a high-frequency acoustic sensing system, resolves boundary-layer gusts and turbulence structures that are overlooked by operational networks, providing more realistic near-surface information for initializing numerical weather prediction model simulations.‎5

Together, these studies form a consistent multiscale narrative showing how advances in dynamical understanding, high-resolution modelling, and improved boundary-layer observations can jointly advance the predictability of extreme weather.

Earth's landscapes are shaped by competition between tectonic plates that push bedrock upward and river networks that remove mass. Transport of countless rock fragments is a fundamental aspect of this action, resonating with many other near-surface processes across the hydrosphere, biosphere, and geosphere. Identifying how efficiently rock fragments are transported away, considering their properties and ecohydrological feedbacks during weather events, has remained a persistent scientific challenge since the dawn of computational geomorphology. With recent advances in terrain remote sensing and analysis techniques, hydroclimate observations/models, and computational methods for describing dynamic topography, a research frontier is emerging, paving the way for a promising new era in the science of surface processes and topographic forms.

The seminar first presents a new application of a theory for heterogeneous sediment transport in mountainous gravel-bed rivers. A set of numerical experiments discovered process-form relations that emerge from sediment grains' lithological heterogeneity. Then, the talk will present a first-of-its-kind Earthcasting approach that integrates high-resolution event-scale rainfall forcing into a Holocene-scale landscape evolution research framework. Within that timescale, the importance of the interaction between soil grains and ecohydrological processes in shaping fast-evolving landforms is highlighted. Lastly, paleo-rainfall regimes capable of triggering erosion-deposition cycles and possible future transitions to a unique climate-erosion state by the 21st century will be demonstrated.

The findings have the potential to shift paradigms in the interpretation of sediment records and landscape forms. The newly developed methodologies enable unprecedented quantification of surface processes with respect to material properties and climate forcings, which open opportunities toward a transformational understanding of landscape evolution.

Despite covering over a third of Earth’s land surface, arid regions remain among the least understood hydrological environments. Practically every component of the desert water cycle is more poorly constrained than its counterpart in wetter regions. Yet deserts are home to over 20% of the global population and are disproportionately vulnerable to hydrometeorological hazards such as droughts, floods, and the accelerating impacts of climate change. A better understanding of the desert water cycle is therefore not only a scientific challenge, but a critical need for sustainable water resource and risk management in drylands.

In this talk, I will present three studies that illuminate different aspects of the desert water cycle:

(a)  how satellite observations can be used to infer the (underwater) topography — and thus the water volume — of remote desert lakes;

(b) what atmospheric ingredients link moisture, rain, and floods in the hyperarid Sahara, and how these relate to the desert's paleo- (and future?) climate; and

(c)  how misjudged flood risk management on the desert margin contributed to the deadliest hydrometeorological disaster of the 21st century in Derna, Libya.

Together, these studies illustrate how unconventional combinations of satellite data and modelling can overcome the challenges of limited in situ observations to reconstruct, quantify, and ultimately understand hydrological processes in deserts. They also challenge longstanding assumptions about runoff generation and risk mitigation in arid regions, pushing the boundaries of what we thought we could know in some of the world's most water-scarce landscapes.

Advances in spectral and structural remote sensing are transforming how

we study and monitor plant ecophysiology across scales, from individual

trees to entire agricultural regions. This lecture will explore how

hyperspectral imaging, LiDAR-based 3D canopy modeling, and artificial

intelligence can be integrated to quantify plant functional traits, monitor

crop dynamics, and support precision agriculture. Through three case

studies, we will demonstrate the power of these approaches in capturing

structural and physiological complexity: (1) Satellite-based detection of

bloom shifts and phenological patterns in California’s almond orchards,

revealing climate-driven variations in flowering dynamics; (2) Fusion of

thermal, multispectral, and LiDAR data to estimate plant water status and

its relationship to fruit cracking, linking spectral signals with physiological

stress responses; and (3) Crop-type mapping and multi-year monitoring

of Israeli agricultural systems using Sentinel-1 and Sentinel-2 data

combined with machine learning for national-scale agricultural

assessment. Together, these studies illustrate how spectral

ecophysiology, combining remote sensing and artificial intelligent, offers

new opportunities to bridge plant function, management, and

sustainability in agricultural landscapes under changing environmental

conditions.