2025-06-13 | | Total: 8
Accurate retrieval of surface solar radiation (SSR) from satellite imagery critically depends on estimating the background reflectance that a spaceborne sensor would observe under clear-sky conditions. Deviations from this baseline can then be used to detect cloud presence and guide radiative transfer models in inferring atmospheric attenuation. Operational retrieval algorithms typically approximate background reflectance using monthly statistics, assuming surface properties vary slowly relative to atmospheric conditions. However, this approach fails in mountainous regions where intermittent snow cover and changing snow surfaces are frequent. We propose an attention-based emulator for SSR retrieval that implicitly learns to infer clear-sky surface reflectance from raw satellite image sequences. Built on the Temporo-Spatial Vision Transformer, our approach eliminates the need for hand-crafted features such as explicit albedo maps or cloud masks. The emulator is trained on instantaneous SSR estimates from the HelioMont algorithm over Switzerland, a region characterized by complex terrain and dynamic snow cover. Inputs include multi-spectral SEVIRI imagery from the Meteosat Second Generation platform, augmented with static topographic features and solar geometry. The target variable is HelioMont's SSR, computed as the sum of its direct and diffuse horizontal irradiance components, given at a spatial resolution of 1.7 km. We show that, when provided a sufficiently long temporal context, the model matches the performances of albedo-informed models, highlighting the model's ability to internally learn and exploit latent surface reflectance dynamics. Our geospatial analysis shows this effect is most powerful in mountainous regions and improves generalization in both simple and complex topographic settings. Code and datasets are publicly available at https://github.com/frischwood/HeMu-dev.git
It has been suggested that Earth's present water budget formed from oxidation reactions between its initial hydrogen-rich primordial atmosphere and its magma ocean. Here we examine this hypothesis by building a comprehensive atmosphere-magma ocean model. We find that water formation is unlikely for two reasons. First, any water formed from oxidation reactions in the magma ocean would quickly outgass because of the water-poor atmosphere above. Second, the top boundary layer of the magma ocean becomes stable against convection because the oxidation reactions produce metallic iron, which sinks to the core of a growing Earth. This iron loss makes the top boundary layer significantly more buoyant than the rest of the magma, thus becoming stable against mixing. Our results suggest that hydrogen dissolution is unlikely to play a major role in the formation of Earth's oceans.
Starting from the Navier-Stokes equation in the $f$-plane approximation, we provide an exact and explicit solution of the governing equations at leading order for fluid flows in the upper layer of the ocean, driven by a wind stress. Such a solution highlights the presence of a mean Ekman current superimposed to trochoidal oscillations and a background geostrophic current.
Machine learning (ML)-based weather models have rapidly risen to prominence due to their greater accuracy and speed than traditional forecasts based on numerical weather prediction (NWP), recently outperforming traditional ensembles in global probabilistic weather forecasting. This paper presents FGN, a simple, scalable and flexible modeling approach which significantly outperforms the current state-of-the-art models. FGN generates ensembles via learned model-perturbations with an ensemble of appropriately constrained models. It is trained directly to minimize the continuous rank probability score (CRPS) of per-location forecasts. It produces state-of-the-art ensemble forecasts as measured by a range of deterministic and probabilistic metrics, makes skillful ensemble tropical cyclone track predictions, and captures joint spatial structure despite being trained only on marginals.
We used precomputed steady states obtained by a spin-up for a global marine ecosystem model as training data to build a mapping from the small number of biogeochemical model parameters onto the three-dimensional converged steady annual cycle. The mapping was performed by a conditional variational autoencoder (CVAE) with mass correction. Applied for test data, we show that the prediction obtained by the CVAE already gives a reasonable good approximation of the steady states obtained by a regular spin-up. However, the predictions do not reach the same level of annual periodicity as those obtained in the original spin-up data. Thus, we took the predictions as initial values for a spin-up. We could show that the number of necessary iterations, corresponding to model years, to reach a prescribed stopping criterion in the spin-up could be significantly reduced compared to the use of the originally uniform, constant initial value. The amount of reduction depends on the applied stopping criterion, measuring the periodicity of the solution. The savings in needed iterations and, thus, computing time for the spin-up ranges from 50 to 95\%, depending on the stopping criterion for the spin-up. We compared these results with the use of the mean of the training data as an initial value. We found that this also accelerates the spin-up, but only by a much lower factor.
Understanding the plausible upper bounds of extreme weather events is essential for risk assessment in a warming climate. Existing methods, based on large ensembles of physics-based models, are often computationally expensive or lack the fidelity needed to simulate rare, high-impact extremes. Here, we present a novel framework that leverages a differentiable hybrid climate model, NeuralGCM, to optimize initial conditions and generate physically consistent worst-case heatwave trajectories. Applied to the 2021 Pacific Northwest heatwave, our method produces temperature anomalies up to 3.7 $^\circ$C above the most extreme member of a 75-member ensemble. These trajectories feature intensified atmospheric blocking and amplified Rossby wave patterns--hallmarks of severe heat events. Our results demonstrate that differentiable climate models can efficiently explore the upper tails of event likelihoods, providing a powerful new approach for constructing targeted storylines of extreme weather under climate change.
This chapter introduces flood and drought through the understanding of the water cycle. In addition to the water cycle, we consider the energy cycle. The floods and droughts have strong regional and seasonal characteristics. The causes of the unbalanced water conditions can occur under the various meteorological phenomena, which have strong regional and seasonal varieties. For an understanding of the cause of floods and drought, we first consider the geographical characteristics of the floods and droughts. At the same time, we focus on the spatial and temporal time-scale of the flood or drought. Moreover, because floods and drought can be considered as excess and shortage of water, respectively, they are opposite. However, their spatial and temporal scales are asymmetric.
We assess the impact of a multi-scale loss formulation for training probabilistic machine-learned weather forecasting models. The multi-scale loss is tested in AIFS-CRPS, a machine-learned weather forecasting model developed at the European Centre for Medium-Range Weather Forecasts (ECMWF). AIFS-CRPS is trained by directly optimising the almost fair continuous ranked probability score (afCRPS). The multi-scale loss better constrains small scale variability without negatively impacting forecast skill. This opens up promising directions for future work in scale-aware model training.