arXiv subjects

Jan Keller

Publications and source records attributed to Jan Keller.

4 recordsLinked to original sources

arXiv2026

AICON: An operational global machine learning weather forecasting model

We introduce AICON, a global machine learning weather prediction (MLWP) model which generates forecasts at 13 km spatial resolution with a 3-hour time step, trained on the high-resolution, non-hydrostatic ICON-DREAM dataset. AICON is in full operational use at Deutscher Wetterdienst since 2nd of March 2026. The model employs a graph neural network (GNN) architecture with an encoder-processor-decoder structure, where node updates are performed using a graph attention mechanism. A key feature of AICON is its use of an icosahedral multi-mesh derived from the native grid of the ICON model, ensuring consistency with the training data. ICON's terrain-following vertical SLEVE coordinate is one of the major distinctions from existing emulators. AICON's training strategy prioritizes small-scale fidelity by avoiding autoregressive multi-step rollout and longer forecast horizons during training, a design choice motivated by the hypothesis that this approach preserves fine-scale features often damped in models optimized for longer-range forecasts. We describe the prognostic and diagnostic variables used for training, the transfer learning protocol employed to accelerate convergence, and the model's performance across a range of evaluation metrics. An extensive evaluation, including routine verification against observation, a tropical cyclone case and spectral analysis reveal the strengths and limitations in the representation of atmospheric variability across scales. Routine verification against observations demonstrates competitive skill relative to the operational ICON model, particularly for near-surface variables in the short to medium forecast range.

physics.ao-ph↗

arXiv2026

Near 13% efficient semitransparent Cu(In,Ga)S2 solar cells with band gap of 1.6 eV on transparent back contact

Wide-gap Cu(In,Ga)S2 solar cells with In2O3:Sn (ITO) as transparent back contact are evaluated for the application as top cells in tandem devices. The effect of Na on the solar cell performance is investigated by supplying additional Na by NaF co-evaporation or exclusively by Na diffusion from glass. An efficiency of 12.7% is achieved for a semitransparent solar cell with a band gap of 1.6 eV, with sufficient Na diffusion from glass only, allowed by a thin ITO layer. Absorber grown with additional NaF co-evaporation during Cu(In,Ga)S2 growth on thicker ITO show a comparable efficiency of 12%. High temperature growth at Tsub = 630{\deg}C enhances overall absorber quality and results in wide-gap absorbers, with photoluminescence quantum yield improved to 1.5 x 10-5, two orders of magnitude higher than absorber grown at low temperature. NaF co-evaporation is effective in suppressing deep defects, thereby reducing non-radiative recombination and enhancing photoluminescence quantum yield further. A GaOx interfacial layer is formed at the rear contact, likely contributing to the passivation of the back contact. With the presence of thick GaOx layer, current blocking effects are visible in the current-voltage curves. On the contrary, a thinner ITO tends to result in thinner GaOx layer and no current blocking is observed.

cond-mat.mtrl-sci↗

arXiv2023

Explaining heatwaves with machine learning

Heatwaves are known to arise from the interplay between large-scale climate variability, synoptic weather patterns and regional to local scale surface processes. While recent research has made important progress for each individual contributing factor, ways to properly incorporate multiple or all of them in a unified analysis are still lacking. In this study, we consider a wide range of possible predictor variables from the ERA5 reanalysis, and ask, how much information on heatwave occurrence in Europe can be learned from each of them. To simplify the problem, we first adapt the recently developed logistic principal component analysis to the task of compressing large binary heatwave fields to a small number of interpretable principal components. The relationships between heatwaves and various climate variables can then be learned by a neural network. Starting from the simple notion that the importance of a variable is given by its impact on the performance of our statistical model, we arrive naturally at the definition of Shapley values. Classic results of game theory show that this is the only fair way of distributing the overall success of a model among its inputs. With this approach, we find a non-linear model that explains 70% of reduced heatwave variability, 27% of which are due to upper level geopotential while top level soil moisture contributes 15% of the overall score. In addition, Shapley interaction values enable us to quantify overlapping information and positive synergies between all pairs of predictors.

physics.ao-ph↗

arXiv2014

A lipid bound actin meshwork organizes liquid phase separation in model membranes

The eukaryotic cell membrane is connected to a dense actin rich cortex. We present FCS and STED experiments showing that dense membrane bound actin networks have severe influence on lipid phase separation. A minimal actin cortex was bound to a supported lipid bilayer via biotinylated lipid streptavidin complexes (pinning sites). In general, actin binding to ternary membranes prevented macroscopic liquid-ordered and liquid-disordered domain formation, even at low temperature. Instead, depending on the type of pinning lipid, an actin correlated multi-domain pattern was observed. FCS measurements revealed hindered diffusion of lipids in the presence of an actin network. To explain our experimental findings, a new simulation model is proposed, in which the membrane composition, the membrane curvature, and the actin pinning sites are all coupled. Our results reveal a mechanism how cells may prevent macroscopic demixing of their membrane components, while at the same time regulate the local membrane composition.

cond-mat.soft↗