Michael Jähn

High-resolution river discharge provides relevant information for evaluating regional climate models and assessing the hydrological impact of climate change. In this study, we generate high-resolution European river discharges by forcing the hydrological discharge (HD) model with runoff fields from the Icosahedral non-hydrostatic climate limited-area mode (ICON-CLM) regional climate model. We then evaluate the quality of the simulated discharge obtained from a reanalysis-driven hindcast by comparing it with observations. Moreover, we assess how well the historical discharge trends are represented across six simulations in which ICON-CLM was driven by different global climate models (GCMs). A comparison of the results with observed river discharge across Europe shows that ICON-CLM‐based simulations exhibit widespread dry biases, which are partly attributable to underestimation of precipitation in these regions, and partly to a too weak surface runoff component in the land-surface scheme of ICON-CLM. Using the HydroPy hydrology model as an alternative representation of land-surface hydrology improves the simulated discharge in many regions. However, biases remain in areas with regulated rivers, as the HD model does not take regulations into account. Analysis of discharge trends from 1960–2014 reveals robust, large-scale patterns in the reference hindcasts, such as an increase in mean and low-flow discharge in Northern Europe and a decrease in Southern Europe. While some historical simulations reproduce these patterns, others show opposing or inconsistent signals, highlighting the strong influence of the driving GCMs. Our results underscore the suitability of ICON-CLM atmospheric forcing for hydrological impact studies and suggest the need for improved model tuning with regard to precipitation, including undercatch corrections in reference observations. They also emphasize the importance of multi-model ensembles to reliably capture climate-induced hydrological changes.

Atmospheric composition models require efficient computation of surface fluxes from anthropogenic emissions and terrestrial biosphere exchange. This is particularly true for applications such as ensemble-based inverse emission modeling where these calculations constitute significant performance bottlenecks. We present the directive-based GPU acceleration and comprehensive refactoring of two critical surface-flux modules within the ICON-ART atmospheric modeling framework: the Online Emission Module (OEM) for anthropogenic emissions and the Vegetation Photosynthesis and Respiration Model (VPRM) for biospheric Carbon Dioxide (CO2) exchange. The implementation employs OPENACC compiler directives to expose inherent data parallelism while maintaining a single portable source code for both CPU and GPU architectures. The optimization strategy addresses fundamental incompatibilities between the original CPU-oriented algorithms and GPU execution requirements through three principal modifications: (i) elimination of runtime string comparisons via pre-computed integer-indexed lookup tables constructed during initialization, (ii) temporal decomposition of computations into sections with distinct update frequencies (initialization (once), boundary conditions (periodic), temporal scaling (hourly), and emission kernels (every timestep)) to minimize redundant calculations, and (iii) reorganization of loop hierarchies to maximize memory coalescing across ICON’s horizontal grid dimension. The hybrid CPU–GPU architecture executes initialization and I/O-intensive boundary updates on the host while offloading the computationally dominant temporal scaling and emission kernels to the device, with all tracer tendencies accumulated in device-resident arrays to eliminate host–device transfers within the physics integration loop. Scientific equivalence between CPU and GPU implementations is validated using a probabilistic testing framework, which establishes tolerance envelopes from perturbed CPU ensemble runs. This is consistent with spatial comparisons of simulated CO2 concentration fields over a European domain, which demonstrate pixel-wise agreement within ±0.01 %, corresponding to the numerical noise floor from floating-point arithmetic variations. Performance benchmarks on NVIDIA Grace–Hopper (GH200) superchips of the ALPS supercomputer in Switzerland reveal speedups of ∼1.9× on a single GPU node to 5.8× on five GPU nodes relative to a 278-rank CPU baseline. Despite 54–93 % higher absolute energy consumption on GPUs, the energy-delay product improves by 20–67 %, demonstrating favorable computational efficiency for time-critical applications. These advances enable ensemble-based atmospheric inversions with hundreds of tracers at kilometer-scale resolution within operationally feasible time constraints, supporting assimilation of dense satellite observations from instruments such as TROPOMI and the forthcoming CO2M mission. The GPU-enabled modules have been successfully deployed in recent ICON-ART applications for regional greenhouse-gas inversion and ensemble data assimilation. The single-source, directive-based implementation ensures long-term maintainability and portability across GPU generations while providing a reusable template for accelerating additional model components.

Numerical weather prediction and climate models require continuous adaptation to take advantage of advances in high-performance computing hardware. This paper presents the port of the ICON model to GPUs using OpenACC compiler directives for numerical weather prediction applications. In the context of an end-to-end operational forecast application, we adopted a full-port strategy: the entire workflow, from physical parameterizations to data assimilation, was analyzed and ported to GPUs as needed. Performance tuning and mixed-precision optimization yield a 5.5× speed-up compared to the CPU baseline in a socket-to-socket comparison. The ported ICON model meets strict requirements for time-to-solution and meteorological quality, in order for MeteoSwiss to be the first national weather service to run ICON operationally on GPUs with its ICON-CH1-EPS and ICON-CH2-EPS ensemble forecasting systems. We discuss key performance strategies, operational challenges, and the broader implications of transitioning community models to GPU-based platforms.

Numerical weather prediction and climate models require continuous adaptation to take advantage of advances in high-performance computing hardware. This paper presents the port of the ICON model to GPUs using OpenACC compiler directives for numerical weather prediction applications. In the context of an end-to-end operational forecast application, we adopted a full-port strategy: the entire workflow, from physical parameterizations to data assimilation, was analyzed and ported to GPUs as needed. Performance tuning and mixed-precision optimization yield a 5.6x speed-up compared to the CPU baseline in a socket-to-socket comparison. The ported ICON model meets strict requirements for time-to-solution and meteorological quality, in order for MeteoSwiss to be the first national weather service to run ICON operationally on GPUs with its ICON-CH1-EPS and ICON-CH2-EPS ensemble forecasting systems. We discuss key performance strategies, operational challenges, and the broader implications of transitioning community models to GPU-based platforms.

Numerical weather prediction and climate models require continuous adaptation to take advantage of advances in high-performance computing hardware. This paper presents the port of the ICON model to GPUs using OpenACC compiler directives for numerical weather prediction applications. In the context of an end-to-end operational forecast application, we adopted a full-port strategy: the entire workflow, from physical parameterizations to data assimilation, was analyzed and ported to GPUs as needed. Performance tuning and mixed-precision optimization yield a 5.6x speed-up compared to the CPU baseline in a socket-to-socket comparison. The ported ICON model meets strict requirements for time-to-solution and meteorological quality, in order for MeteoSwiss to be the first national weather service to run ICON operationally on GPUs with its ICON-CH1-EPS and ICON-CH2-EPS ensemble forecasting systems. We discuss key performance strategies, operational challenges, and the broader implications of transitioning community models to GPU-based platforms.

Numerical weather prediction and climate models require continuous adaptation to take advantage of advances in high-performance computing hardware. This paper presents the port of the ICON model to GPUs using OpenACC compiler directives for numerical weather prediction applications. In the context of an end-to-end operational forecast application, we adopted a full-port strategy: the entire workflow, from physical parameterizations to data assimilation, was analyzed and ported to GPUs as needed. Performance tuning and mixed-precision optimization yield a 5.6x speed-up compared to the CPU baseline in a socket-to-socket comparison. The ported ICON model meets strict requirements for time-to-solution and meteorological quality, in order for MeteoSwiss to be the first national weather service to run ICON operationally on GPUs with its ICON-CH1-EPS and ICON-CH2-EPS ensemble forecasting systems. We discuss key performance strategies, operational challenges, and the broader implications of transitioning community models to GPU-based platforms.

Emission inventories serve as crucial input for atmospheric chemistry transport models. To make them usable for a model simulation, they have to be pre-processed and, traditionally, provided as input files at discrete model time steps. In this paper, we present an “online” approach, which produces a minimal number of input data read-in at the beginning of a simulation and which handles essential processing steps online during the simulation. For this purpose, a stand-alone Python package “emiproc” was developed, which projects the inventory data to the model grid and generates temporal and vertical scaling profiles for individual emission categories. The package is also able to produce “offline” emission files if desired. Furthermore, we outline the concept of the online emission module (written in Fortran 90) and demonstrate its implementation in two different atmospheric transport models: COSMO-GHG and COSMO-ART. Simulation results from both modeling systems show the equivalence of the online and offline procedure. While the model run time is very similar for both approaches, input size and pre-processing time are greatly reduced when online emissions are utilized.

Emission inventories serve as crucial input for atmospheric chemistry transport models. To make them usable for a model simulation, they have to be pre-processed and, traditionally, provided as input files at discrete model time steps. In this paper, we present an online approach, which produces a minimal number of input data read in at the beginning of a simulation and which handles essential processing steps online during the simulation. For this purpose, a stand-alone Python package emiproc was developed, which projects the inventory data to the model grid and generates temporal and vertical scaling profiles for individual emission categories. The package is also able to produce offline emission files if desired. Furthermore, we outline the concept of the online emission module (written in Fortran 90) and demonstrate its implementation in two different atmospheric transport models, COSMO-GHG and COSMO-ART. Simulation results from both modeling systems show the equivalence of the online and offline procedure. While the model run-time is very similar for both approaches, disk storage and pre-processing time are greatly reduced when online emissions are utilized.

Airborne mineral dust is a key player in the Earth system and shows manifold impacts on atmospheric properties such as the radiation budget and cloud microphysics. Investigations of smoke plumes originating from wildfires found significant fractions of mineral dust within these plumes – most likely raised by strong, turbulent fire-related winds. This study presents and revisits a conceptual model describing the emission of mineral dust particles during wildfires. This is achieved by means of high-resolution large-eddy simulation (LES), conducted with the All Scale Atmospheric Model (ASAM). The impact of (a) different fire properties representing idealized grassland and shrubland fires, (b) different ambient wind conditions modulated by the fire's energy flux, and (c) the wind's capability to mobilize mineral dust particles was investigated. Results from this study illustrate that the energy release of the fire leads to a significant increase in near-surface wind speed, which consequently enhances the dust uplift potential. This is in particular the case within the fire area where vegetation can be assumed to be widely removed and uncovered soil is prone to wind erosion. The dust uplift potential is very sensitive to fire properties, such as fire size, shape, and intensity, but also depends on the ambient wind velocity. Although measurements already showed the importance of wildfires for dust emissions, pyro-convection is so far neglected as a dust emission process in atmosphere–aerosol models. The results presented in this study can be seen as the first step towards a systematic parameterization representing the connection between typical fire properties and related dust emissions.