Search

The north and south of the Alps, as well as the eastern shores of the Adriatic Sea, are hot spots of severe convective storms, including hail and lightning associated with deep convection. With advancements in computing power, it has become feasible to simulate deep convection explicitly in climate models by decreasing the horizontal grid spacing to less than 4 km. These kilometer-scale models improve the representation of orography and reduce uncertainties associated with the use of deep convection parameterizations. In this study, we perform km-scale simulations for eight observed cases of severe convective storms (seven with and one without observed hail) over the Alpine-Adriatic region. The simulations are performed with the climate version of the regional model Consortium for Small-scale Modeling (COSMO) that runs on graphics processing units (GPUs) at a horizontal grid spacing of 2.2 km. To analyze hail and lightning we have explored the hail growth model (HAILCAST) and lightning potential index (LPI) diagnostics integrated with the COSMO-crCLIM model. Comparison with available high-resolution observations reveals good performance of the model in simulating total precipitation, hail, and lightning. By performing a detailed analysis of three of the case studies, we identified the importance of significant meteorological factors for heavy thunderstorms that were reproduced by the model. Among these are the moist unstable boundary layer and dry mid-level air, the topographic barrier, as well as an approaching upper-level trough and cold front. Although COSMO HAILCAST tends to underestimate the hail size on the ground, the results indicate that both HAILCAST and LPI are promising candidates for future climate research. [ABSTRACT FROM AUTHOR]

Progress in leveraging current and emerging high-performance computing infrastructures using traditional weather and climate models has been slow. This has become known more broadly as the software productivity gap. With the end of Moore's law driving forward rapid specialization of hardware architectures, building simulation codes on a low-level language with hardware-specific optimizations is a significant risk. As a solution, we present Pace, an implementation of the nonhydrostatic FV3 dynamical core and GFDL cloud microphysics scheme which is entirely Python-based. In order to achieve high performance on a diverse set of hardware architectures, Pace is written using the GT4Py domain-specific language. We demonstrate that with this approach we can achieve portability and performance, while significantly improving the readability and maintainability of the code as compared to the Fortran reference implementation. We show that Pace can run at scale on leadership-class supercomputers and achieve performance speeds 3.5–4 times faster than the Fortran code on GPU-accelerated supercomputers. Furthermore, we demonstrate how a Python-based simulation code facilitates existing or enables entirely new use cases and workflows. Pace demonstrates how a high-level language can insulate us from disruptive changes, provide a more productive development environment, and facilitate the integration with new technologies such as machine learning. [ABSTRACT FROM AUTHOR]

Progress in leveraging current and emerging high-performance computing infrastructures using traditional weather and climate models has been slow. This has become known more broadly as the software productivity gap. With the end of Moore's Law driving forward rapid specialization of hardware architectures, building simulation codes on a low-level language with hardware specific optimizations is a significant risk. As a solution, we present Pace, an implementation of the nonhydrostatic FV3 dynamical core which is entirely Python-based. In order to achieve high performance on a diverse set of hardware architectures, Pace is written using the GT4Py domain-specific language. We demonstrate that with this approach we can achieve portability and performance, while significantly improving the readability and maintainability of the code as compared to the Fortran reference implementation. We show that Pace can run at scale on leadership-class supercomputers and achieve performance speeds 3.5–4 times faster than the Fortran code on GPU-accelerated supercomputers. Furthermore, we demonstrate how a Python-based simulation code facilitates existing or enables entirely new use-cases and workflows. Pace demonstrates how a high-level language can insulate us from disruptive changes, provide a more productive development environment, and facilitate the integration with new technologies such as machine learning. [ABSTRACT FROM AUTHOR]

Simulation software in geophysics is traditionally written in Fortran or C++ due to the stringent performance requirements these codes have to satisfy. As a result, researchers who use high-productivity languages for exploratory work often find these codes hard to understand, hard to modify, and hard to integrate with their analysis tools. fv3gfs-wrapper is an open-source Python-wrapped version of the NOAA (National Oceanic and Atmospheric Administration) FV3GFS (Finite-Volume Cubed-Sphere Global Forecast System) global atmospheric model, which is coded in Fortran. The wrapper provides simple interfaces to progress the Fortran main loop and get or set variables used by the Fortran model. These interfaces enable a wide range of use cases such as modifying the behavior of the model, introducing online analysis code, or saving model variables and reading forcings directly to and from cloud storage. Model performance is identical to the fully compiled Fortran model, unless routines to copy the state in and out of the model are used. This copy overhead is well within an acceptable range of performance and could be avoided with modifications to the Fortran source code. The wrapping approach is outlined and can be applied similarly in other Fortran models to enable more productive scientific workflows. [ABSTRACT FROM AUTHOR]

Inverse modeling of anthropogenic and biospheric CO2 fluxes from ground-based and satellite observations critically depends on the accuracy of atmospheric transport simulations. Previous studies emphasized the impact of errors in simulated winds and vertical mixing in the planetary boundary layer, whereas the potential importance of releasing emissions not only at the surface but distributing them in the vertical was largely neglected. Accounting for elevated emissions may be critical, since more than 50 % of CO2 in Europe is emitted by large point sources such as power plants and industrial facilities. In this study, we conduct high-resolution atmospheric simulations of CO2 with the mesoscale Consortium for Small-scale Modeling model extended with a module for the simulation of greenhouse gases (COSMO-GHG) over a domain covering the city of Berlin and several coal-fired power plants in eastern Germany, Poland and Czech Republic. By including separate tracers for anthropogenic CO2 emitted only at the surface or according to realistic, source-dependent profiles, we find that releasing CO2 only at the surface overestimates near-surface CO2 concentrations in the afternoon on average by 14 % in summer and 43 % in winter over the selected model domain. Differences in column-averaged dry air mole XCO2 fractions are smaller, between 5 % in winter and 8 % in summer, suggesting smaller yet non-negligible sensitivities for inversion modeling studies assimilating satellite rather than surface observations. The results suggest that the traditional approach of emitting CO2 only at the surface is problematic and that a proper allocation of emissions in the vertical deserves as much attention as an accurate simulation of atmospheric transport. [ABSTRACT FROM AUTHOR]

Inverse modeling of anthropogenic and biospheric CO2 fluxes from ground-based and satellite observations critically depends on the accuracy of atmospheric transport simulations. Previous studies emphasized the impact of errors in simulated winds and vertical mixing in the planetary boundary layer, whereas the potential importance of releasing emissions not only at the surface but distributing them in the vertical was largely neglected. Accounting for elevated emissions may be critical, since more than 50% of CO2 in Europe is emitted by large point sources such as power plants and industrial facilities. In this study we conduct high-resolution atmospheric simulations of CO2 with the mesoscale model COSMO-GHG over a domain covering the city of Berlin and several coal-fired power plants in eastern Germany, Poland and the Czech Republic. By including separate tracers for anthropogenic CO2 emitted only at the surface or according to realistic, source-dependent profiles, we find that releasing CO2 only at the surface overestimates near-surface CO2 concentrations in the afternoon on average by 14% in summer and 43% in winter over the selected model domain. Differences in column mean dry air mole fractions XCO2 are smaller, between 5% in winter and 8% in summer, suggesting smaller yet non-negligible sensitivities for inversion modeling studies assimilating satellite rather than surface observations. The results suggests that the traditional approach of emitting CO2 only at the surface is problematic and that a proper allocation of emissions in the vertical deserves as much attention as an accurate simulation of atmospheric transport. [ABSTRACT FROM AUTHOR]

Climate models project an increase in heavy precipitation events in response to greenhouse gas forcing. Important elements of such events are rain showers and thunderstorms, which are poorly represented in models with parameterized convection. In this study, simulations with 12 km horizontal grid spacing (convection-parameterizing model, CPM) and 2 km grid spacing (convection-resolving model, CRM) are employed to investigate the change in the diurnal cycle of convection with warmer climate. For this purpose, simulations of 11 days in June 2007 with a pronounced diurnal cycle of convection are compared with surrogate simulations from the same period. The surrogate climate simulations mimic a future climate with increased temperatures but unchanged relative humidity and similar synoptic-scale circulation. Two temperature scenarios are compared: one with homogeneous warming (HW) using a vertically uniform warming and the other with vertically dependent warming (VW) that enables changes in lapse rate. The two sets of simulations with parameterized and explicit convection exhibit substantial differences, some of which are well known from the literature. These include differences in the timing and amplitude of the diurnal cycle of convection, and the frequency of precipitation with low intensities. The response to climate change is much less studied. We can show that stratification changes have a strong influence on the changes in convection. Precipitation is strongly increasing for HW but decreasing for the VW simulations. For cloud type frequencies, virtually no changes are found for HW, but a substantial reduction in high clouds is found for VW. Further, we can show that the climate change signal strongly depends upon the horizontal resolution. In particular, significant differences between CPM and CRM are found in terms of the radiative feedbacks, with CRM exhibiting a stronger negative feedback in the top-ofthe- atmosphere energy budget. [ABSTRACT FROM AUTHOR]

The best hope for reducing long-standing global climate model biases is by increasing resolution to the kilometer scale. Here we present results from an ultrahighresolution non-hydrostatic climate model for a near-global setup running on the full Piz Daint supercomputer on 4888 GPUs (graphics processing units). The dynamical core of the model has been completely rewritten using a domainspecific language (DSL) for performance portability across different hardware architectures. Physical parameterizations and diagnostics have been ported using compiler directives. To our knowledge this represents the first complete atmospheric model being run entirely on accelerators on this scale. At a grid spacing of 930m (1.9 km), we achieve a simulation throughput of 0.043 (0.23) simulated years per day and an energy consumption of 596MWh per simulated year. Furthermore, we propose a new memory usage efficiency (MUE) metric that considers how efficiently the memory bandwidth - the dominant bottleneck of climate codes - is being used. [ABSTRACT FROM AUTHOR]

The best hope for reducing long-standing global climate model biases, is through increasing the resolution to the kilometer scale. Here we present results from an ultra-high resolution non-hydrostatic climate model for a near-global setup running on the full Piz Daint supercomputer on 4888 GPUs. The dynamical core of the model has been completely rewritten using a domain-specific language (DSL) for performance portability across different hardware architectures. Physical parameterizations and diagnostics have been ported using compiler directives. To our knowledge this represents the first complete atmospheric model being run entirely on accelerators at this scale. At a grid spacing of 930 m (1.9 km), we achieve a simulation throughput of 0.043 (0.23) simulated years per day and an energy consumption of 596 MWh per simulated year. Furthermore, we propose the new memory usage efficiency metric that considers how efficiently the memory bandwidth - the dominant bottleneck of climate codes - is being used. [ABSTRACT FROM AUTHOR]

Climate models project an increase in heavy precipitation events in response to greenhouse gas forcing. Important elements of such events are rain showers and thunderstorms, which are poorly represented in models with parameterized convection. In this study, simulations with 12 km horizontal grid spacing (convection-parameterizing model, CPM) and 2 km grid spacing (convection-resolving model, CRM), and with either a one-moment microphysics scheme (1M) or a two-moment microphysics scheme (2M) are employed to investigate the change in the diurnal cycle of convection with warmer climate. For this purpose, simulations of 11 days in June 2007 with a pronounced diurnal cycle of convection are compared with surrogate simulations from the same period. The surrogate climate simulations mimic a future climate with increased temperatures, but unchanged relative humidity and synoptic-scale circulation. Two temperature scenarios are compared, one with homogeneous warming (HW) using a vertically uniform warming, the other with vertically-dependent warming (VW) that enables changes in lapse rate. The two sets of simulations with parameterized and explicit convection exhibit substantial differences, which are well known from the literature. These include differences in the timing and amplitude of the diurnal cycle of convection, and the frequency of precipitation with low intensities. There are also significant differences in terms of the response to the surrogate warming. For CRM, an increase of hourly heavy precipitation events is found for both surrogate scenarios and microphysics schemes. The intensification is consistent with the Clausius-Clapeyron relation. For cloud type frequencies, virtually no changes are found for HW, but a substantial reduction in high clouds is found for VW. Some of the CPM sensitivities differ significantly. Importantly, the increase of heavy precipitation events simulated by CPM is larger than suggested by the Clausius-Clapeyron relation. Moreover, significant differences between CPM and CRM are found in terms of the radiative feedbacks, with CRM exhibiting a stronger negative feedback in the top of the atmosphere energy budget. [ABSTRACT FROM AUTHOR]

The representation of moist convection in climate models represents a major challenge, due to the small scales involved. Using horizontal grid spacings of O(1km), convection-resolving weather and climate models allows one to explicitly resolve deep convection. However, due to their extremely demanding computational requirements, they have so far been limited to short simulations and/or small computational domains. Innovations in supercomputing have led to new hybrid node designs, mixing conventional multi-core hardware and accelerators such as graphics processing units (GPUs). One of the first atmospheric models that has been fully ported to these architectures is the COSMO (Consortium for Small-scale Modeling) model. Here we present the convection-resolving COSMO model on continental scales using a version of the model capable of using GPU accelerators. The verification of a week-long simulation containing winter storm Kyrill shows that, for this case, convection-parameterizing simulations and convectionresolving simulations agree well. Furthermore, we demonstrate the applicability of the approach to longer simulations by conducting a 3-month-long simulation of the summer season 2006. Its results corroborate the findings found on smaller domains such as more credible representation of the diurnal cycle of precipitation in convection-resolving models and a tendency to produce more intensive hourly precipitation events. Both simulations also show how the approach allows for the representation of interactions between synopticscale and meso-scale atmospheric circulations at scales ranging from 1000 to 10 km. This includes the formation of sharp cold frontal structures, convection embedded in fronts and small eddies, or the formation and organization of propagating cold pools. Finally, we assess the performance gain from using heterogeneous hardware equipped with GPUs relative to multi-core hardware. With the COSMO model, we now use a weather and climate model that has all the necessary modules required for real-case convection-resolving regional climate simulations on GPUs. [ABSTRACT FROM AUTHOR]