No abstract

Very high-resolution spectral transform models are believed to become prohibitively expensive because of the relative increase in computational cost of the Legendre transforms compared to the gridpoint computations. This article describes the implementation of a practical fast spherical harmonics transform into the Integrated Forecast System (IFS) at ECMWF. Details of the accuracy of the computations, of the parallelization, and memory use are discussed. Results are presented that demonstrate the cost effectiveness and accuracy of the fast spherical harmonics transform, successfully mitigating the concern about the disproportionally growing computational cost. Using the new transforms, the first T7999 global weather forecast (equivalent to '2.5-km horizontal grid size) using a spectral transform model has been produced.

A specific interface between the radiation transfer calculations and the rest of the ECMWF model was introduced in 2003, potentially providing substantial economy in computer time by reducing the spatial resolution at which radiation transfer is evaluated, without incurring some of the deficiencies produced by the sampling strategy previously used in the ECMWF model. The introduction of a new more-computerintensive radiation package (McRad) in June 2007 has led to a differentiated use of this interface depending on the applications. The history of the interface, how it is used, and its impact when using the new radiation scheme are discussed here. For a given model resolution, the impact of a lower-resolution radiation grid on the model behavior is studied here, in the context of 10-day forecasts at high resolution (T L 799L91), of medium-resolution forecasts (T L 399L62) used in the Ensemble Prediction System (EPS), and of lowresolution simulations (T L 159L91) as used for model development and seasonal forecasts with an interactive ocean. Results for the high-resolution forecasts are compared in terms of objective scores and of the quality of "surface" parameters (total cloud cover, 2-m temperature and specific humidity, and 10-m wind) usually verified in a meteorological context. For the medium-resolution forecasts, the impact of the radiation grid is studied in terms of the potential increase in the efficiency of the EPS system without deteriorating the probabilistic skill. The impact of changes in the radiation grid resolution on the low-resolution versions of model is discussed in terms of cloud-radiation interactions and ocean surface temperature. For these operational applications, a radiation grid with a coarsening factor even as large as 2.5 for T L 799L91 and T L 159L91 and 4.2 for the EPS T L 399L62 is shown to give results free of any systematic differences linked to the spatial interpolation and to the coarser resolution of both the inputs to and the outputs from the radiation transfer schemes.

No abstract

Abstract. In the simulation of complex multi-scale flows arising in weather and climate modelling, one of the biggest challenges is to satisfy strict service requirements in terms of time to solution and to satisfy budgetary constraints in terms of energy to solution, without compromising the accuracy and stability of the application. These simulations require algorithms that minimise the energy footprint along with the time required to produce a solution, maintain the physically required level of accuracy, are numerically stable, and are resilient in case of hardware failure. The European Centre for Medium-Range Weather Forecasts (ECMWF) led the ESCAPE (Energy-efficient Scalable Algorithms for Weather Prediction at Exascale) project, funded by Horizon 2020 (H2020) under the FET-HPC (Future and Emerging Technologies in High Performance Computing) initiative. The goal of ESCAPE was to develop a sustainable strategy to evolve weather and climate prediction models to next-generation computing technologies. The project partners incorporate the expertise of leading European regional forecasting consortia, university research, experienced high-performance computing centres, and hardware vendors. This paper presents an overview of the ESCAPE strategy: (i) identify domain-specific key algorithmic motifs in weather prediction and climate models (which we term Weather & Climate Dwarfs), (ii) categorise them in terms of computational and communication patterns while (iii) adapting them to different hardware architectures with alternative programming models, (iv) analyse the challenges in optimising, and (v) find alternative algorithms for the same scheme. The participating weather prediction models are the following: IFS (Integrated Forecasting System); ALARO, a combination of AROME (Application de la Recherche à l'Opérationnel à Meso-Echelle) and ALADIN (Aire Limitée Adaptation Dynamique Développement International); and COSMO–EULAG, a combination of COSMO (Consortium for Small-scale Modeling) and EULAG (Eulerian and semi-Lagrangian fluid solver). For many of the weather and climate dwarfs ESCAPE provides prototype implementations on different hardware architectures (mainly Intel Skylake CPUs, NVIDIA GPUs, Intel Xeon Phi, Optalysys optical processor) with different programming models. The spectral transform dwarf represents a detailed example of the co-design cycle of an ESCAPE dwarf. The dwarf concept has proven to be extremely useful for the rapid prototyping of alternative algorithms and their interaction with hardware; e.g. the use of a domain-specific language (DSL). Manual adaptations have led to substantial accelerations of key algorithms in numerical weather prediction (NWP) but are not a general recipe for the performance portability of complex NWP models. Existing DSLs are found to require further evolution but are promising tools for achieving the latter. Measurements of energy and time to solution suggest that a future focus needs to be on exploiting the simultaneous use of all available resources in hybrid CPU–GPU arrangements.