Parallel GPU architecture framework for the WRF Single Moment 6-class microphysics scheme

Huang, M.; Huang, Bormin; Gu, Lingjia; Huang, Hung Chang; Goldberg, Mitchell D.

doi:10.1016/j.cageo.2015.06.014

Cited by 11 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We followed the physics option set recommended for 1-4 km grid distances in the WRF model user's guide (https://www2.mmm.ucar.edu/wrf/users/ docs/user_guide_V3.9/contents.html) (accessed on 2 December 2023), with adjustments made to surface layer and boundary layer parameter schemes. Consequently, the physical parameterizations utilized in this study included the WRF Single-Moment 6-Class Microphysics Scheme (WSM6) [28], the YSU boundary layer scheme [29], the RRTMG longwave and shortwave radiation scheme [30], the unified Noah land surface scheme [31], and the revised MM5 Monin-Obukhov surface layer scheme [32]. Cumulus parameterization was deactivated.…”

Section: Variational Assimilation Methods and Observation Operatorsmentioning

confidence: 99%

Error Model for the Assimilation of All-Sky FY-4A/AGRI Infrared Radiance Observations

Pu,

2024

Sensors

View full text Add to dashboard Cite

The Advanced Geostationary Radiation Imager (AGRI) carried by the FengYun-4A (FY-4A) satellite enables the continuous observation of local weather. However, FY-4A/AGRI infrared satellite observations are strongly influenced by clouds, which complicates their use in all-sky data assimilation. The presence of clouds leads to increased uncertainty, and the observation-minus-background (O−B) differences can significantly deviate from the Gaussian distribution assumed in the variational data assimilation theory. In this study, we introduce two cloud-affected (Ca) indices to quantify the impact of cloud amount and establish dynamic observation error models to address biases between O−B and Gaussian distributions when assimilating all-sky data from FY-4A/AGRI observations. For each Ca index, we evaluate two dynamic observation error models: a two-segment and a three-segment linear model. Our findings indicate that the three-segment linear model we propose better conforms to the statistical characteristics of FY-4A/AGRI observations and improves the Gaussianity of the O−B probability density function. Dynamic observation error models developed in this study are capable of handling cloud-free or cloud-affected FY-4A/AGRI observations in a uniform manner without cloud detection.

show abstract

Section: Variational Assimilation Methods and Observation Operatorsmentioning

confidence: 99%

Error Model for the Assimilation of All-Sky FY-4A/AGRI Infrared Radiance Observations

Pu,

2024

Sensors

View full text Add to dashboard Cite

show abstract

“…An implementation of WSM6 on CUDA [11] obtained a speedup of up to 246 using a K40 GPU and up to 295 using two GPUs when compared to a single-threaded CPU. Kim et al [12] implemented the WSM6 using OpenACC for Model for Prediction Across Scales (MPAS) [1].…”

Section: Background and Related Workmentioning

confidence: 99%

Runge-Kutta Method and WSM6 Microphysics for Weather Prediction on Hybrid Parallel Platform

Silva,

Stefanes,

Capistrano

2023

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

In Numerical Weather Prediction (NWP) models we need to model the dynamics of the atmosphere and the physical variables that take place in a moment. In grid point models the temporal evolution of the model variables are calculated in a 3D grid which covers the atmosphere from the surface up to the model top. NWP models include two import routine namely Runge-Kutta method and microphysics scheme WSM6. This paper describes advances in the performance of the Runge-Kutta method and WSM6 microphysics by exploiting multi-level parallelism using CUDA-based GPU on hybrid parallel platform. We applied pipeline parallelism technique, workload balancing and asynchronous data exchange strategy each demonstrating be useful to improve performance. Our experiments show that the solution is scalable. We also realized an analysis of the accuracy of our implementation with good results.

show abstract

“…The WRF five-layer thermal diffusion scheme was accelerated using CUDA C, and a 311× speedup was obtained on one Tesla K40 GPU [26]. The WRF Single Moment 6-class microphysics scheme was also accelerated using CUDA C, and obtained a 216× speedup on one Tesla K40 GPU [27].…”

Section: Related Workmentioning

confidence: 99%

A Novel GPU-Based Acceleration Algorithm for a Longwave Radiative Transfer Model

et al. 2020

View full text Add to dashboard Cite

Graphics processing unit (GPU)-based computing for climate system models is a longstanding research area of interest. The rapid radiative transfer model for general circulation models (RRTMG), a popular atmospheric radiative transfer model, can calculate atmospheric radiative fluxes and heating rates. However, the RRTMG has a high calculation time, so it is urgent to study its GPU-based efficient acceleration algorithm to enable large-scale and long-term climatic simulations. To improve the calculative efficiency of radiation transfer, this paper proposes a GPU-based acceleration algorithm for the RRTMG longwave radiation scheme (RRTMG_LW). The algorithm concept is accelerating the RRTMG_LW in the g- p o i n t dimension. After implementing the algorithm in CUDA Fortran, the G-RRTMG_LW was developed. The experimental results indicated that the algorithm was effective. In the case without I/O transfer, the G-RRTMG_LW on one K40 GPU obtained a speedup of 30.98× over the baseline performance on one single Intel Xeon E5-2680 CPU core. When compared to its counterpart running on 10 CPU cores of an Intel Xeon E5-2680 v2, the G-RRTMG_LW on one K20 GPU in the case without I/O transfer achieved a speedup of 2.35×.

show abstract

Parallel GPU architecture framework for the WRF Single Moment 6-class microphysics scheme

Cited by 11 publications

References 14 publications

Error Model for the Assimilation of All-Sky FY-4A/AGRI Infrared Radiance Observations

Error Model for the Assimilation of All-Sky FY-4A/AGRI Infrared Radiance Observations

Runge-Kutta Method and WSM6 Microphysics for Weather Prediction on Hybrid Parallel Platform

A Novel GPU-Based Acceleration Algorithm for a Longwave Radiative Transfer Model

Contact Info

Product

Resources

About