Meteorology
Scenarios
High-performance computers have long been essential to the meteorology industry. In weather forecast, many factors need to be identified by numerical predictions, which lead to massive computing requirements. Three methods, including statistical forecast, synoptic forecast, and numerical weather prediction (NWP) have been used for weather forecast. Among them, the statistical forecast is based on the meteorological observation data accumulated over the years. It is generally used for mid- and long-term (more than 10 days) forecast, but the accuracy is low. In the mid- (3–10 days) and short-term (12 hours – 3 days) and ultra-short-term (12 hours) weather forecast, the synoptic forecast and NWP are mainly used. In particular, NWP becomes the most important method for more accurate weather forecast. Table 1 describes the weather forecast models and application.
Mesoscale Weather Forecast Model |
Description |
Application |
|---|---|---|
MM5 |
MM5 is a fifth-generation mesoscale weather forecast model jointly developed by the National Atmospheric Research Center (NCAR) and Pennsylvania State University. |
MM5 is a mesoscale and microscale non-hydrostatic dynamical model designed for meteorological simulations. It is the most widely used mesoscale forecast model. The final iteration of the model is MM5V3. However, its active research and development have concluded as the community has transitioned its focus to the next-generation WRF model. |
WRF |
The WRF model is a next-generation mesoscale model developed based on the MM5 model. It represents a collaborative research initiative involving numerous U.S. research institutions and universities, functioning as both an advanced forecasting model and a sophisticated data assimilation system. |
The WRF model combines portability with easy maintenance and use. It also features high scalability and efficiency, making it a widely used model. As a public-domain framework, WRF is maintained and supported by the NCAR and remains freely accessible to the public. |
GRAPES |
The Global/Regional Assimilation and Prediction System (GRAPES) model is a next-generation NWP system developed by the China Meteorological Administration (CMA). |
The long-term mission of GRAPES is the sustainable research and development of a next-generation NWP system tailored for 21st-century requirements. By steadily enhancing the forecasting accuracy, GRAPES provides the scientific and technical underpinning necessary to elevate the CMA's weather and climate forecast services, ultimately meeting the ever-growing demands of economic construction and social development. |
AREMS |
To address the evolving demands of meteorological operations, the Institute of Heavy Rain, China Meteorological Administration (IHR/CMA) and the Institute of Atmospheric Physics, Chinese Academy of Sciences (IAP, CAS) collaboratively developed the AREMS mesoscale rainstorm numerical prediction system, leveraging the REM (ETA) model as its foundational framework. |
The AREMS model demonstrates high prediction accuracy for torrential rain events in the Yangtze-Huai River Basin. It accurately forecasts the position, orientation, and movement trends of rain belts. Its forecasts for rainstorm centers and intensities align closely with observed data. Notably, for extreme precipitation within the 12–36 hour lead time, its Threat Score (TS) significantly outperforms competing models. The model has proven highly effective in forecasting trials across both basins. |
Challenges
- Massive computing workloads
Mesoscale meteorological models (MM5, WRF, GRAPES, and AREMS) require massive computing capacity. Due to the time-sensitive nature of weather forecast, these models must complete execution within a strict window, typically under two hours. Furthermore, the escalating demand for meteorological accuracy has shifted grid resolutions from hundreds of kilometers to a few kilometers, leading to a substantial increase in computing workloads. A two-fold increase in accuracy theoretically results in a sixteen-fold increase in computing intensity. These demands necessitate parallel computing for NWP, as single-CPU or conventional systems are insufficient. Consequently, models have been parallelized via message passing (MPI) or shared memory (OpenMP) frameworks, and high-performance parallel computers are required. Currently, most mesoscale models have been parallelized. While MM5 and WRF support parallel MPI, parallel OpenMP, and hybrid MPI+OpenMP execution modes, GRAPES and AREMS have also implemented MPI parallelization.
- Intensive communication
As these models are parallel software applications that generally employ finite-difference grid schemes for parallel computing, the communication overhead between CPUs during mesoscale weather forecast is significant, requiring extremely high communication performance. In models such as MM5 and WRF, communication encompasses both inter-domain exchanges between the parent and nested domains, as well as intra-domain communication across different data partitions. This makes the high-performance communication network a must-have for computers.
- Requirements for high real-time performance and scheduled running
The computing cluster for the meteorology industry needs to deliver good real-time performance to support accurate weather forecast. In addition, the forecast system needs to run automatically at fixed time without manual intervention. Generally, the scheduled jobs are completed in two to four fixed time segments every day. Each time segment lasts for two hours. These jobs run at the same time every day, and you must ensure that they can be completed on time.
- Heavy computing workload in main mode
The software processing procedure can be divided into three phases: pre-processing, main mode, and post-processing. The pre-processing phase includes document download and data assimilation. The post-processing phase covers the graphical processing of generated products. These two phases do not have high requirements on computers. The main mode is the most important part of the entire system and demands for the most computing requirements and high performance of computers.