As one of the world’s leading operational meteorology and climate research services, the Korea Meteorological Administration (KMA) has a 365-day-a-year responsibility with global scope. A governmental organization of the Republic of Korea, KMA helps protect the nation’s safety and economic welfare with reliable meteorological observation, analysis and forecast services. In addition, the international community looks to KMA for its meteorological data and research in weather and climate prediction, climate change, air pollution monitoring, and earthquake and marine meteorology.
Weather forecasting is regarded as an important element affecting the socioeconomic welfare of the globe — a situation demanding a highly reliable weather forecasting system. In response, KMA scientists and researchers use computationally demanding numerical weather prediction (NWP) models to calculate the atmospheric movements and physical processes that cause weather changes.
Specifically, KMA runs the Unified Model (UM) developed by the United Kingdom Met Office. The KMA UM operational forecast suite spans a wide range of time and space scales, including kilometer-scale mesoscale models, limited-area regional models, global models, seasonal forecasts, global and regional climate predictions and an ensemble prediction system. This operational forecast suite and its accompanying research workload require performance pushing a petaflop and multi-petabyte data management capabilities.
Given its demanding operational and research requirements, KMA needed superior sustained application performance and a robust overall solution. They selected the powerful Cray® XC40™ supercomputer, featuring the Aries™ interconnect. KMA’s system is composed of three Cray XC40 supercomputers (for a combined total of 6.2 PF of peak performance) with a multi-tier, multi-petabyte Cray® Sonexion® Lustre® storage solution and additional archive and back-up systems. This configuration gives KMA the necessary performance to advance its NWP work. For example, with the Cray XC40 supercomputer KMA can significantly increase the spatial resolutions of the medium range global UM model from N512 (25 km) to N768 (17 km), enhance the resolutions of the data assimilation, both enhance resolution and number of members in the medium-range ensemble prediction system and move to a newer UM dynamical core. Overall these improvements enhanced the performance of the global deterministic and ensemble prediction systems by about 10 percent. The Cray XC40 systems replace KMA’s previous Cray® XE6™ supercomputers.
KMA’s Cray XC40 system is the largest integrated solution for operational numerical weather prediction in the Asia-Pacific region and one of the world’s largest, strongly positioning KMA as a leader in operational meteorology.
The KMA-Cray Earth System Research Center
KMA, as an operational meteorology leader, and Cray, as a supercomputing leader, continue to work together on the successful Earth System Research Center (ESRC), a cooperative venture established in 2005 to advance the science of earth system modeling over the East-Asia Pacific region and promote the use of advanced high performance computing facilities.