Abstract:

[Context] With the rapid development of meteorological informatization, the required computing power has continued to increase and become increasingly heterogeneous, and distributed across multiple locations, raising the challenges such as difficult resource integration, inefficient management, and imbalanced workload. [Objective] To address the limitations of traditional usage paradigm, a unified management service platform for meteorological computing power was designed and implemented, aiming to provide comprehensive support for numerical weather prediction and research. [Methods] The platform adopts a hierarchical design, consisting of resource, management, and service layers. It integrates unified identity authentication for multi-dimensional authorization and automatic permission matching, offering web-based unified access and REST API-based scheduling and job distribution. [Results] The platform has been developed and deployed in the Numerical Weather Prediction Center, helping to improve resource utilization and management efficiency. [Conclusions] The platform provides a novel approach to resource utilization for meteorological forecasting and research, enhancing computing power services and support capabilities, and demonstrating significant practical value and broad application potential.

Key words: meteorological computing resources, unified computing resource management and service, unified identity authentication, resource access control, REST API