基于超算的全球电离层模型快速并行解算

为应对日益丰富的观测数据以及数据再处理对高性能计算的需求，开发了基于OpenMP以及MPI（Message Passing Interface）并行计算的全球电离层快速建模算法。采用武汉大学超级计算机对全球电离层建模效率进行了不同并行计算方案的实验。结果表明，采用多节点MPI并行计算能够极大地提高数据处理效率，相比传统单节点串行计算提高了近30倍，相比单节点OpenMP并行计算提高了近3~4倍。MPI并行计算方案充分利用了丰富的计算机资源来提高全球电离层建模效率，对电离层建模算法的快速测试、产品的重新再处理具有重要作用，对多系统全球导航卫星系统（Global Navigation Satellite System，GNSS）快速精密定轨、大规模GNSS网解也有较好的参考价值。

Rapid Parallel Estimation of Global Ionospheric Model Based on Supercomputer

An Algorithm of global ionospheric rapid modeling based on OpenMP and MPI (message passing interface) parallel computing is developed to meet the requirement of high performance computing for the massive data and reprocessing. The supercomputer of Wuhan University is used for modeling the global ionosphere with kinds of parallel computing approaches. The results indicate that it can significantly improve the efficiency of data processing by applying multi nodes with MPI parallel computing technology. Compared to traditional single node with serial computing and parallel computing, the efficiency has increased by 30 times and 3 to 4 times, respectively. The MPI parallel computing approach takes full advantage of rich computing resources to promote the efficiency of global ionospheric modeling. It is very useful for rapid test of algorithms and reprocessing of modeling. Also, it has reference significance to precise orbit determination of multi GNSS (Global Navigation Satellite System) as well as huge GNSS network solutions.