摘要
时空离散化并行计算是一种效率较高的并行计算方法,但以往的研究主要基于MPI框架和多树编码规则。为了使应用广泛的Pfafstetter编码的分布式水文模型实现时空并行计算,采用OpenMP编程模型,为WEP⁃L分布式水文模型汇流过程设计了基于Pfafstetter编码的分层模拟单元方法和时空离散化方法。首先分层模拟单元方法根据Pfafstetter编码规则建立子流域拓扑表,使子流域按照从上游到下游的顺序进行分层,然后在此基础上考虑时间因素重新分层实现时空离散化,最后将同一层中时间及空间上互不影响的子流域模拟任务分配给不同的线程实现并行计算。选取黄河流域作为研究区域,在不同线程条件下对并行效果进行了测试,结果表明,汇流并行化提高了模型模拟效率。基于20核的CPU,仅考虑分层模拟单元方法时,加速比最大为4.14,而采用时空离散化方法时,加速比能够达到8.17。
Temporalspatial discretization parallel computing is a parallel method with high efficiency.However,previous researches are mainly based on the MPI framework and multitree coding rules.In order to realize temporalspatial parallel computation in distributed hydrological model by widely used Pfafstetter,this paper adopted the OpenMP programming model,the layered simulation unit method and temporalspatial discretization method based on Pfafstetter coding were designed for the flow routing process of WEPL distributed hydrological model.The layered simulation unit method established the subwatershed topology table according to the Pfafstetter coding rules.The subbasin was stratified according to the order of the upstream to the downstream.Then,on this basis,the time factor was taken into account to restratify to realize temporalspatial discretization.Finally,the subwatershed simulation tasks which were independent of each other in time and space of the same layer were assigned to different threads to realize parallel computing.The Yellow River basin was selected as the research area and the parallel effect was tested with a different number of threads.The experimental results show that the flow routing parallelization improves the simulation efficiency of the model.Based on a CPU with 20 cores,when only the layered simulation unit algorithm is considered,the maximum acceleration ratio is 4.14.However,the acceleration ratio can reach to 8.17 when temporalspatial discretization is considered.
作者
秦泽宁
周祖昊
刘明堂
向东
刘佳嘉
QIN Zening;ZHOU Zuhao;LIU Mingtang;XIANG Dong;LIU Jiajia(North China University of Water Resources and Electric Power,Zhengzhou 450046,China;State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin,China Institute of Water Resources and Hydropower Research,Beijing 100038,China)
出处
《人民黄河》
CAS
北大核心
2020年第8期15-20,共6页
Yellow River
基金
国家重点研发计划项目(2016YFC0402405)
国家自然科学基金资助项目(91647109,51679257,51779270)
江西省水利科技重大项目(KT201501,KT201508)
云南省院士工作站建设专项(2015C013)
坪山河干流综合整治及水质提升工程专项课题(CSCEC-PSH-2017-03)
流域水循环模拟与调控国家重点实验室基金资助项目(SKL2018TS04)。
