A Parallel JPWSPC Algorithm for Hydrodynamic Simulation of River Network

摘要

To diminish the hydrodynamic simulation time of complex river network, a parallel computation algorithm based on JPWSPC (joint point water stage prediction and correction) is proposed. In the iteration of solving procedure of PJPWSPC algorithm, computation of each reach can be performed in slave process, and water level prediction and correction of joint point is executed in the master process with the MPI API employed in the communication between master process and slave process. This parallel framework for river network hydrodynamic simulation can be harnessed in the loop network as well as the tributary network. The relation between speed-up ratio and average segment count of river network calculated in single process is Preliminary analyzed and the grouping method of branches for task scheduling is constructed by analyzing the computing process of each reach and communication between master process and slave process. Two hydrodynamic scheme simulation of natural river network is carried out. The estimation method can be employed in the grouping of branched in the hydrodynamic simulation of river network and the multiobjective computing resources allocation in the massive river network hydrodynamic simulation in the cloud computing based framework (HydroMP).