摘要
Numerical modeling is of crucial importance in understanding the behavior of regional groundwater system. However, the demand on modeling capability is intensive when performing high-resolution simulation over long time span. This paper presents the application of a parallel program to speed up the detailed modeling of the groundwater flow system in the North China Plain. The parallel program is implemented by rebuilding the well-known MODFLOW program on our parallelcomputing framework, which is achieved by designing patch-based parallel data structures and algorithms but maintaining the compute flow and functionalities of MODFLOW. The detailed model with more than one million grids and a decade of time has been solved. The parallel simulation results were examined against the field observed data and these two data are generally in good agreement. For the comparison on solution time, the parallel program running on 32 cores is 6 times faster than the fastest MICCG-based MODFLOW program and 11 times faster than the GMG-based MODFLOW program. Therefore, remarkable computational time can be saved when using the parallel program, which facilitates the rapid modeling and prediction of the groundwater flow system in the North China Plain.
Numerical modeling is of crucial importance in understanding the behavior of regional groundwater system. However, the demand on modeling capability is intensive when performing high-resolution simulation over long time span. This paper presents the application of a parallel program to speed up the detailed modeling of the groundwater flow system in the North China Plain. The parallel program is implemented by rebuilding the well-known MODFLOW program on our parallelcomputing framework, which is achieved by designing patch-based parallel data structures and algorithms but maintaining the compute flow and functionalities of MODFLOW. The detailed model with more than one million grids and a decade of time has been solved. The parallel simulation results were examined against the field observed data and these two data are generally in good agreement. For the comparison on solution time, the parallel program running on 32 cores is 6 times faster than the fastest MICCG-based MODFLOW program and 11 times faster than the GMG-based MODFLOW program. Therefore, remarkable computational time can be saved when using the parallel program, which facilitates the rapid modeling and prediction of the groundwater flow system in the North China Plain.
基金
supported by the National Basic Research Program (973 Program) of China (Nos. 2010CB428804 and 2011CB309702)
the Key Projects of National Natural Science Foundation of China (No. 61033009)