A Workflow-Serialized Parallel Spatial IDW Interpolation on Windows HPC
In order to implement the spatial Inverse Distance Weighting (IDW) interpolation in parallel, we use the quadtree approach to decompose the spatial domain and execute the IDW interpolation on these quads through the parametric tasks on Windows HPC platform. The whole implementation includes the executables of QuadDC, IDW, output gathering and data visualization which should be run serially. Compared with the traditional method to integrate each executable manually, the Job Manager of Windows HPC utilizes multi-task job to serialize all the executables as a workflow. The experiments test two cases of IDW0 and IDW6 with different domain clustering regions, and show that our implementation improves the functionality integration and efficiency of the spatial IDW interpolation.
Y. J. Xia et al., "A Workflow-Serialized Parallel Spatial IDW Interpolation on Windows HPC", Applied Mechanics and Materials, Vols. 20-23, pp. 370-375, 2010