A two-dimensional interpolation function for irregularly-spaced data
ACM '68 Proceedings of the 1968 23rd ACM national conference
A quadtree approach to domain decomposition for spatial interpolation in grid computing environments
Parallel Computing - Special issue: High performance computing with geographical data
Guest Editors' Introduction: Parallelism on the Desktop
IEEE Software
| Hi-index | 0.00 |
Spatial interpolation is a widely used GIS function for estimating values at locations where observed values are not available or adequate. One popular method for spatial interpolation is inverse distance weighted, which calculates estimated values based on a weighted sum of the values of a number of nearest neighbors that have observed values. This research focuses on solving a large-scale interpolation problem with a global coverage based on the inverse distance weighted method. Specifically, spherical distance is calculated instead of normal Euclidean distance commonly used in GIS software, which is necessary to find correct neighbors in the regions along the 180° longitude and in the polar areas. The computation of the global-scale interpolation based on spherical distance is intensive especially for achieving high-resolution results. This paper introduces how to accelerate such computation by exploiting massive parallelism provided by Graphic Processing Units (GPUs) with significant improvement of computational performance reported.