5d-odetlap: a Novel Five-dimensional Compression Method on Time-varying Multivariable Geospatial Dataset

A five dimensional (5D) geospatial dataset consists of several multivariable 4D datasets, which are sequences of time-varying volumetric 3D geographical datasets. These datasets are typically very large in size and demand a great amount of resources for storage and transmission. In this paper, we present a lossy compression technique for 5D geospatial data as a whole, instead of applying 3D compression method on each 3D slice of the 5D dataset. Our lossy compression technique efficiently exploits spatial and temporal similarities between 2D data slices and 3D volumes in 4D oceanographic datasets. 5D-ODETLAP, which is an extension of, but essentially different from, the Laplacian partial differential equation, solves a sparse overdetermined system of equations to compute data at each point in (x,y,z,t,v) space from the data given at a representative set of points. 5D-ODETLAP is not restricted to certain types of datasets. For different datasets, it has the flexibility to approximate each one according to their respective data distributions by using suitable parameters. The final approximation is further compressed using Run Length Encoding. We use different datasets and metrics to test 5D-ODETLAP, and performance evaluations have shown that the proposed compression technique outperforms current 3D-SPIHT method on our selected datasets, from the World Ocean Atlas 2005. Having about the same mean percentage error, 5D-ODETLAP’s compression result produces much smaller maximum error than 3D-SPIHT. A user-defined mean or maximum error can be set to obtain desired compression in the proposed method, while not in 3D-SPIHT.