Volume IV-4/W2
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-4/W2, 7-13, 2017
https://doi.org/10.5194/isprs-annals-IV-4-W2-7-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 4.0 License.
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-4/W2, 7-13, 2017
https://doi.org/10.5194/isprs-annals-IV-4-W2-7-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 4.0 License.

  19 Oct 2017

19 Oct 2017

COMPARISON OF SPATIOTEMPORAL MAPPING TECHNIQUES FOR ENORMOUS ETL AND EXPLOITATION PATTERNS

R. Deiotte1 and R. La Valley1,2 R. Deiotte and R. La Valley
  • 1ISSAC Corp, 6760 Corporate Drive, Ste. 240, Colorado Springs, CO 80922, USA
  • 2OGSystems, Inc., 14291 Park Meadow Drive #100, Chantilly, VA 20151, USA

Keywords: Spatiotemporal Encoding, Space-Filling Curves, Manifold Coverings, Computational Cost, Space-timeIndexing, Space-time Encoding Efficiency, Space-time Encoding Utility

Abstract. The need to extract, transform, and exploit enormous volumes of spatiotemporal data has exploded with the rise of social media, advanced military sensors, wearables, automotive tracking, etc. However, current methods of spatiotemporal encoding and exploitation simultaneously limit the use of that information and increase computing complexity. Current spatiotemporal encoding methods from Niemeyer and Usher rely on a Z-order space filling curve, a relative of Peano’s 1890 space filling curve, for spatial hashing and interleaving temporal hashes to generate a spatiotemporal encoding. However, there exist other space-filling curves, and that provide different manifold coverings that could promote better hashing techniques for spatial data and have the potential to map spatiotemporal data without interleaving. The concatenation of Niemeyer’s and Usher’s techniques provide a highly efficient space-time index. However, other methods have advantages and disadvantages regarding computational cost, efficiency, and utility. This paper explores the several methods using a range of sizes of data sets from 1K to 10M observations and provides a comparison of the methods.