Morton-PELT: Efficient De-Noising Spatial-Change Point Detection for Large Scale Normalized Digital Elevation Model Data Set

In this paper, we propose a work flow for processing and analysing large-scale Digital Elevation Model (DEM) data that uses an infrastructure for change point detection methods based on a space filling curve representation of point patterns. The Digital Elevation Model usually maps to a three dimensional data set, and falls into the category of geo-spatial data. In a different realm, the field of statistical sequence analysis developed change point analysis methods, designed for time series, that are able to detect discontinuities in the data. Since the detection of discontinuities in geo-spatial data would be a worthwhile task, the purpose of this research is to find ways to combine geo-spatial processing with change-point detection in order to perform change-point detection on geo-spatial data. If successful, this endeavor would greatly extend the toolset available for geo-spatial data. In more detail, we propose to use a space filling curve method (Morton Order) to transform the high dimensional geospatial data into a time series. Once it is in sequential format, existing change point detection methods can be directly applied. This paper implements this suggested method and demonstrates its practical utility. To make it more effective, the HDBSCAN(Hierarchical Density-Based Spatial Clustering of Applications with Noise) algorithm is also applied in its capacity of de-noising algorithm to remove the noisy points in the intermediate phase. The resulting work flow is robust and has the capability to detect sea mountains among large-scale DEM data. Furthermore, it returns visualized reports to end users.