nD-PC

The main challenge in this proposal is to realize a distributed Open Point Cloud Map (OPCM) scalable infrastructure with High Performance/Throughput Computing (HPC/HTC) and enable interactive visualization using perspective views without data density shocks, continuous zoom-in and out and progressive data streaming between server and client.

Big geo-data requires good spatio-temporal data organization, including levels of detail that allow to zoom in from high-level overviews (complete countries/continents) to the smallest detail (as the curb stones of a sidewalk) and everything in between. The world’s largest point cloud data sets, despite their potential high value, are heavily underexploited due to the problematic data management, access, and limited software tools that are able to directly employ them.

This project aims to realize the paradigm shift from raster and vector representations to the new, highly efficient, nD-PointCloud representation with deep integration of space, time, scale/continuous Levels of Importance (cLoI) dimensions.

The current state of the art is organizing large point clouds in discrete levels of data pyramids.
We propose to generate random cLoI values according to ideal distribution function for continuous levels, add this as dimension, and use high-resolution nD space-filling curves, which have shown to be beneficial in organising multi-dimensional spaces and for task decomposition in, but which have not been applied yet to
these volumes and higher dimensions (>3D).

To assess and validate the nD-PointCloud/OPCM, it will be tested with the Deltares applications in the domain of water management: point cloud (=reality) enables change detection (of dunes, bathymetry, sediments using AHN, ICESat-2/GEDI temporal point clouds) more efficient based on cLoI, flood models with better simulation results, more clear visualizations, etc.