Multivariate Probabilistic Range Queries for Scalable Interactive 3D Visualization

Abstract

Fig. 1. Probabilistic queries. For the meteorological data set shown on the left (N > 1G cells), on the right we show three consecutive query results. (1) Without filtering the data; (2) Filtering for temperature in [−24, −8] and relative humidity in [15, 20]; 1% false positives (shown in red). (3) A d = 5-dimensional query, in addition filtering height in [0, 4000] and two vector field components. (Inset) We improve both the query performance and the false positive rate via a hierarchical early-out strategy using a novel concept of supercells.