Uncertainty Quantification and Visualization: Geo-Spatially Registered Terrains and Mobile Targets Suresh Lodha Computer Science, University of California, Santa Cruz • Common consistent representation of multiple views of geo-spatially registered terrains • Low uncertainty compression algorithms preserving line features within terrains • Visualization of uncertainty of GPS-tracked mobile targets • Integration of mobile targets and terrains with geographic databases for decision-making Accomplishments - I • Development of GIS infrastructure for context-aware situational visualization • Development of GPS infrastructure for mobile visualization • Work on consistency and uncertainty issues in mobile situational (GIS-GPS) visualization Accomplishments - II • Modeling and quantifying uncertainty – Probability-based uncertainty (collaboration with Pramod Varshney, Syracuse University) – Spatio-temporal GPS uncertainty – Low uncertainty line preserving compression algorithms for terrains (extension from point preserving algorithms from previous year) Accomplishments - III • Integration of data and uncertainty within a global geospatial system (collaboration with Georgia Tech) • Application to – Geospatial visualization – General Aviation • Continuing work on – Multimodal interaction (speech) – Database querying – Wireless networks for communicating and visualizing data and information with associated uncertainty GIS Infrastructure • Aerial Imagery (DOQQs) • Elevation Data – Digital Elevation Models (DEMs) – LIDAR Data • Architectural Drawings • Street Maps • Schematic Diagrams GIS Images: Aerial Imagery and LIDAR GIS Images: DEM and AutoCAD GPS Infrastructure • Ashtech Z-12/G-12 Sensors – Standalone (1 meter) / Differential (1 cm) – Velocity (.1 knots) – L1/L2 frequency (ionospheric delay correction) – RTK/RTCM messages – 10 Hz update rate GPS Receiver Equipment Consistency and Uncertainty in Mobile Situational Visualization • Disparate data sources • Different data formats • Different coordinate systems • Different resolutions/ sampling/ sizes • Different accuracy • Different time stamps • Communication time lags Common Consistent Representation: Multiple Views of Terrains Aerial Imagery AutoCAD Drawing LiDAR Data • Common Coordinate System • Geo-Spatial Registration • Accuracy Common Consistent Representation: Multiple Views of Terrains Modeling and Visualizing Uncertainty • Probability-based uncertain particle movement • GPS-based spatio-temporal uncertainty in particle movement • Low uncertainty compression algorithms preserving line features within terrains Algorithmic Computation • Compute the probability of target at a point x after time t – Probability at an initial location (p) – Probability of movement along a direction (d) – Probability of speed (s) – Final probability = p * d * s Computation of Probabilistic Locational Uncertainty Uncertain Probabilistic Shapes GPS Sources of Uncertainty • Measurement Errors – Satellite clock drift, receiver clock drift, satellite location error, atmospheric effects, multipath effect, selective availability • GPS Availability Issues • GPS Integrity Anomalies and Vulnerability Parameters • Mode – Standalone / Differential • Environment – Urban / Foliage • Movement – Stationary – Moving (Constant Velocity, Random) Modeling: Static Data • Number of accessible/used satellites – Urban higher than foliage – Standalone same as differential • SNR (Signal to Noise Ratio) values – Urban higher than foliage – Standalone same as differential • DOP (Dilution of Precision) values – Urban smaller than foliage – Standalone smaller than differential Satellite Availability Dilution of Precision • Satellite Geometry and Orientation Good satellite geometry Poor satellite geometry SNR Modeling Observations and Analysis: Constant Velocity Data Visualization Visualization Terrain Uncertainty • Point feature preserving compression algorithms (last year MURI) • Line feature preserving compression algorithms – EMD (earth movers distance) concept extended to line features – More efficient local algorithm – Line preservation (coastlines etc.) Topology Degradation Metric • EMD (Rubner et al. ‘98, Batra et al. ‘98, ’99, Lodha et al. 2000) – amount of work required to move one set of lines to another (similarity) – Variables • # features • Location of features • Feature Attributes – Length, Orientation Line EMD Error Line Preserving Compression Unconstrained Coastline preserving Line Preserving Compression Original Unconstrained Coastline preserving Hierarchical Line Simplification Integration of Data and Uncertainty within VGIS • Hierarchical zooming from the globe into the UCSC Campus (1/2 foot resolution imagery) • Real-time visualization of GPS-tracked objects and associated uncertainty within VGIS Hierarchical Zooming into UCSC Campus Real-Time Mobile Uncertainty Visualization within VGIS Uncertainty Quantification, Visualization and Communication: Continuing Work • Heterogeneous Geo-Spatial Uncertainty • Mobile Temporal Uncertainty • Multi-Sensor Data Fusion (Images, LIDAR) • Multi-modal Interaction (speech) • Database Querying • Wireless Networks Collaborations - I • Worked with Pramod Varshney on probabilistic uncertain particle movement (1 joint paper and 1 jointly supervised student); continuing to collaborate on uncertainty with mobility constraints • Worked with Bill Ribarsky on integration of uncertainty within VGIS (1 joint paper and 1 jointly supervised student); continuing to collaborate on uncertainty in mobile situational visualization Collaborations - II • Worked with Ulrich Neumann on development of GPS infrastructure • Worked with Avideh Zakhor on acquistion of LIDAR data • continuing to collaborate on uncertainty in GPS, LIDAR and image data Major Accomplishments • Computation and visualization of uncertainty for terrains while preserving point and line features of terrains, • Computation and visualization of uncertain mobile GPS-tracked targets embedded within a GIS Environment, and • Embedding and visualization of uncertainty within the VGIS.