At the OGC member meeting in Delft, Netherlands, my team from Gaia3D and I shared our experiences and the challenges we faced while visualizing large spatio-temporal datasets in digital twins. In conclusion, we discussed the necessity for a new standard, referred to as 'Voxel Tiles,' for visualizing spatio-temporal data.
Do we need a new standard for visualizing the invisible?
1. Do we need a new
standard for visualizing
the invisible?
The 128th OGC Member Meeting
Sanghee Shin(shshin@gaia3d.com)
Hakjoon Kim(hjkim@gaia3d.com)
25th March 2024
Meeting Sponsors
2. Agenda
• Background: Blending Spatio-Temporal Data with Digital Twins
• The Hurdles: The Big Data Dilemma and Standards
• Tackling the Problems: Smart Fixes for Data Overload
• Lessons Learned: Map Tiles, Vector Tiles, 3D Tiles, and Voxel Tiles(?)
3. Background
• We’ve been developing web based digital twins for decision makers and citizens.
• To support informed decision-making and to provide comprehensive understanding of
real-world problems, it is essential to reconstruct the real-world in digital twins from
diverse data sources in an intuitive manner with terrain, buildings, spatio-temporal
data, and others.
• We’ve developed digital twins on top of OGC standard blocks such as WMS, WMTS,
WFS, and 3D Tiles.
• When we tried to meld spatio-temporal data with digital twins, we realized there are
problems!
4. Spatio-Temporal Data
Chemical Diffusion Prediction
Noise Propagation Simulation Drainage and Flooding Simulation Before and After Construction
Wind Flow Simulation Before and After Construction
5. The Hurdles
• There are challenges with data volume in aerodynamics modeling or computational
fluid dynamics(CFD), e.g., wind field, air pollution, odors spreading, or chemical
diffusion simulations.
• The size of these data are very huge by 2 reasons.
1. Multi-layered(3D gridded) data covering very large areas, from hundreds to thousands
of square kilometers.
2. Datasets consisting of numerous temporal snapshots.
6. In addition….
• Our visualization process required the use of all snapshots in their entirety to
accurately represent changes over time.
• Users expressed a desire to track the temporal evolution of the data.
• Decision makers and citizens preferred animation as the optimal method to
illustrate these dynamic changes.
• And hard to find a standardized method to visualize these spatio-temporal
data
7. Real Cases
<Chemical diffusion prediction during a chemical leak incident> <NO2 diffusion simulation for EIA of a landfill construction>
Source data spec:
- Coverage area : 15km x 15km(cell size 100m x 100m)
- Coverage height : 0~3km (8 layers)
- Time duration : 3 hours (1 min interval, 180 timestamps)
- Data size : 1.3GB (ASCII format)
Source data spec:
- Coverage area : 30km x 30km(cell size 200m x 200m)
- Coverage height : 0~200m (7 layers)
- Time duration : 1 year(1 hour interval, 8,760 timestamps)
- Data size : 114GB (ASCII format)
8. Tackling the Problems
• Implementing value coding to reduce and compress the data size.
• Converted the original data to PNG format, achieving a reduction to
approximately 1.5% of the original size.
9. And,
• Spatial and temporal interpolation were performed on the client side to
minimize discontinuities between vertical layers and temporal snapshots.
spatial interpolation temporal interpolation
10. And,
• Each frame of the animation was created using volume rendering.
concept of volume rendering
4 steps of volume rendering
(1. ray casting, 2. sampling, 3.shading, 4. compositing)
13. Lessons Learned
• We need lightweight data handling strategies for effective visualization of large
size spatio-temporal data.
• For dynamic data that evolves over time, considering a streaming approach is
imperative to ensure efficient data delivery.(In the case of the NO2 diffusion simulation,
retrieving the complete set of PNGs took between 12 to 18 minutes, which we deemed unsatisfactory. Consequently,
we are now in the process of developing a streaming service to efficiently deliver this data.)
• We’ve used Map Tiles, Vector Tiles, 3D Tiles to represent the real-world
features and phenomena. So, how about having a new standard for visualizing
the spatio-temporal data, maybe named Voxel Tiles?
Acknowledgment: This work was supported by Korea Environment Industry & Technology Institute(KEITI) through 'Advanced Technology Development Project for Predicting
and Preventing Chemical Accidents' program and 'The Decision Support System Development Project for Environmental Impact Assessment' program, funded by Korea Ministry
of Environment(MOE)(grant number: RS-2022-KE002176 and 2020002990005).