Spatiotemporal Data Management, Prediction, and Visualization for Coastal Digital Twins
Project Description
Hong Kong’s coastal areas generate and accumulate large-scale spatiotemporal data from multiple sources, including historical coastal records, wave and tidal information, weather conditions, sensor observations, simulation outputs, and spatial map data. These data reflect the temporal evolution and spatial distribution of coastal conditions, and can be used to understand historical trends, monitor current risks, and support future coastal hazard prediction.This project aims to develop a prototype for managing, predicting, and visualizing spatiotemporal data in a coastal digital twin system with four core tasks. First, students will organize and preprocess different types of spatiotemporal data. Second, they will design spatiotemporal data management structures, such as indexing, querying, and retrieval methods for different locations and time periods. Third, they will explore deep learning models for predicting coastal risk indicators, such as flooding level, inundation area, or infrastructure condition. Fourth, they will build an interactive visualization interface to display spatial distributions, temporal trends, and prediction results.Through this project, students will gain hands-on experience in spatiotemporal data management, predictive modeling, digital twin system design, and visual analytics.
Supervisor
ZHOU, Xiaofang
Quota
2
Course type
UROP1000
UROP1100
UROP2100
UROP3100
UROP3200
UROP4100
Applicant's Roles
The applicant is expected to:
1.Clean and organize spatiotemporal datasets from multiple sources, such as weather records, sensor data, map grids, and physical simulation outputs.
2.Build a data management module to support storage, indexing, and query of spatial and temporal data.
3.Implement deep learning prediction models for spatiotemporal forecasting tasks, such as risk level prediction or condition index estimation.
4.Develop visualization components to show historical data, real-time data, and predicted results on maps, charts, or dashboards.
5.Document the system design, implementation process, and experimental results clearly.
1.Clean and organize spatiotemporal datasets from multiple sources, such as weather records, sensor data, map grids, and physical simulation outputs.
2.Build a data management module to support storage, indexing, and query of spatial and temporal data.
3.Implement deep learning prediction models for spatiotemporal forecasting tasks, such as risk level prediction or condition index estimation.
4.Develop visualization components to show historical data, real-time data, and predicted results on maps, charts, or dashboards.
5.Document the system design, implementation process, and experimental results clearly.
Applicant's Learning Objectives
By completing this project, the applicant will learn how to:
1.Manage large-scale spatiotemporal data using index structures to support efficient storage, query, and retrieval.
2.Apply deep learning models, such as graph neural networks and attention mechanisms, to spatiotemporal prediction problems.
3.Build visual analytics tools to present spatial patterns, temporal changes, and prediction results.
4.Connect data management, prediction, and visualization into a digital twin prototype.
1.Manage large-scale spatiotemporal data using index structures to support efficient storage, query, and retrieval.
2.Apply deep learning models, such as graph neural networks and attention mechanisms, to spatiotemporal prediction problems.
3.Build visual analytics tools to present spatial patterns, temporal changes, and prediction results.
4.Connect data management, prediction, and visualization into a digital twin prototype.
Complexity of the project
Moderate