Spatio-temporal analysis
Spatio-temporal data is graphic and image data that contains both time and space information. It captures multidimensional attributes of objects — locations, shapes, changes, and size distribution.
The traditional dichotomy classifies spatio-temporal data into vector data and raster data. It has been used in industries built around GPS, geographic information systems (GIS), and Remote Sensing (RS). As IoT and smart terminals have spread across industries, a new category — perceptual spatio-temporal data — has emerged. Its role has expanded from location-based service (LBS) to multidimensional joint analysis and spatio-temporal pattern mining.
Spatio-temporal data types
| Type | Examples |
|---|---|
| Vector data | Digital maps and digital elevation model (DEM) data |
| Raster data | Remote sensing imagery and panoramic images |
| Perceptual data | Location data from smart terminals and laser point cloud data |
Spatio-temporal data models
AnalyticDB for PostgreSQL supports three data models to cover the full range of spatio-temporal workloads:
-
Geometry model: Complies with OpenGIS standards and supports 2D (x, y), 3D (x, y, z), and 4D (x, y, z, m) geometries.
-
Raster model: Represents data as a matrix of cells or pixels organized into rows and columns. Each cell stores a value, such as temperature. Raster data includes digital aerial photographs, satellite images, digital images, and scanned maps.
-
Trajectory model: Records the location information of a moving feature, such as a vehicle or a person.
Use cases
Business intelligence
-
Select store locations by analyzing foot traffic patterns and local demographics
-
Track and optimize delivery routes based on courier trajectories
-
Push targeted messages or ads based on a user's LBS context
-
Identify spatial correlations between consumers and their active areas
Transportation and logistics
-
Provide real-time location services for public transportation, Internet plus transportation, and intelligent logistics
-
Analyze traffic flow on roads and intersections, predict flows of people and vehicles, and estimate time of arrival (ETA)
-
Aggregate vehicle origin and destination data for demand analysis
-
Optimize vehicle monitoring, scheduling, and dispatch
-
Match and analyze similar vehicle trajectories
-
Visualize transport capacity and real-time movement heat maps
-
Manage, monitor, and trigger alerts for geofences
Public security
-
Track special groups and manage child custody records
-
Track special vehicles and identify abnormal vehicles
-
Push safety alerts and danger warnings
-
Monitor and manage travel records with health QR codes
Autonomous driving
-
Store, retrieve, analyze, and mine laser point cloud data for spatio-temporal patterns
-
Match trajectories with high precision and plan local paths
-
Produce and manage high-precision maps
Spatio-temporal engines
AnalyticDB for PostgreSQL provides two extension modules for storing, indexing, querying, analyzing, and computing spatial and spatio-temporal data: PostGIS and GanosBase.
Engine comparison
| Capability | PostGIS | GanosBase |
|---|---|---|
| Geometry model | Full OpenGIS compliance; rich objects, indexes, functions, and operators | Fully compatible with PostGIS geometry models; existing applications migrate without changes |
| Raster model | Standard raster functions and operators | Supports Object Storage Service (OSS) as a data source; richer management, analysis, and computation functions |
| Trajectory model | Not supported | Full support with dedicated data types, functions, and stored procedures |
Benefits
Easy to use: Switch between PostGIS and GanosBase without changing your application. Migrate data from a standalone PostgreSQL database to an AnalyticDB for PostgreSQL instance.
Cost-effective: The massively parallel processing (MPP) architecture lets a single AnalyticDB for PostgreSQL instance handle more spatio-temporal data than a standalone PostgreSQL database based on the same performance metrics and computing resources. OSS integration enables tiered storage of hot and cold data, which reduces storage costs for raster data and laser point cloud data without sacrificing query performance.