Spatial Databases: A soft Introduction

Lecturer: Malumbo Chaka Chipofya

Material for each class of the course can be found here on the spatial databases Classes page

Course Description

Spatial databases are a core component of any medium to large scale deployment of geographic information systems and services. A Spatially enabled Database Management System (SDBMS) provides capabilities for storage, replication, and analysis of geospatial data. “Spatial is special” goes the not-so-old adage. As such spatial data require special methods for storage, search, and query formulation. For example, the sorts of questions that one may want to pose to a spatial database are often significantly more complex as compared to the question one may ask, for instance, a sales records database. Understanding database management systems in general and spatial databases in particular are therefore essential for any Geoinformatics professional.

This course will first cover the basics of database management systems and introduce spatial database management systems with a focus on the relational database model. In parallel to the study of database concepts, the course develop particular skills in qualitative spatial reasoning that will be used in query formulation. The qualitative reasoning concepts are useful for implementing support for intelligent query capabilities and improving human interaction. For example instead of searching for a place using coordinates it is more natural for human beings to refer to the general area or territory where the place located. Finally, the course will explore, briefly but in a practical way, spatial extensions in alternative database models. This follows some of the trends that can be seen in practice. Initially, the relational database model earned unquestioned prominence in the database domain. However, with the changing technological landscape, other database models are beginning to find use in popular applications. These include array databases, document-oriented databases, and graph databases among others. All material will illustrated with suitable examples.

The course will take place in two parallel sessions. One will be focused on spatial database management systems and the other will focus on qualitative spatial knowledge representation and reasoning. The two fields are connected by distinct utility of qualitative spatial representation in spatial database query formulation and evaluation.

Note: It is not possible to take both this course and the course “Qualitative Spatial Knowledge Representation and Reasoning” On Wednesdays, the course will be offered in StudLab 126 (together with “Qualitative Spatial Knowledge Representation and Reasoning”).

Learning Objectives

The main learning objectives of this course are

to gain a general understanding of what Spatial Databases and Database Management Systems are and what they do
to learn some applications where Spatial Databases they are used and how (using spatial relational DB combo Postgres/Postgis)
explore a few contemporary database models

Main Topics and Tentative Schedule

Material for each class of the course can be found on the spatial databases Classes page

Prelude: Data and problem solving in science and applications
1. A small gallery of geospatial applications to be used throughout the course
  1. Land use change modelling and detection (place holder - will depend on data: what happened/is happening/will happen? how it happened/is happening/will happen?)
  2. Logistics and Transportation (place holder - will depend on data: specify the problem, determine the network, optimize the delivery cost)
  3. Social and business analysis: demographics (place holder - will depend on data: where is happening? as in where's whatsapp trending? next?)
2. Spatial Data Models: Field/Object vs. Raster/Vector dichotomy - different problems different representations.
3. An overview of Spatial Analysis methods and the required input data models and formats
4. Database models for GIS: A historical overview

(1 Session: 23.10.14)

The Relational Database Model
1. Recap and Overview of Database Systems
2. Relations, Tuples, Attributes, Keys and References (Relationships), and Relation Schemas
3. Relational Database Design
  1. Functional Dependencies and Normalization (1st Normal Form plus a note about other normal forms - up to BCNF if time allows)
  2. Problem analysis and data modelling
  3. Database schema design: E-R Model

(1 Session 3.a-b.: 30.10.14; 1 Session 3.c.i-ii.: 06.11.14)

Interacting with Relational Databases
1. Relational Query Language: Structured Query Language
2. Spatial Extensions of Relational Databases: Example - Postgresql and Postgis
3. Designing SQL queries - simple queries, query nesting, query chaining
4. Database access from applications (just a mention with examples)
  1. connecting to a DBMS from a GIS
  2. connecting to a DBMS using scripting/programming languages - Java, Python

(2 Sessions: 13.11.14, 20.11.14,)

Spatial Relational Database Systems:
1. Spatial data storage formats (WKB, WKT)
2. OGC Simple Features Standard
3. Spatial queries
  1. simple postgis queries
  2. complex queries
  3. built-in functions
  4. extensions - pgrouting

(1 Session 5.a-c.: 27.11.14; 2 Sessions 5.d. : 4.12.14, 11.12.14)

Applications: Terraview and Terralib: Prof. Dr. Gilberto Camara

(2 Sessions: 18.12.14, 8.01.15)

A sample of Nosql Databases
1. Array databases: brief introduction and example application (SciDB)
2. Document databases: brief introduction and example application (search locations of events: MongoDB)
3. Graph databases: brief introduction and example application (routing - compare with pgrouting or qualitative storage and query - directly store qcn and fetch subnetwork given in the query: Neo4J)

(3 Sessions: 15.01.15, 22.01.15, 29.01.15)

8. Conclusion: Summary of all lectures given and tips for future directions for research and further use cases. (1 Session: 5.02.15)