Spatial Data Modelling Transformation of OS Master Map

Spatial Data Modelling: Transformation of OS Master. Map to 1: 250 000 scale database EEO- Hutton Club Seminar Series 23 Nov 2007 Omair Zubair Chaudhry Institute of Geography University of Edinburgh O. Chaudhry@sms. ed. ac. uk EEO-Hutton Club – Nov 2007 1 of 48

Structure Outline: – Introduction: Scale and Geographic Phenomena – Generalisation: Model & Cartographic – Methodology: Database Transformation – Results, Evaluation and Implementation – Conclusions EEO-Hutton Club – Nov 2007 2 of 48

Introduction: Scale & Pattern EEO-Hutton Club – Nov 2007 3 of 48

Introduction: Need of Generalisation EEO-Hutton Club – Nov 2007 4 of 48

Art and Science of Cartography EEO-Hutton Club – Nov 2007 5 of 48

Current Research The Challenge: Phenomena at 1: 250, 000 from OS Master. Map 1: 250, 000 Ordnance Survey Master. Map (Sourced from 1: 1250 & 1: 10, 000 mapping) Objectives • To demonstrate the utility of the transformed database • To demonstrate importance of relationships in terms of database transformation • Handling large volumes of data EEO-Hutton Club – Nov 2007 6 of 48

Critical Factors of Model Generalisation EEO-Hutton Club – Nov 2007 7 of 48

Target Data Model • Classes of Target Data Model EEO-Hutton Club – Nov 2007 8 of 48

Source Data Model • Source Database: OS Master. Map Topography Layer EEO-Hutton Club – Nov 2007 9 of 48

Objects and Relationship • Semantic Relationships: Taxonomic Relationships ‘is-a’ EEO-Hutton Club – Nov 2007 10 of 48

Objects and Relationship • Partonomic Relationships ‘part of’ EEO-Hutton Club – Nov 2007 11 of 48

Conditions and Constraints • Model generalisation is a application and scale dependent process. • Effect the parameter settings of the output classes • Topological Consistency • Geometrically (space exhaustive) and thematically partitioned. EEO-Hutton Club – Nov 2007 12 of 48

Operation • Aggregation EEO-Hutton Club – Nov 2007 13 of 48

Methodology EEO-Hutton Club – Nov 2007 14 of 48

Container Boundaries • Container Boundaries for composite Classes of Target Data Model: – Settlement – Forest – Hills and Ranges EEO-Hutton Club – Nov 2007 15 of 48

Buildings to Settlement EEO-Hutton Club – Nov 2007 16 of 48

Settlement Container Boundary EEO-Hutton Club – Nov 2007 17 of 48

Settlement Container Boundary EEO-Hutton Club – Nov 2007 18 of 48

Settlement Container Boundary • Modelling Citiness 50 closest buildings EEO-Hutton Club – Nov 2007 19 of 48

Settlement Container Boundary Provided EEO-Hutton Club – Nov 2007 20 of 48

Settlement Container Boundary EEO-Hutton Club – Nov 2007 21 of 48

Seeing Wood from Trees EEO-Hutton Club – Nov 2007 22 of 48

Forest Container Boundary EEO-Hutton Club – Nov 2007 23 of 48

Forest Container Boundary • Area Patch Methodology • (Müller and Wang, 1992) Eq 1 ci = ai/(pi/4π) If >FT then Expand (eq 1) Else Check density If density >FT Expand (eq 1) Else Contract (eq 1) EEO-Hutton Club – Nov 2007 24 of 48

Forest Container Boundary EEO-Hutton Club – Nov 2007 25 of 48

Mountains from Mole Hills EEO-Hutton Club – Nov 2007 26 of 48

Hill / Range Container Boundary EEO-Hutton Club – Nov 2007 27 of 48

Absolute Height Above 150 m EEO-Hutton Club – Nov 2007 28 of 48

Hill / Range Container Boundary • Prominence of a Summit EEO-Hutton Club – Nov 2007 29 of 48

Hill / Range Container Boundary • Morphology • (Wood 1996) EEO-Hutton Club – Nov 2007 30 of 48

Hill / Range Container Boundary EEO-Hutton Club – Nov 2007 31 of 48

Hill / Range Container Boundary EEO-Hutton Club – Nov 2007 32 of 48

Hill / Range Container Boundary EEO-Hutton Club – Nov 2007 33 of 48

Database Enrichment EEO-Hutton Club – Nov 2007 34 of 48

Database Enrichment • Partonomic Relationships EEO-Hutton Club – Nov 2007 35 of 48

Selection and Aggregation EEO-Hutton Club – Nov 2007 36 of 48

Results and Evaluation EEO-Hutton Club – Nov 2007 37 of 48

Results EEO-Hutton Club – Nov 2007 38 of 48

Results EEO-Hutton Club – Nov 2007 39 of 48

Utility • • Spatial Analysis SELECT a. Geometry FROM Source_database a, Partonomy_table b, Name_database c WHERE a. object_id=b. objet_id and a. descgroup=’Building’ and b. composite_object_id=c. composite_object_id and c. object_name=’East Calder’; EEO-Hutton Club – Nov 2007 40 of 48

Utility Higher Level of Abstraction EEO-Hutton Club – Nov 2007 41 of 48

Utility EEO-Hutton Club – Nov 2007 42 of 48

Implementation • All algorithms implemented in Java • Oracle Spatial 10 g used for data storage and spatial functions: Nearest Neighbours, Aggregation, Distance, Topological Relationships, Area intersection. • R-Tree Spatial Indexing • Land. Serf (by Jo Wood) java libraries integrated with our code. • Geomedia and Arc. GIS used for visualisation EEO-Hutton Club – Nov 2007 43 of 48

Findings • • • Creation of higher order phenomena Automated Aggregation Intelligent Links Meaningful Queries Results of Model generalisation input to Cartographic generalisation EEO-Hutton Club – Nov 2007 44 of 48

Future Work • Expanding the number of classes in the target data model • Evaluating Techniques • Developing Generalisation Services • Ontological Driven Generalisation • Developing dataset partitioning Techniques EEO-Hutton Club – Nov 2007 45 of 48

Conclusions • “Generalisation is more than just mimicry of human cartographer, it is about modelling geographic space” • Importance of meaning and scale at different levels of detail • Modelling not only needed to create visual output but also for spatial analysis, data mining and meaningful interrogation of spatial data. EEO-Hutton Club – Nov 2007 46 of 48

Acknowledgements • William Mackaness, Nick Hulton for their constant support. • Ordnance Survey and the University of Edinburgh for funding this research. • Institute of Geography for all the required facilities and a great working environment. Special thanks to Steve Dowers for help on Oracle Spatial. • My family and friends for their love and support. • And Specially grateful to…. . EEO-Hutton Club – Nov 2007 47 of 48