Integrating geoinformation

Dr Nigel TroddCoventry University

an integrated geodatabase

1st Law of [a] GIS [salesman]*

2 sets of GI, 1 combination

20 data sets, 190 pairs, 1 million+ combinations

The more data linked, many more potential applications

* You get something for nothing by bringing together GI from different sources and using it in combination

BUT

Can raise complex problems of GI ownership

produces uncertainty where data collected to different standards

Data linking procedures may partially determine results

To integrate spatial data you need…

A compatible geodata file formatA specified data projection & coordinate systemA common spatial data modelTo understand the meaning of your geographical objects

Aim & objectives

To explain key processes & issues in integrating geospatial data

We will identify & exemplify techniques toapply a data projection & coordinate systemgeoregister dataconvert spatial data modelsspecify ontologies for geographical objects

Geodata stream

spatial referencing

Some (most?) georeferences are metric They define location using measures of

distance from fixed placese.g. distance from the Equator or from the Greenwich

Meridian

Others are based on ordering E.g. street addresses in most parts of the world

order houses along streets, sometimes even numbers one-side and odd numbers the other

Others are only nominal Placenames do not involve ordering or

measuring

Georeferencing for mapping and analysis

Placenames can be converted to coordinates using gazetteersStreet addresses must be converted to coordinates Using address-matching and geocoding

functions

Metric referencing systems can be converted between map projections Using mathematical transformations

Not surprisingly, these are standard GIS functions

Metric Geographic Coordinates: Lat / Long

(0,0)Equator

Prime Meridian

Latitude and Longitude

Longitude line (Meridian)N

S

W E

Range: 180ºW - 0º - 180ºE

Latitude line (Parallel)N

S

W E

Range: 90ºS - 0º - 90ºN(0ºN, 0ºE)

Equator, Prime Meridian

But… Earth to Globe to Map

Real world simplified as

a sphere

Sphere ‘flattened’ to

a map

Map projection system Datum Coordinate system

- direction

distance

area

shape

Metric georeferencing systems

More than one map projection?

John Savard’s homepage explains the basics: http://members.shaw.ca/quadibloc/maps/mapint.htm

Peter Dana’s notes for the Geographers Craft website (using map projections): http://www.colorado.edu/geography/gcraft/notes/mapproj/mapproj_f.html

Henry Bottomley does an impressive job at demonstrating the effects of different map projections: http://www.btinternet.com/~se16/js/mapproj.htm

a way of representing a spherical world using only a flat piece of paper.

equidistant conic projection

The Universal Transverse Mercator (UTM) Projection

A type of cylindrical projectionImplemented as an internationally standard coordinate system Initially devised as a military standard

Uses a system of 60 zones Maximum distortion is 0.04%

Transverse Mercator because the cylinder is wrapped around the Poles, not the Equator

Zones are each six degrees of longitude, numbered as shown at the top, from W to EUniversal Transverse Mercator

System

100 km

Easting 524

Northing 739

SP524739

OSGB National Grid

Georegistering spatial data

1. Identify common points on two data layers – at least one (the registered control) with a known projection, datum & coordinate system

Describe a (mathematical) relationship between the points

2. Apply that relationship to transform the entire unregistered data layer.

The unregistered layer inherits the georeferencing system of the registered layer

3. (for raster) Resample to a common grid

Ground control points (GCPs)

Rotate

Flip

Stretch

Translate

Transformations

transforming an imagerubber sheeting

resampling a grid

resampling a grid: NN

resampling an image

NN

Bilinear

Original

Geodata stream

But… why bother?

The real world is 3-dimensionalModern data capture technologies record XYZ coordinatesUse a 3D model for data storageVisualise geospatial data in 3DOnly project the data in 2D when you have to publish a flat paper map!

If only it was that simple…

..Archived data….Maps used as ‘base maps’……Images assume a ‘flat’ Earth

You will still need to understand projections & coordinate systems for many years

Converting spatial data models

Most, but not all, GIS can handle raster & vector data

Many GIS functions have been implemented for only one data model

Users of GIS frequently have to rasterise vectors & vectorise rasters

rasterisation

a complication?

Length or area measurements

Topology

Vectorisation: many more complications

Data interoperability

Low level: data file format

Medium level: data standards

High level: ontologies

interoperabilityWhat? Why?In a perfect world all data would come in one data format… but the world is not perfect!!Therefore, we adopt standard procedures & protocols to describe dataIf a GIS complies with these standards then it can read the data, …no matter what its’ internal format.

Partial interoperability

Making interoperability happen: formats & standards, procedures & protocols

de facto industry formats software vendors e.g. ESRI ‘shp’, Google Earth KML

National / international standardse.g. Ordnance Survey NTF

Web-based open standards… HTML, XML Open Geospatial Consortium (OGC) specifications

GML, WMS, WFS http://www.opengeospatial.org/

Semantic difficultiesMultiple application domainsMeans multiple languagesSimple questions… what is a road?Numerous answers…

Transport route Source of air pollution Drainage pathway Lump of concrete

How do you ‘translate’ data between languages?

Ontologiesan ontology is a specification of a conceptIn GI systems, we use concepts like land parcel, highway, lake, etc.In Architecture Engineering and Construction (AEC) systems we use concepts like building, room, garden, backyard, etc.

If our specification includes unique, unambiguous names for concepts, and descriptions of the meaning of those names

then we could develop a mapping between ontologies

AEC has produced well-specified BIMs e.g. IFCGIS has produced…. very little

Geo-ontologies

GIS has produced…. very little e.g. ESRI Geodatabase Schemas Data Models for GIS Users http://www.esri.com/software/arcgis/geodatabase/

about/data-models.html

Most of the time we (GISers) create implicit geo-ontologies on-the-fly… we know what we mean!

… what is a mountain?

and… perfect integration is not always possible

…the geography of data sources

summary

• We have tools to support spatial data integration

Geodata file format exchange

Georeferencing

Geodata model conversion

X Geospatial ontologies

• and (sometimes) we lack data worthy of integration