Presentation from the AGI Geocommunity 2008 conference, September 2008.

1. If we have Google, do weIf we have Google, do wegg
need the OGC?need the OGC?
J M lJeremy Morley
jmorley@cege.ucl.ac.ukj y@ g
Dept. Civil, Environmental & GeomaticDept. Civil, Environmental & Geomatic
Engineering, UCL

2. G lGoal
• To explore the places of Google and the
Open Geospatial Consortium (OGC) inOpen Geospatial Consortium (OGC) in
geospatial information systems, especially
web mapping
(And by extension Microsoft Virtual Earth• (And by extension, Microsoft Virtual Earth,
Multimap, OS OpenSpace, etc.)p p p )

3. O tliOutline
• [ Web mapping under the hood ]
G l ' ti l t• Google's geospatial systems
• Introduction to the OGCIntroduction to the OGC
• Google Maps versus OGC Web Map
Services
B d b i• Beyond web mapping

4. G l ' ti l tGoogle's geospatial systems
• Google Maps
G l E th• Google Earth
• KMLKML
• SketchUp
– 3D design sketching software
Recent addition: ArcGIS connector– Recent addition: ArcGIS connector
– Collada (XML 3D exchange format)( g )

5. Google Maps

6. Particular features of Google MapsParticular features of Google Maps
• AJAX programming => interactivityAJAX programming > interactivity
• Good UI (simple, uncluttered)
• Data!
Fi d l l i• Fixed zoom levels, raster mapping
(<=> server farms cartography)(< > server farms, cartography)
• Well integrated functionality,
– e.g. route planning with vector overlay
Published public free (to some) access• Published, public, free (to some) access,
Javascript API

7. Public API -> Mashups
Wikimapia
www.wikimapia.org
Public API Mashups

9. Google Earthg

10. Scientific data in Google EarthScientific data in Google Earth
The Visual Exploration of Insurance Data in Google Earthp g
Aidan Slingsby, Jason Dykes, Jo Wood, Matthew Foote, Mike Blom
Proceedings of GISRUK 2008

11. KML Now an OGC standard
KML
• Used by Google to describe geographic features &
services, particularly for Google Maps & Earth, p y g p
• An XML application to:
– Specify icons and labels to identify locations on the planet surface
– Create different camera positions to define unique views for each of
your features
– Use image overlays attached to the ground or screenUse image overlays attached to the ground or screen
– Define styles to specify feature appearance
– Write HTML descriptions of features, including hyperlinks and
embedded imagesembedded images
– Use folders for hierarchical grouping of features
– Dynamically fetch and update KML files from remote or local
network locations
– Fetch KML data based on changes in the 3D viewer
– Display COLLADA textured 3D objects (SketchUp)Display COLLADA textured 3D objects (SketchUp)
• http://earth.google.com/kml/whatiskml.html

12. Introduction to the OGCIntroduction to the OGC
• Open Geospatial ConsortiumOpen Geospatial Consortium
• Consortium of vendors, users, academics involved
fin geospatial information technologies
• Includes the major GIS vendors, the largestIncludes the major GIS vendors, the largest
players in the IT world, small companies, national
agencies universitiesagencies, universities
– E.g. Mapinfo, ESRI, Intergraph, Cadcorp,..
Google Microsoft Oracle– Google, Microsoft, Oracle
– NASA, ESA, USGS, OS, BGS, NERC
– UCL Nottingham EDINA City University– UCL, Nottingham, EDINA, City University

13. OGC Obj tiOGC Objectives
• Promote the use of “interoperable geoprocessing”
throughout the Information Technology marketplacethroughout the Information Technology marketplace.
• Synchronize “geoprocessing technology” with commercial
“Information Technology standards”Information Technology standards
• Arrange cooperation of “GI product suppliers” and “GI
” t d l i t bl ft i t fusers” to develop interoperable software interfaces
• Involve the entire community in the “interoperability
”process”
• Provide an “industry forum” for “partnerships” and
cooperative business development projects.

14. S h t h th OGC i ?So, what has the OGC given us?
• GML for geographic data exchange
• Standardised methods for interacting with
GIS web servicesGIS web services
– NB: care over the phrase "web services"
• Wide range of GIS web services:
Web Map Service– Web Map Service
– Web Coverage & Feature Services
– Catalog Service for the Web

15. GML –
Geographic Markup Language
• GML is one of the older standards
• Concerned with data, not representation
(cf KML)(cf. KML)
• Has evolved with time to add functionality
• GML 2:
– XML for Simple FeaturesXML for Simple Features
– "Simple features" 2D vector data only + feature
collectionscollections
– Version 2.1.2 used as foundation for OS Mastermap

16. GML 3GML 3
• In addition to simple 2D linear features:
features with complex, non-linear, 3D geometry; featuresg y
with 2D topology, features with temporal properties,
dynamic features, coverages, and observations;
• represent spatial and temporal reference systems, units
of measure and standards information;
• use reference system, units and standards information in
the representation of geospatial phenomena,p g p p ,
observations, and values;
• represent default styles for feature and coveragerepresent default styles for feature and coverage
visualization.

17. OGC W b M iOGC Web Mapping
Th diff t i f i t t• Three different services of interest:
– Web Map Service (WMS), Web Feature Service (WFS),
Web Coverage Service (WCS)
• Server-side mapping methodsServer side mapping methods
• Map (and other) requests are made through
standard HTTP methodsstandard HTTP methods
– URL
– Parameters supplied within URL (GET) or hidden in
document (POST)
– One request gives one response (stateless)

18. Example –
UCL's ICEDSUCL s ICEDS
Partially transparent GOES imagery overlaid on SRTM topography – iceds.ge.ucl.ac.uk

19. Cascaded services
• ICEDS can act as client as well as server
• Requests for non-local layers are passed
through to other services on the webthrough to other services on the web
• This is known as cascadingThis is known as cascading
WMS
UCL ICEDSWEB BROWSER UCL ICEDS
Acting as proxy: cascade
requests, assemble map
WEB BROWSER
ICEDS portal Web requests, assemble map
OGC server
ICEDS portal Web
site
OGC Services,
e g CIESIN
WMS
OGC server
Geospatial
(DHTML +
Javascript)
WMS
e.g. CIESIN
WFS/
WCS
Geospatial
data storage
p )

20. Wh OGC?Why OGC?
• Standardised access to data and map
servicesservices
• Independence from vendor/database/OSp
=> choice, flexibility, upgrade, 'free' prototyping
• Clients can start to access different
providers' services with minimal extraproviders services with minimal extra
programming

21. Best of breed,,
software independence
These are the OGC t d d l
UCL ICEDS
These are the
components we use.
Chosen best of breed
OGC standards also
give independence
from spatial
Acting as proxy: cascade
requests, assemble map
Chosen best of breed
open source elements
BUT could easily
from spatial
database / file
systems in
ti d t &
Stores case annotations
Mapserver
substitute different
components, e.g.
vendor solutions
presenting data &
catalogue to the
Internet
Geoserver Files
vendor solutions Internet
PostGIS

22. Google Maps versus OGC WMSg p
1 - Interfaces
• GUI
WMS t i i d d t f GUI– WMS, etc., is independent of GUI
– Open source interfaces are available to buildp
upon, e.g. OpenLayers, similar to Google Maps
ArcGIS Explorer NASA WorldWind globes– ArcGIS Explorer, NASA WorldWind globes
• Programming: what's in an API?g g
– The Google API has clearly encouraged uptake
Si il API il bl f th i l &– Similar APIs available for other commercial &
open source clients, e.g. OpenLayers, ESRI

23. Google Maps versus OGC WMSg p
2 - Data
• The principle advantage of Google (or
Microsoft OS Multimap etc )Microsoft, OS, Multimap etc.)
• However a number of cascadable, open, p
access global data WMS
O St t– OpenStreetmap
– JPL's OnEarth global satellite mapsg p
– UCL ICEDS for terrain
b t 'll h t f & h t d t t t– but you'll have to pay for & host data to get
same resolution as Google (etc)

24. Google Maps versus OGC WMSg p
- Mixed mode
• Possible to mix Google Maps data layers
with OGC WMSwith OGC WMS
– Still needs a Google Maps API key
http://www openlayers org/dhttp://www.openlayers.org/d
ev/examples/example-
list.html

25. OGC i h t i tiOGC service characteristics
"Complex"• "Complex"
– Needn't be more complex to set up than vendor web-
GIS solutions
– Possibly, if using separate client and server
components, but standardised interfaces
– Issues of security, authentication, availability in cross-y y
institution systems. Fair comparison?
• "Slow"Slow
– If unconstrained client, server will have to work to
generate maps Difficult to cachegenerate maps. Difficult to cache.
– So constrain zoom, using cached tiles (Ka-Map)

26. W b d tWeb data access
• We can move beyond simple Web client
access to map imagesaccess to map images
• Most GIS vendors are OGC members
• Therefore most support connection to
OGCOGC services
– WMS common and relatively matureWMS common and relatively mature

27. WMS Connection from ArcGISWMS Connection from ArcGIS
ArcGIS 8.3…

28. W b d t (2)Web data access (2)
• What if we want to access source data?
W b C S i (WCS)• Web Coverage Service (WCS) –
raster/coveragesg
• Web Feature Service (WFS) –
/fvector/features

29. Cadcorp – WMS & WFSp

30. Comsine Ltd.:
WCS demonstrationWCS demonstration
wcs_url = 'http://iceds.ge.ucl.ac.uk/cgi-
bin/wcs?map=wcs map&SERVICE=WCS&REQUEST=GetCoverbin/wcs?map=wcs.map&SERVICE=WCS&REQUEST=GetCover
age&COVERAGE=srtm&FORMAT=GEOTIFFINT16'
res = max(tif_pixsize_x, tif_pixsize_y)
wcsRequest = wcs_url + "&CRS=EPSG:" +
str(epsg) + "&BBOX=" + bbox + "&RESX=" +
str(res) + "&RESY=" + str(res)
logger.info("WCS REQUEST " + wcsRequest)
if (dem == None):if (dem == None):
demFilename = str(time.clock()) + ".tif"
urllib.urlretrieve(wcsRequest, filename)
cleanup = 1

31. Beyond Web MappingBeyond Web Mapping
• Web processingp g
• Service chaining
• Sensor webs• Sensor webs
– http://www.opengeospatial.org/projects/groups/sensorweb
– http://vast.uah.edu/SensorML
• City & building modelling (CityGML)y g g ( y )
• ISO TC/211: geospatial standards

32. GEOSSGEOSS
• Global Earth Observation System of Systems
(http://www earthobservations org/)(http://www.earthobservations.org/)
• An attempt to create an umbrella systemp y
giving access to data from national &
international remote sensing space agenciesinternational remote sensing space agencies
• Hence distributed data, and probablyHence distributed data, and probably
distributed processing
• Each agency maintains own services

33. GEOSS componentsp

34. ConclusionsConclusions
• If we have Google do we need the OGC?• If we have Google, do we need the OGC?
• Yes and no – depends on your applicationp y pp
• Probably not for mass-market / consumer
( )web mapping (but, mixed system mash-ups)
• But yes if you want to:• But yes if you want to:
– use own base maps, or complex overlays;
– connect desktop GIS to GIS web services;
start mixing data from different services; or– start mixing data from different services; or
– create complex geospatial models or services.

35. W b P iWeb Processing
• In future we will want to do more than just
display data off the webdisplay data off the web
• What if we want to insert some processingp g
(from rendering through to data analysis)?
And do this in flexible on the flyAnd do this in flexible, on-the-fly
architectures?
• Web Processing Services (WPS)
R t ti l d l ti l– Runs geocomputational models or geospatial
operations on user-supplied datasets

36. S i h i iService chaining
How do we compose a sequence of• How do we compose a sequence of
processes, a service chain?p
• Adopt technology from the IT industry:
• BPEL: Business Processes Execution
LanguageLanguage
• Increasingly important for distributed data &
distributed processing
This is just part of the interoperable• This is just part of the interoperable
infrastructure needed

37. Fine dust (PM10) prediction maps
Discover
Log in
Access
Order
SSE Portal
Order
Discover
Service
Request
access
Results
View SOS/WMS
BPEL Workflow
Web Service at IRCEL
Web Service at Vito
OVL
Forecast
Model
RIO
Interpolation
Model
SOS
PM10
WMS
PM10
Maps
WCS/WMS
AOD
SOS
Forecasts
Model Model Maps
Slide from Yves Coen,
GEOSS ADC Architecture
W k h 2/2008 AOD
ToolkitPM10
measurements
OVL Forecasts
Workshop, 2/2008

38. Fine Dust Forecast - Result
Slides from Yves Coen,
GEOSS ADC Architecture
Workshop, 2/2008

39. W b i d th b tWeb mapping – under the bonnet
• Web mapping is a client/server system
Cli t PC d it b b• Client: your PC and its web browser
• Server: the web servers (hardware andServer: the web servers (hardware and
software)
Th t d i f b i d i• The art and science of web mapping design
is especially about the balance betweenp y
client and server processing

40. Simple server-side mapsp p
Client Server
1. Browser requests map URL
4. Map tiles are
requested to
build Web page
2. Web
server calls
Internet
map script
via CGI
3 S i t
Browser software Web server
3. Script
returns
HTML doc
referring to
Files /
Map script -
returns custom
HTML document
specific
map tile
images
Files /
D'base
Example: Streetmap.co.uk

41. Server-side interactive mappingpp g
Client Server
2
1
2
3
Web server
5
Internet
Web server
Servlet Exec
Browser software Internet
Map
Web
page
script
4
Fil /
Map
Server
Files /
D'base
Example: USGS Coastal Mapping (ArcIMS)

42. Client/server systemy
Client Server
1
2
5
3
4
Web server
6
Internet
Browser software
Web server
Servlet Exec
Internet
Map
Web
page
script
Browser
Files /
Map
ServerJava GIS app.
Browser
plug-in
Files /
D'base
Java VM
Example: Flash Earth