Sharing and Visualizing Environmental Data using Virtual Globes

1
Sharing and Visualizing Environmental Data using
Virtual Globes

• Jon Blower, Alastair Gemmell, Keith Haines
• Reading e-Science Centre, ESSC, University of
  Reading
• Peter Kirsch, Nathan Cunningham, Andrew Fleming
• British Antarctic Survey, Madingley Road,
  Cambridge
• Roy Lowry
• British Oceanographic Data Centre, Liverpool

2
E-Science Connected science

• We need to be able to connect data across the
  environmental sciences (right)
• Visualization is very important
• But complicated by plethora of file formats and
  tools
• Geographic Information Systems (GIS) often
  complex and vendor-specific
• Environmental science data are four-dimensional
• GIS historically focussed on land surface

3
How important is visualization?

• (Microsoft case study with BP Hurricane
  Management System)
• "This solution is changing the way we do
  business. When the data is presented through a
  map-based interface, its amazing. It gives you a
  richer, bigger, more intelligent picture of
  whats going on. " Steve Fortune, BP
• http//www.microsoft.com/casestudies/casestudy.asp
  x?casestudyid201427

4
The Virtual Solar System

• gt 30 Virtual Globes currently available!

5
Combining data on a Virtual GlobeHurricane
Katrina, August 2005

• Storm track positions (analysed from ECMWF
  vorticity data) by Lizzie Froude, ESSC
• Sea surface temperature data from UK Met Office
  FOAM model
• Combination shows cooling of surface waters on
  right-hand side of cyclonic storm track
• High winds cause upwelling of cool, deep water

6
Keyhole Markup Language (KML)

• KML encodes simple geographic features in XML
• "Points, lines and polygons"
• Image overlays
• ltdescriptiongt field can hold HTML with links,
  images etc
• Contains several types of information
• Geolocation
• Style
• Camera control

7
Some currently-availableenvironmental datasets
in KML format
8
Virtual Globes as Discovery andBrowse tools

• BODC uses Google Earth as spatial metadata
  browsing tool for in-situ measurements
• Can easily check for errors
• E.g. ocean data located on land
• Misplaced component of linear ship track
• Displays light metadata, with links to more
  sophisticated information
• Developed in under a week!

9
Real-time data visualization

• British Antarctic Survey (BAS) used KML data
  feeds to support 2 scientific cruises in 2006
• Multiple data streams (ship location, sea temp
  and salinity, air temp and pressure) streamed to
  Google Earth in near real time
• Enabled real-time decision-making (e.g. tracking
  of predators, left)
• Saved ship time, hence

King penguin track overlain with concurrent
chlorophyll and satellite imagery
10
Beyond visualizationExtending the capabilities
of VGs

• Approach 1 Extend the client
• Only possible if VG exposes an API or is open
  source
• ArcGIS Explorer (.NET API)
• NASA WorldWind (Java API, Open Source)
• Approach 2 Do processing on a server
• Expose results as KML via a Web Service
• Can use web interface to control the Web Service
• Useful for closed-source systems (e.g. Google
  Earth)

11
Example ocean data assimilation

• Scientists need to compare ocean models and
  observations
• A Web Service performs comparison, then outputs
  results in KML
• Red dots show bad model-obs fits, green dots are
  good fits
• Clicking on an observation brings up more
  information
• Driving real improvements in the model

http//www.nerc-essc.ac.uk/alg/OceanDIVA.html
12
Virtual Globes pros and cons

• Pros
• Easy to use
• Easy data formats
• Compelling visualizations
• Cons
• Generally no data analysis functions
• Vertical and temporal dimensions not always
  handled properly (left)
• Patchy support for OGC Web Services
• (But products vary widely)

Vertical structure of Gulf Stream shown above sea
level in Google Earth
13
What is the best scientific use for VGs?
Have an idea
Discuss/explore
Do the work
Publish
Virtual Globes
MATLAB, IDL etc
14
Alternatives to Virtual Globes WebGIS

• Lots of systems now available for displaying
  geographic data on the Web
• Google/Yahoo Maps
• Microsoft Virtual Earth
• OpenLayers
• FreeEarth
• WorldKit
• Usually based on Javascript API
• Usually read imagery from map tile services
• Can often overlay simple features (KML, GeoRSS)

ICEDS (UCL)
Godiva2 (ReSC)
15
Did Hurricane Felix cause the seasurface to cool?

• Storm track from www.hurricane-tracking.co.uk
• Sea surface temperature from OSTIA (analysed
  satellite product), UK Met Office
• Little visible evidence of cooling.
• Why different behaviour from Katrina?

http//gearthblog.com/blog/archives/2006/07/weathe
r_and_sto.html http//lovejoy.nerc-essc.ac.uk8080
/ncWMS/godiva2.html?datasetOSTIAvariableanalyse
d_sst http//ghrsst-pp.metoffice.com/pages/latest_
analysis/ostia.html
16
The Bigger Picture
17
Beyond KMLGeography Markup Language (GML)
ProfileSeriesFeature

• KML GML v2 (simple features)
• GML v3 adds much more sophistication (complex
  features)
• Communities create profiles of GML ("Application
  Schema")
• Describes features of interest
• NERC Data Grid created CSML
• Climate Science Markup Language

Thanks to Andrew Woolf (STFC and NDG)
18
Open Geospatial Web Services

• Open Geospatial Consortium publishes open
  standards for geospatial Web Services

(plus many more!)
19
Importance of sharing geospatial data

• "Approximately 80 of business and government
  information has some reference to location, but
  until recently the power of geographic or spatial
  information and location has been underutilized
  as a vital resource for improving economic
  productivity, decision-making, and delivery of
  services" OGC vision
• EU INSPIRE directive mandates that all public
  bodies must provide
• Discovery services
• View services
• Download services

20
What I haven't discussed

• Security
• Need solution based on simple HTTP GET
• Licensing
• Public outreach

21
Summary

• Virtual Globes allow easy simultaneous
  visualization of lots of types of environmental
  data
• Well-suited to "exploratory" phases of scientific
  work
• Good way for data providers to advertise their
  wares
• KML is simple but limited way of encoding data
• It's a visualization format really
• GML should be used to encode data for true
  interoperability
• But needs concerted community effort to define
  feature types
• Simple interfaces are good!
• "scriptability" enables mashups

22
Also see

• Reading e-Science Centre booth
• Satoshi Sekiguchi, "GEO Grid"
• Keynote, Tuesday 0945
• Tim Foresman, "Digital Earth The New Digital
  Commons"
• Keynote, Wednesday 1500
• Gen-Tao Chiang, "Driving Google Earth from
  Fortran"
• Session 2.2, Tuesday 1510
• Gobe Hobona, "Workflow enactment of grid-enabled
  geospatial Web Services"
• Session 3.4, Wednesday 1125
• Geolinking demo, White Rose Grid
• Demo session, Wednesday 1430
• E-Science Highlights, Issue 2 (in your delegate
  pack!)