Digital Earth as a Digital Twin
Michael F. Goodchild
University of California
Santa Barbara
What is a digital twin?
• “A virtual representation of the real world,
including physical objects, processes,
relationships, and behaviors”
• “GIS is foundational for any digital twin”
– Esri, https://www.esri.com/en-us/digital-
twin/overview
Danette Allen, NASA
https://ntrs.nasa.gov/api/citations/20210023699/downloads/ASME%20Digital%20
Twin%20Summit%20Keynote_final.pdf
Issues with definitions
• Esri: a representation
– including processes
– no representation can be perfect
– what is the purpose of the representation?
• more than visualization?
• Replica: “fully describes…from the micro
atomic level to the macro geometrical level”
– “The map is not the territory” (Korzybski, 1933)
• A digital twin can only be fraternal
– fraternal twins share only part of the genome
– even identical twins are not identical
The purpose of digital twins
• Accurate simulation of a system
– in order to evaluate what-if scenarios
– predicting the impact of proposals
– replicate, simulate, evaluate
The Gore speech of 1998
• “I believe we need a ‘Digital Earth’. A multiresolution,
three-dimensional representation
of the planet, into which we can embed vast
quantities of geo-referenced data.”
– http://www.zhanpingliu.org/research/terrainvis/digi
talearth.pdf
– all of the data will be of limited resolution, hence
imperfect
– no reference to processes, simulation, what-if
experiments
• But travel back to the first ISDE…
Implementing Digital Earth:
A Research Agenda
Michael F. Goodchild
University of California
Santa Barbara
Perspectives on Digital Earth
 1. An immersive environment
– “I believe we need a 'Digital Earth'. A multiresolution,
three-dimensional
representation of the planet, into which we
can embed vast quantities of
geo-referenced data.” U.S. Vice President
Gore, 1/98
 Spin, zoom, pan
– "fly-by" technology
Immersive environments
 Head-mounted devices
 Immersadesk
 The "cave"
 Standard computer displays
– 2D window on manipulable 3D objects
– Nick Faust, Georgia Tech
– SRI Digital Earth, Terravision
– powerful processors, 3D graphics
Research challenges
 Smooth zoom
– 10km to 1m resolution
– consistent data structures
smooth transitions to more detailed data
color matches
– projections
orthographic for the globe
projected for local detail
Georgia State: nested azimuthal projections
Research challenges (2)
 Visualization
– renderable data
– non-renderable data
iconic representation indicating presence
symbolic representation
– user-centered views
reduce resolution in periphery
avatar
A dynamic Digital Earth
 Simulations of past and future
conditions
 A library of simulation models
– applied to local conditions represented by
data
 A tool with enormous educational value
 PCRaster demonstrations
– University of Utrecht, Peter Burrough
Modelling uplift in Sabah,
Malaysia.
UCEL
Over a period of several million years
movement along the faults has created long
sediment-filled valleys
UCEL
Faults
Relative
vertical
displacement
Sediment
The demo illustrates:
 A simplified model of normal faults and
landform before uplift
 Reaction of landform to gradual vertical
displacement along the parallel normal faults
 Erosion and deposition as a result of vertical
movements (red is erosion - blue is
deposition)
 Emergent behaviour of rivers leading to
development of braided streams
tectonics
Research challenges
 Data structures and modeling
– no finite difference models on the curved
surface of the planet
– finite element models based on triangles?
– object-based models
 Describing models
– metadata
– libraries of models
Research challenges (2)
 Software environments
– PCRaster
 Calibration, verification, accuracy
 Integration across domains
– coupling models
– distinct ontologies
Summary: four perspectives
 An immersive environment
 A metaphor for information organization
 A distributed database transparent to
the user
 A representation of the planet's
dynamics
Is a digital twin distinctive?
• Does it have distinct principles?
– or is it just more of?
• finer resolution
• more accurate process models
• more layers and variables
– compare “big data”
• bigger than small data?
• too big to handle?
• Is there a threshold that merits the term
“digital twin”?
– of data resolution, functionality, accuracy…?
The uncertainty problem
• How to visualize uncertainty?
– are spatial resolutions already finer than those of
the human eye?
• How to incorporate uncertainty into
predictions?
– uncertainty will come from:
• data
• the process of integrating or fusing data
• simulation models
• the means of communicating or presenting the data
Dealing with uncertainty
• To ignore it is unethical
• Fitness for use
– is the level of uncertainty acceptable for my
particular use case?
– must digital twins be tied to particular use cases
• and never repurposed?
• Propagate uncertainty into predictions
– using methods of sensitivity analysis
– using simulation
Some ethical issues
• Uncertainty
• The potential for misinformation
– deep fakes
– false inferences
• Privacy
– of individuals
• Lack of transparency
– proprietary (black box) software
– provenance of data
The missing pieces
• Data fusion and integration
• Interoperability of process models
• Search for data and process models
• Integration of digital twins
• Education
– what are the principles?
– lack of software for demonstration
Some takeaways
• There is abundant and rapidly growing
interest across industry and academia
• To date there has been little interest in
academic GIScience
• The imperfect nature of all digital twins raises
issues
– there are no standards for what can be claimed to
be a digital twin
• There are strong links between digital twins
and Digital Earth