A Framework for Geospatial Uncertainty: From Concept to Communication
Adam Chilton, Dr. Ed Figura, Dr. Adel Bolbol, Dr. Gobe Hobona, Prof. Mike Jackson
Advanced Geospatial Information &
Intelligence Services Research
Research Programme and the Team
AGIS:
A
dvanced
G
eospatial
I
nformation &
I
ntelligence
S
ervices
Research
An industry led research programme funded by Dstl (Defence Science and Technology Labs) who as part of
the MOD (Ministry of Defence) provide science and technology advice into the MOD.
“Is the road 2 lane?”
“All weather?”
“What are the
chances of being
spotted?”
“Where is it safe to step?”
“How old is this information?”
“How accurate are
these coordinates?”
“Where did the
information come
from?”
“What is missing?”
Different Opinions
“I don’t want to see more confusion”
“I’m not interested in uncertainty – show me the facts”
“Whoa – information overload!”
“What does this all mean?”
“Of what use is it to me, to be shown uncertainty?”
Framework for Uncertainty
ISO 19115
ISO 19139
ISO 19157
UncertML
MGMP
WMS
SLD
KML
WxS
ISO
=
I
nternational
O
rganization for
S
tandardization
MGMP
=
M
OD
G
eospatial
M
etadata
P
rofile
UncertML
= a conceptual model and XML encoding designed for encapsulating probabilistic
uncertainties
Sources of Uncertainty
There is uncertainty in the data
from the moment it is measured...
… this propagates through
subsequent processes
… including how the data
is presented to the user
Uncertainty Representation
•
MGMP (
M
OD
G
eospatial
M
etadata
P
rofile)
– Data Quality Element of the profile
– Within Data Quality is an element called Quantitative
Result (i.e. a data quality analysis with a quantitative
value)
– Add a new quantitative result measurement for data
quality called: Uncertainty Statistics
Uncertainty Representation
class Fig A.4 : Data quality information
DQ_Element + nameOfMeasure: CharacterString [0..*] + measureIdentification: MD_Identifier [0..1] + measureDescription: CharacterString [0..1] + evaluationMethodType: DQ_EvaluationMethodTypeCode [0..1] + evaluationMethodDescription: CharacterString [0..1] + evaluationProcedure: CI_Citation [0..1] + dateTime: DateTime [0..*] + result: DQ_Result [1..2]
"report" or "lineage" role is mandatory if scope.DQ_Scope.level = 'dataset'
DQ_DataQuality
+ scope: DQ_Scope
LI_Lineage Metadata entity set information::MD_Metadata
«CodeList» DQ_Ev aluationMethodTypeCode + directInternal + directExternal + indirect DQ_Result "levelDescription" is
mandatory if "level" notEqual 'dataset' or 'series' DQ_Scope + level: MD_ScopeCode + extent: EX_Extent [0..1] + levelDescription: MD_ScopeDescription [0..*] DQ_Quantitativ eResult + valueType: RecordType [0..1] + valueUnit: UnitOfMeasure + errorStatistic: CharacterString [0..1] + value: Record [1..*] DQ_ConformanceResult + specification: CI_Citation + explanation: CharacterString + pass: Boolean MGMP_Quantitativ eResult + uncertaintyStatistic: UncertML_Statistic [0..*] +report 0..* +lineage 0..1 +dataQualityInfo 0..*