Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Topic:
AN APPROACH TO LAND COVER CLASSIFICATION
the LCML (Land Cover Meta Language) model
by
Antonio Di Gregorio
FAO-NRCE
Workshop on Land Cover Classification
(FAO LCML/LCCS v.3)
Paramarivo - Suriname 9-13 March
This presentation gives an overview on the basic
concepts to understand the logic of LCML Land Cover
Meta Language
(Part of the standardization in ISO TC211).
It address the characterization of land features and
harmonization of different Land Cover Classification
Systems or Legends, so that data from multiple
sources, and data prepared in different application
environments can be compared and integrated.
Individual data sets are developed using ISO 19109
compliant object oriented Application Schema
described using the LCML meta model.
SAME BASIC DEFINITIONS
Definition
Land cover is the observed (bio)physical cover
on the earth’s surface.
It includes vegetation and man-made features as well as bare rock. bare soil and inland water surfaces.
Land Cover is the basic information for geospatial data bases.
It can be considered as a boundary object to link different disciplines.
It has been always considered the key geographically explicit
feature which other disciplines may use as geographical reference.
Land cover is one of the most important element for description and study of the environment
the main resource controlling primary productivity for terrestrial ecosystems
can be defined in terms of land (and its cover)
land cover is the easiest detectable indicator of human interventions on the land
Land cover is a critical parameter for environmental databases
land cover changes quickly over time
GEO societal benefits and land cover observations
Climate Land change & GHG
emis. Water+energy exchanges
Weather Land–surface climate int. Vegetation characteristics
Health Land change / disease
vectors / boundary cond.
Disasters Fire monitoring
Land degradation assess.
Agriculture Cultivation pattern+forestry
Land degradations
Ecosystems Change environment cond.
Services + accounting
Energy Bio-energy/biomass
Wind/hydro power assess.
Water Water resources / quality Land+water use pattern
Biodiversity Ecosystem characteristics Habitats + fragmentation
Observing land cover as ECV
Integrated land cover observations
IN-SITU (+ IKONOS type)
periodically (usually 1-10 yrs)
Detailed physionomy
Floristics and species distribution
Crop type and rotation etc.
Thematic detail
Spatial deta
il
high
high low
LANDSAT/SPOT – type
inter-annual (1-5 yrs)
Vegetation
physionomy
Global archives
In situ database
Assuming observation
continuity and
consistency
Global daily observations
Land type/
Phenology
Completed and endorsed by IGOS partnership and GEO in 2007
CHARACTERIZATION AND CLASSIFICATION OF
LAND FEATURES
To classify is a human activity.
Classification is the means whereby we order knowledge. Our lives are surrounded with systems of classification, limned by standards,
formats, etc.
The oldest method to communicate knowledge is, no doubt, human language
and conversation, specific language elements or specialized terms are created
to exchange particular types of information.
A body of shared knowledge as a basis for communication is therefore
part of most sciences, and historically we find ample evidence of specialized
terminology, hierarchical thinking and classifications established
within those disciplines.
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
Sokal (1974) defines a classification as:
“the ordering or arrangement of objects into groups or sets on the basis
of their relationships”, and
Bowker and Star (1999) as:
“a spatial temporal or spatio-temporal segmentation of the world”. They define
a ‘classification system’ as “a set of boxes (metaphorical or literal) into which things
can be put in order to then do some kind of work bureaucratic or knowledge
production”.
The “real world” Teoretical representation
of a classification system
A CLASSIFICATION WILL NEVER BE ABLE TO REPRESENT ALL THE ASPECTS OF THE
REAL WORLD. IT MUST BE CLEAR THAT A CLASSIFICATION REFLECT (JUST) A
SPECIFIC SCOPE (OR FEATURES OF A SPECIFIC GEOGRAPHIC AREA) FOR WHICH HAS
BEEN DEVELOPED
WE MUST RECOGNIZE “THE BALANCING ACT” INHERENT IN
CLASSYFYING.
THE CLASSICAL WAY HOW TO REPRESENT THE REAL WORLD :
MAP
STATISTICAL REPORT
LEGEND
1a Forest
2b Agriculture
3c Urban
4d Rock
5e Water
6f ...........
7g............
Legend CLASSES
are a generalization
of the reality explicated
(formalization of meaning)
using a narrative text
(class definition).
13
•It should be recognized that no classification system can reflect the
social and/or the natural world fully accurately
•Classification (categorization) is an highly dynamic process related
to geographic areas, time and culture
•There are and it will be always multiple ways to categorize (segment) the
real world phenomena, all of them have the same legitimacy
AN HISTORICAL EVIDENCE
Single ontology methods to formalize a class meaning is proved
ineffective to fully describe the thematic content of a feature.
It is (usually) source of vagueness and ambiguity.
It is not functional in a modern data management.
•Different terms used for same concepts (Synonymy).
•Different understanding of homonymous concepts (Polysemy)
(e.g. the various meanings of the term ‘forest’ for forestry, env. mod. etc.)
•Different understandings of the relations among common concepts.
•Common instances across databases assigned to different
concepts in different ontologies.
SEMANTIC PROBLEMS ARE CONSTANT PART OF THE HUMAN
SOCIAL RELATIONS (AND HUMAN LANGUAGE )
COFFEE
AMERICAN
ESPRESSO
CAFFELLATTE
CAPPUCCINO
LUNGO
CORRETTO CON GOCCIA
MACCHIATO
IN VETRO IN CUP CALDO (WARM)
FREDDO (COLD)
RISTRETTO
•Categories (classes) are usually limited in number.
This forces the map producer to drastically generalize reality.
•Class definitions are imprecise, ambiguous or absent. The build up of the definition in the form of a narrative text is unsystematic (many diagnostic criteria
forming the system are not always applied in a consistent way) and in any case do not always reflect
the full extent of the information.
•Generalization into categories where meaning is very often limited to the class
name, or has only an unclear class description, implies rigidity in the transfer
of information from the data producer to the end user community. End users have limited if any possibility to interact with the data, and must therefore accept them ‘as is’.
The representation of the granularity of the aspects summarizing a specific feature of the real world is
drastically reduced or lost.
•Often some vagueness in the class definition is artificially included by the map
producer to hide some ‘technical anomalies’
SHORTHCOMINGS of CURRENT “SINGLE ONTOLOGY” SYSTEMS
STANDARDIZATION ISSUE A FIRST AND RECOGNISABLE
LIMITOF THE ACTUAL CLASSICAL APPROACH
•Mapping (and/or conceptual representation of a particular geographic feature)
is a local activity, so at one level it can be understood why there is the tendency
to establish unique classification systems to fit local conditions
•Any land surface is at a certain level (or scale of observation) heterogeneous
and the standards to represent and generalize those land characteristics
are about as diverse as the land itself
•In geographic information truth as a distinct, incontrovertible and correct
fact cannot exist
HARMONIZATION AND SEMANTIC
INTEROPERABILITY OF GEOGRAPHIC DATA SETS
A BRIEF OVERVIEW
Many classifications of geographic phenomena are often a black box to anyone
outside the immediate group involved in the classification process.
In the worst cases LU/LC information are treated as “data” by many users who don’t
fully understand its inherent relativism
PREMISES
L.C. Semantic and Semantic Interoperability between L.C. Systems
the core of the problem
Despite the great need of data harmonization there is a
huge problem of compatibility and comparability
between land cover (LC) products.
PREMISES
Harmonization should be the process whereby
differences among existing definitions of land
characterization are identified, clarified and
inconsistencies reduced. However, this is not the
actual case, where current maps exist mostly as
independent and incompatible data sets.
Causes of data bases heterogeneities
different coverage (level of
detail) due to different scope –
user needs
different relations due to
different classification
perspectives
different semantics due to
different conceptualizations
1 2
0
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
11 4
5 6
0
28 7
9
10
1 2
0
3 4
5 6
0
211 2
0
1 2
0
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
3 4
5 6
0
21
11 4
5 6
0
28 7
9
10
3 4
5 6
0
21
3 4
5 6
0
21
11 4
5 6
0
28 7
9
10
11 4
5 6
0
28 7
9
10
FA
O -
LA
ND
CO
VE
R T
OP
IC C
EN
TR
E -
GL
CN
Interoperability Definition
Interoperability is the ability of “systems” to operate in conjunction, on the exchange
and reuse of available resources, …, according to the intended use of their
providers, in order to fulfill the requirements of a specific task (Kavouras and Kokla,
2008)
Basic Points of Interoperability Definition
“systems”: NOT technical systems (e.g., hardware/software or
products), BUT any organized field of human activity (e.g.,
organizations, geospatial databases, infrastructures, etc.).
“according to intended use”: essential that users (humans or
computers) understand geographic information without ambiguity
understanding involves the study of meanings, i.e., semantics.
the semantics of geographic information are described by a
geospatial ontology.
interoperability focus: semantic interoperability between geospatial
ontologies.
SAME EXAMPLES OF SEMANTIC INTEROPERABILITY
EXAMPLES OF DIFFERENT FOREST DEFINITIONS
Global land cover
datasets
Important factors for disagreement:
1. Land cover/forest definitions:
IGBP legend : percent tree cover >60%
GLC2000 legend : percent tree cover >15%
2. Spatial heterogeneities
Comparison of forest classes
History: independent, incompatible datasets
FA
O -
LA
ND
CO
VE
R T
OP
IC C
EN
TR
E -
GL
CN
CHINA 2000-2005
1. Forest (arbor forest, mangrove forest, bamboo forest),
2. Open forest land,
3. Shrub land,
4. Unestablished forest,
5. Nursery land,
7. Other land
6. Forest suitable land,
Classes
SRI LANKA 1992-1999
Classes
1. Lowland rain forest
2. Moist monsoon Forest
3. Dry monsoon Forest
11. Teak
6. Mangroves
7. Riverine dry forest
12. Mahogany
5. Sub Montane forest
9. Conifers
10. Eucalypts
4. Montane forest
8. Sparse and open forest
Classes Classes
NEPAL Jafta 2000
S. KOREA 1999-2006
1. Sal (Shorea
robusta)
2. Tropical Mixed
Hardwood
3. Upper and
Lower Mixed
Hardwood
4. Chir Pine
5. Blue Pine/
Cypress/Yew
6. Fir/ Hemlock/
Spruce/ Cedar
7. Shrub
8. Agri/Grass
10. Water bodies
11. Bare Land
12. Snow 5. Rocky Area
2. Cultivated land on steep slopes
1. Stocked
forest
3. Unstocked forest land
4. Denuded forest land ??
1. Conifers
2. Hardwood
3. Mixedwood
5. Oaks
4. Alpine Conifers
(RSS 1996)
Classes
NEW ZEALAND LCDB 2
Classes
MYANMAR 2000-2005
1. Evergreen Forest
4. Deciduous Dipterocarp Forest
5. Pine forest
6. Hill forest
7. Bamboo Forest
3. DUMD (Dry Upper Mixed Deciduous) Forest
2. MUMD (Moist Upper Mixed Deciduous) Forest
14. Shifting Cultivation
8. Mangrove Forest
9. Evergreen Forest / Open
10. MUMD Forest / Open
11. Dry Forest
12. Mangrove Open
13. Scrub Land
Classes
INDIA 2003
1. Very Dense
Forest (VDF)
2. Moderately
Dense Forest
(MDF) ??
3. Open
Forest(OF)
4. Scrub
Classes
MONGOLIA 1975-2005
1. natural forest
2. planted forest
3. shrubs
4. forest area damaged by fire
5. forest damaged by insects
6. logging area
7. open forest
8. area for reforestation
9. non forest area
50 Fernland 51 Gorse / Broom 52 Manuka and or
Kanuka 53 Matagouri 55 Sub Alpine
Shrubland 56 Mixed Exotic
Shrubland 57 Grey Scrub
54 Broadleaved Indigenous Hardwoods 68 Deciduous Hardwoods 69 Indigenous Forest
15 Alpine Grass-/Herbfield
41 Low Producing Grassland
43 Tall Tussock Grassland
70 Mangrove
...........follow
SAME EXAMPLE OF FOREST CLASSIFICATIONS IN ASIA
EXAMPLES OF LAND COVER CLASSIFICATONS IN SOUTH ASIA
STANDARDIZATION OF A COMMON SET OF CLASSES
TO OVERPASS THE PROBLEM OF SEMANTIC
INTEROPERABILITY?
A WRONG PARADIGM
LAND COVER AS ANY OTHER GEOGRAPHIC FEATURE CAN BE CONSIDERED AS A
“ CONTINUUM”
A CLASS OR CATEGORY IS A “PARTITION”OF THIS CONTINUUM. ANY PARTITION
(CATEGORY) AS IS OWN LEGITIMACY
COVER % 1 100
COSTA RICA (70-100%) US (30-100%)
FAO (10-100%)
A NEW APPROACH TO SEMANTIC HARMONIZATION:
•IT IS IMPOSSIBLE ( AND ALSO NAIVE) TO THINK TO STANDARDIZE
CLASSES
•INSTEAD OF TO TRY TO STANDARDIZE CATEGORIES (CLASSES) IT IS MORE
IMPORTANT AND EFFICIENT TO STANDARDIZE THE WAY HOW A SPECIFIC
CATEGORY HAS BEEN CREATED
•THIS IMPLIES THE FORMALIZATION OF THE RULES AND CONDITIONS
ON HOW A SPECIFIC CLASS HAS BEEN CONCEPTUALIZED
•A COMMON PROTOCOL OF RULES TO DESCRIBE THE “ REAL WORLD”
WILL GUARANTEE COMPLETE FLEXIBILITY IN THE CREATION OF
SPECIFIC CATEGORIES AND IN THE SAME TIME WILL ASSUIRE
SEMANTIC INTEROPERABILITY BETWEEN THEM
•THE SYSTEM SHOULD ALSO ASSURE MIGRATION FROM
THE HUMAN LANGUAGE TO MACHINE REPRESENTATION OF THE RULES
AND CONDITIONS WITH WHICH THE “FORMALIZATION OF THE MEANING”
OF A GIVEN CATEGORYHAS BEEN CREATED
Real world
FOREST
Thematic
chategory
As part of a
applied legend
Agriculture
Forest
Wetland
Urban
..........
..........
Legend 2
Legend 3
Database
by chategories
OTHER LEGENDS DIFF: FROM
THE ORIGINAL MUST RELATE
TO THE ORIGINAL TO APPROACH
THE DATABASE
ONTOLOGICAL PROBLEMS
SUBJECTIVITY OF TRANSLATION
TOO LARGE CONCEPTUAL JUMP
SOURCE OF AMBIGUITIES,
VAGUENES AND INCONSISTENCES
From the real world to the construction of a database
From the real world to the construction of a database
BEFORE TO CREATE CONCEPTUAL
CATEGORIES THE REAL WORLD
BE REPRESENTED IN A MORE
SIMPLE AND EFFICIENT WAY
Real world Database
by objects Objects
FOREST
WOODLAND
SAVANNA
USER DEFINED
CATEGORIES
EN
D U
SE
RS
THE LCML/LCCS CONCEPT
•LCCS adheres to the concept that it is deemed as more important to standardize
the attribute terminology rather than the final categories.
•LCCS works by creating a set of standard diagnostic attributes to create
or describe different LC classes.
•They act as standardized building blocks and can be combined to
describe the more complex semantics of each LC class in any separate application
ontology (classification system)
•The emphasis is no longer on the class name but on the set of clearly quantifiable
attributes
IT IS FROM 1998 THAT FAO STARTED TO DEVELOP THIS NEW STRATEGY
CREATING DIFERENT VERSIONS OF LCCS (Land Cover Classification System)
From LCCS v.2 to LCML
•Complete the passage from a “classification system” to a reference
“Meta-Language”.
•Develop a full “Object oriented” syntax to support a (possible use of)
relational data base
•Extend the descriptive power of the system
•Instantiate the “formalization of the meaning” using a “standard” modelling
language
•Rationalize the “language elements”.
•Improve the capacity of the system to be extended with a large series of
“application related” attributes
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
MAJOR CHARACTERISTCS OF LCML (LCCS v3)
Rigorous categorization of the language elements
•BASIC OBJECTS purely based on physionomic aspect
BIOTIC ABIOTIC
•PROPERTIES of basic objects (further physionomic characterization
of basic objects as height, cover etc)
•BASIC OBJECTS CHARACTERISTICS (descriptive elements of
the basic objects not directly related to its physiognomic
characterization as veg artificiality etc.)
•COVER CLASS CHARATERISTICS (descriptive elements of the land
cover class as a whole as climate, landform etc.)
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
LCML is a sophisticated Land Cover description engine
that can be part of many different FDM
Development of LCML has focus on four main aspects:
•Very high descriptive power of different Land Cover situations
•Rigorous syntax ( categorization of language elements and
rules to relate them)
•Clear classification criteria
•Avoiding as much as possible use of complex categories
(definitions) and constrains for basic elements
Majority of the elements (and their definition) are derived from
the previous versions of LCCS. They are more than 10 years in
use and have gained a considerable acceptance world wide
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
MAJOR CHARACTERISTCS OF LCML (LCCS v3)
Simple classification criteria
Fundamental idea: a predefined set of basic elements
(BIOTIC and ABIOTIC) and their properties enriched in their
semantic significance with “element” and “ class”
characteristics can be arranged in different types of vertical
and horizontal patterns to describe a wide variety of
distinctive and detailed land cover situations
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
{ Two LCML_Element(s)
in one LC_Stratum
must be of the same
class unless there is an
LCML_SequentialTemp
oralRelationship
between them or unless
they are of type
LCML_Gramineae or
LCML_Forbs.
{An LCML_Stratum
may have an
LCML_VerticalRelations
hip only with all the
next LCML_Stratum in
the ordered list of
LCML_Stratum that
belongs to the same
LCML_LandCoverClass}
LCML_SequentialTemporalRelationshipTypes
+ sequentialSameYear
+ sequentialOtherYear
(from LCML_LandCoverClassStructure)
<<enumeration>>
{An LCML_Element may have an
LCML_SequentialTemporalRelationship only
with the next LCML_Element in the ordered
list of LCML_Elements which belong to the
same LCML_Stratum}
LCML_Element(from LCML_LandCoverClassStructure)
0..10..1
LCML_SequentialTemporalRelationship
LCML_SequentialTemporalRelationship
- length[0..1] : LCML_PermittedPositiveRealValues
- type[1] : LCML_SequentialTemporalRelationshipTypes
LCML_Stratum(from LCML_LandCoverClassStructure)
0..10..1
LCML_VerticalRelationship
0..* +layerElement0..*
+layer
layered
LCML_HorizPattern
+ PatternType : CharacterString
+ PatternCoverPercentage : LC_PermittedPercentageValues
+ PatternOccurrence : LC_PermittedPercentageValues
(from LCML_LandCoverClassStructure)
1..*+composedOf
1..* {ordered}
+composedBy
composition
LCML_LandCoverClassificationSystem
LCML_ClassCharacteristic(from LCML_ClassCharacteristics)
LCML_LandCoverClass(from LCML_LandCoverClassStructure)
1..*
+composedBy
+composedOf1..*classComposition
1..*+systemClasses 1..*
+systemsystemCollection
+describe+describedBy
Characteristic
Core Structure of the LCML meta model – allowing multiple
stratification and horizontal pattern description.
LCML_MetaLanguage
ClassificationSystem
ApplicationSchema
ApplicationSchemaInstance
<<Instantiates>>
<<Instantiates>>
<<Instantiates>>
Meta Meta Meta Model showing the
Instantialion relationships between the
LCML MetaLanguage and a
Classification Ssytem and an
Application Schema
Land Cover Meta Language describing the common sub
elements and their relations that may be combined to
create any Land Cover Classification System. ISO 19144-2
is at this level of the hierarchy.
A specific Land Cover Classification System that
effectively establishes a dictionary of classifiers that may
be used to create a specific "System" (or Feature
Concept Distionary) for use in defining a specific
Application Schema. ISO 19144-1 addresses the
organization of classifiers and their geometric
representation using a Discrete Coverage at this level of
the hierarchy.
An Application Schema, compliant with ISO 19109 Rules
for Application Schema, that defines the specific set of
feature types used in the Feature Catalogue (Legend) for
a Land Cover data set or series of data sets.
A specific set of Land Cover data, compliant with the
Application Schema.
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
Joint Standards and Registers Both ISO 19144-1 and ISO 19144-2 are joint
standards with the UN and ISO.
This means that the standards and the
registers are joint – the UN may manage
Classification Registers on behalf of ISO and
itself.
The ISO Directives Part 1 Annex H establish rules for establishing a
Registration Authority.
The technical management board designates registration authorities in
connection with International Standards on the proposal of the technical
committee concerned.
Registration authorities should be qualified and internationally
acceptable bodies; ....
Registration authorities should be required to indicate clearly in their
operations that they have been designated by ISO or IEC ...
Registration functions undertaken by the registration authority under the
provisions of the relevant International Standard shall require no
financial contribution from ISO or IEC or their members.
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
Register in 19144-2 LCML
The purpose of Registration in ISO 19144-2 LCML is different than that of 19144-1.
The LCML contains a set of fixed metalanguage elements that are the basic vocabulary for describing different land cover classification systems. This vocabulary has to be stable in order for descriptions of different land cover classification systems to be comparable.
Changes to the properties of land cover element characteristics and land cover class characteristics as expressed in the metalanguage objects may be done by registration.
This provides a simpler route to extend the descriptive aspects of the metalanguage without changing the basic metalanguage elements. It also permits the characteristics and associated code lists to be extended.
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
Control Body Each Register requires a
Control Body to manage the submission of elements into the register.
A joint control body, or bodies is/are proposed for the registers required in 19144-1 and 19144-2.
The UN FAO may establish its own registers addressing its own implementation of the standards, just as any nation or other organization may do.
The Organization Responsible
MultipleSubmitters
A few Register Managers
Several Control Bodies
Many Registers
One Registry Manager & Registry for an organization
LCML IS NOT ONLY A SYSTEM TO HARMONIZE
CLASSIFICATION SYSTEM
BUT AN ADVANCED APPROACH TO FORMALIZE
THE MEANING OF LAND FEATURES
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
Corine L.C.Level 3
1.1.2 Discontinuous urban fabric
Most of the land is covered by structures, Buildings, roads and artificially surface areas associated
with vegetated areas and bare soil, which occupy discontinuous but significant surfaces. Between
10% and 80% of the land is covered by residential structures.
Land Cover elements (basic objects) NOT STRUCTURED, NOT EXPLICIT (in the data base), SOME TIME VAGUE/INCOMPLETE
Land Cover attributes NOT STRUCTURED, NOT EXPLICIT (in the data base), SOME TIME VAGUE/INCOMPLETE
Land Cover el. cover or polygon occupancy SOME TIME VAGUE, NOT EXPLICIT FOR EACH ELEMENT, SOME VALUE NOT STORED IN THE DATA BASE
Land Cover elements horizontal/vertical relationship, NOT STRUCTURED, NOT EXPLICIT (in the data base), VAGUE/INCOMPLETE
SINGLE ONTOLOGY STILL A GOOD APPROACH TO
REPRESENT GEOGRAFIC FEATURES?
A NEW WAY TO REPRESENT THE COMPLEXITY
OF THE REAL WORLD
A MODEL INSTEAD OF
A CLASS NAME AND TEXT DESCRIPTION
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
EXAMPLE 1: VEGETATION LAYERING
Broadleaved deciduous trees with two strata of shrubs
.
EL_ClosedBroadleavedDecidousForest
EL_CBLDFStratum1 EL_CBLDFTrees EL_CBLDFTreesHeightRange
+ maxValue: Real = 7.0 {readOnly}
+ minValue: Real = 5.0 {readOnly}
EL_CBLDFTreesCoverRange
+ maxValue: Real = 100.0 {readOnly}
+ minValue: Real = 70.0 {readOnly}EL_CBLDFTreesNaturalSeminaturalVegetation
EL_CBLDFTreesBroadLeaf
EL_CBLDFStratum2
EL_CBLDFStratum3 EL_CBLDFScrubs3
EL_CBLDFScrubs2
EL_CBLDFScrubs2NaturalSeminaturalVegetation
EL_CBLDFScrubs3NaturalSeminaturalVegetation
EL_CBLDFScrubs2HeightRange
+ maxValue: Real = 5.0 {readOnly}
+ minValue: Real = 3.0 {readOnly}
EL_CBLDFScrubs2CoverRange
+ maxValue: Real = 40.0 {readOnly}
+ minValue: Real = 20.0 {readOnly}
EL_CBLDFScrubs3HeightRange
+ maxValue: Real = 0.5 {readOnly}
+ minValue: Real = 0.3 {readOnly}
EL_CBLDFScrubs3CoverRange
+ maxValue: Real = 20.0 {readOnly}
+ minValue: Real = 10.0 {readOnly}
EL_CBLDFTreesDecidous
+growthFormQuality 1
1
1
1
1
1
11
+growthFormQuality 1
1
+height
1
1
+cover
1
1
+height
1
1
+cover
1
1
+leafPhenology 1
1
111
1
11
+height
11
+cover
1
1
+growthFormQuality1
1
+leafType1
1
Strata 1
Strata 2
Strata 3
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
EXAMPLE 2: Aquatic and regularly flooded vegetation-Closed
mangroves trees
EL_ClosedMangroveForest
EL_CMFStratum1
EL_CMFStratum2
EL_CMFTreesEL_CMFTreesHeightRange
+ maxValue: Real = 7.0 {readOnly}
+ minValue: Real = 5.0 {readOnly}
EL_CMFTreesCoverRange
+ maxValue: Real = 100.0 {readOnly}
+ minValue: Real = 70.0 {readOnly}EL_CMFTreesNaturalSeminaturalVegetation
EL_CMFTreesBroadLeafEL_CMFTreesDecidous
EL_CMFWaterBody
EL_CMFWaterBodyDailyVariation
EL_CMFWaterBodyHeightRange
+ maxValue: Real = 1.0 {readOnly}
+ minValue: Real = 0.2 {readOnly}
EL_CMFWaterBodySalinity
+ type: LC_WaterSalinityTypes = moderatelySaline {readOnly}
EL_CMFWaterBodyArtificiality
+ type: LC_ArtificialityTypes = natural {readOnly}
1
+growthFormQuality1
1
1
1
11
+leafPhenology 1
1
+leafType 1
1
+cover
1
1
11
+dailyVariation
1
1
+height
1
1
+waterAndAssociatedSurfacesQuality1
1
+waterAndAssociatedSurfacesQuality 1
1
+height
11
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
EXAMPLE 3: COMPLEX BIOTIC AND ABIOTIC RELATIONSHIP
Building with garden on top with young trees seasonally covered by snow
MOVING FORWARD AN ADVANCED SYSTEM
DO WE REALLY NEED “CATEGORIES” TO CREATE A
MODERN LAND COVER DATA BASE?
burned water stress
FOREST (according to FAO definition)
DIFFERENT ASPECTS OF THE REAL WORLD ARE GENERALIZED, THE USERS
LOOSE IMPORTANT ASPECTS OF THE REALITY. THE LOST OF INFORMATION
IS IRREVERSIBLE
DO WE REALLY NEED CATEGORIES?
DO WE REALLY NEED CATEGORIES?
•Despite the obvious constraints, categories are useful means whereby
we cope with the “continuous” nature of the real world and its multiplicity
of information.
•Categorization is also a powerful method how we share knowledge.
•Categories/specialized terminology are therefore part of most sciences,
and historically we find ample evidence of specialized terminology,
hierarchical thinking and classifications established within those disciplines.
• THE PROBLEM IS WERE WE POSITION CATEGORIES IN A DATA BASE
AND HOW THEY INTERACT WITH THE POSSIBILITY/CAPABILITY TO
MAXIMIZE THE REPRESENTATION OF THE GRANULARITY OF
INFORMATION OF THE REAL WORLD
Information Technology Division (CIO) Information Systems Architecture Forum
THE OBJECT ORIENTED METHOD TO MAP LAND COVER
ACCORDING TO FAO/GLCN
Real world
(universe of
Discourse acc.
to ISO)
Different methods
to produce a data base
•Predefined list of classes
•Major class groups further
enriched in each polygon
•Land Cover caracterization
using LCML elemens
Data base built up
according to LCML
OBJECTS
Tree
Shrub
Building
Etc.....
AGGREGATION
OF THE OBJECTS
ACCORDING TO
DIFFERENT
ONTOLOGIES
CORINNE
ANDERSON
NATIONAL
CLASSIFIC.
Analysis by rules
LCML (LAND COVER
META LANGUAGE)
ONTOLOGICAL RELATIONSHIP
OF THE OBJECTS
MAP PRODUCERS DOMAIN END USER DOMAIN
LCCS 3
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
X 1
X 2
Y 1
Y
2 X 1
X 2
F 1
Z
2 X 1
X 2
Y 1
Y
2 X 1
X 2
F 1
Z
2 X 1
X 2
Y 1
Y
2 X 1
X 2
F 1
Z
2
X 1
X 2
Y 1
Y
2 X 1
X 2
F 1
Z
2
DIN
AM
IC D
ATA
BA
SE
RE
PR
ES
EN
TE
D
BY
OB
JE
CT
S N
OT
CA
TE
GO
RIE
S
APPLICATION 1
Carbon flux
estination
APPLICATION 2
Agriculture
assessment
APPLICATION 3
Rangeland
Assessment
APPLICATION 3
Ecosystem
accounting
Capture granularity of real world information
with LCML semantic protocol
THE LCML PROVIDERS COMUNITY ------ GLOBAL END USERS COMUNITY
Use the structured amount of information to built ad hoc
categories to develop specific application
TH
E G
RA
NU
LA
RIT
Y O
F I
NF
OR
MA
TIO
N O
F T
HE
RE
AL
WO
RLD
RE
PR
ES
EN
TIN
G T
HE
RE
AL I
TY
WIT
H L
CM
L
US
E L
CM
L F
LE
XIB
ILIT
Y T
O C
RE
AT
E S
PE
CIF
IC
U
SE
R D
EF
INE
D C
AT
EG
OR
IES
USER 1
USER 2
USER 3
USER n…
CATEGORIES
CATEGORIES
CATEGORIES
CATEGORIES
LC
ML S
EM
AN
TIC
EL
EM
EN
TS
BA
SIC
O
BJE
CT
S, P
RO
PE
RT
IES
a
nd
CH
AR
AC
TE
RIS
TIC
S
THE LCML FLUX OF INFORMATION
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
FINAL COMMENTS
•To solve the problem of a modern Land Features characterization
and semantic inter-operability we need a change of perspective •Move away from the vagueness of class names/description
•Up to date “formalization of the meaning”
•Semantic interoperability is a responsibility of “data producers”
•It is not an option is a basic requirement of any geographic
data set
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
FINAL COMMENTS
Be potentially able to converse with other systems. This inherent harmonization
property should not relay (only) on expert judgment but the harmonization
process should be automatized as much as possible.
Recognize the balancing act inherent in classifying (Bowker and Star, 1999).
A classification will never be able to fully represent all the aspects of the real world,
therefore it must be clear it reflect (just) a specific scope for which has been
developed.
Render voice retrieval (Bowker and Star, 1999) by allowing users to detail
and compare classes using a detailed class description (systematically organized
with a list of explicit measurable diagnostic attributes), thus avoiding the risk of
systems being impermeable to the end users.
An Ideal Classification System should:
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
FINAL COMMENTS
Standardization process should focus on the rules and conditions
how a feature is conceptualized rather than acting just on the class name.
The “formalization of the meaning” of the system and its
components should be formulated with the most modern methods of
modellization.
A modern classification should not be considered an isolated structure
but more a functional component of a rather complete system for data
management.
Information Technology Division (CIO) Knowledge Information Systems Architecture Forum (KISAF)
FINAL COMMENTS
Does not matter how big is the distance (10 or 100000 miles), to arrive
You always need to start from the first step (Chinese proverb).
Thank you.