This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Quality assessment ofImage & Corine Land Cover 2000
Guillermo Villa (1)
ge & Co e d Cove 000database in Spain.Lessons learnt from the project. Future actions
15 July 2005 ICC 2005 A Coruña 1
Antonio Arozarena (1)Isabel del Bosque (1)Ana Porcuna (2)Nuria Valcarcel (1)
(1) IGN Spain(2) Tragsatec
1. Background of I&CLC2000 in Spain
Summary
. Bac g ou d o &C C 000 Spa
2. Quality control
3. Results
4. Lessons learnt
5 Future actions: SIOSE Project
215 July 2005 ICC 2005 A Coruña 2
5. Future actions: SIOSE Project
6. Conclusions
2
On 27 June 1985 and by virtue of a decision taken by the Council of Ministers of the European Union (EC/338/85), “an experimental project for the acquisition of data, the coordination and standardization of information on the state of the environment
315 July 2005 ICC 2005 A Coruña 3
and natural resources in the Union” was launched. It was called the CORINE Programme (Coordination of Information on the Environment).
The period from 1985 to 1990 saw the creation of an environmental information system, which included the development of nomenclatures and methodologies regarding land cover agreed by the member states of the EU.
415 July 2005 ICC 2005 A Coruña 4
member states of the EU.
3
Goals
Homogeneity
Databases comparable between different countries
Allow periodic updating
515 July 2005 ICC 2005 A Coruña 5
Principal Objectives I&CLC2000:
Background of I&CLC2000 Spain
• Updating of CLC90 database
• Evaluation of Land Cover changes between 1990 and 2000
615 July 2005 ICC 2005 A Coruña 6
990 d 000
4
715 July 2005 ICC 2005 A Coruña 7
815 July 2005 ICC 2005 A Coruña 8
5
29 Participant countries:
Background of I&CLC2000
915 July 2005 ICC 2005 A Coruña 9
Participant organizations:
Background of I&CLC2000 Spain
ETCETC--TETE
1015 July 2005 ICC 2005 A Coruña 10
A.G.E. CC.AA.
6
CLC in Spain
IGN (Instituto Geográfico Nacional) is Spanish EIONET’s National ReferenceSpanish EIONET’s National Reference Centre for Land Cover
IGN has coordinated CLC90 and I&CLC2000
CLC 90/2000 inSpain: 5th level nomenclature
1115 July 2005 ICC 2005 A Coruña 11
nomenclature
I&CLC2000 finished (EU’s 3 level and Spanish 5 level databases)
1215 July 2005 ICC 2005 A Coruña 12
7
1315 July 2005 ICC 2005 A Coruña 13
1415 July 2005 ICC 2005 A Coruña 14
8
1515 July 2005 ICC 2005 A Coruña 15
1615 July 2005 ICC 2005 A Coruña 16
9
Decentralization in Spain
1715 July 2005 ICC 2005 A Coruña 17
19 Autonomous regions → 19 Production Teams
Organization of CLC2000 in Spain
AEMA (EEA)Comisión de seguimiento
I&CLC2000 Europa
CTE Medioambiente terrestre
CCI (JRC)
PFN-MIMAM
NRC / CNIG – IGNCoordination
A.G.E. CC AA
Comisión de seguimientoI&CLC2000 España
I&CLC2000 Europa
1815 July 2005 ICC 2005 A Coruña 18
A.G.E.•Mº Fomento•Mº Medioambiente•Mº Agricultura•Mº Ciencia y Tecnología•Mº Economía•Mº Defensa
CC.AA.-Galicia -Castilla la Mancha-Principado de Asturias -Región de Murcia-Cantabria -Comunidad Valenciana-País Vasco -Extremadura-Comunidad Foral de Navarra -Andalucía-Aragón -Islas Baleares-Cataluña -Canarias-Castilla y León -Ceuta-La Rioja -Melilla-Comunidad de Madrid
Producto 3a – CLC2000 NacionalProducto 3b – CLC90 nacional revisado Producto 3aNivel5 – CLC2000 Nacional a nivel 5Producto 3bNivel5 – CLC90 nacional revisado a nivel 5
Países miembros
Producto 4 – Cambios CLC nacionalesProducto 4Nivel5– Cambios CLC nacionales a nivel 5
Países miembros
Producto 5 – Mosaico europeo IMAGE2000 EEA/JRC
Producto 6a – CLC2000 europeo
P d 6b CLC90 i dEEA/JRC
2715 July 2005 ICC 2005 A Coruña 27
Producto 6b – CLC90 europeo revisado/J
Producto 7 – Cambios CLC90 europeo EEA/JRC
Producto 8 – CLC raster 250 m EEA/JRC
Producto 9 – CLC raster 100 m EEA/JRC
Producto 10 – CLC estadísticas 1 km2 EEA/JRC
Producto 11 – Metadatos nacionales Países miembros
Producto 12 – Metadatos europeos EEA/JRC
Product 1: Landsat 7 ortho-rectified scenes
2815 July 2005 ICC 2005 A Coruña 28
15
Product 2: National IMAGE 2000 Mosaic
2915 July 2005 ICC 2005 A Coruña 29
Product 3: CLC90 corrected + CLC2000 national
3015 July 2005 ICC 2005 A Coruña 30
16
Product 4: National changes database
3115 July 2005 ICC 2005 A Coruña 31
Product 5: IMAGE 2000 european mosaic
3215 July 2005 ICC 2005 A Coruña 32
17
Product 6: European CLC2000 and CLC90corrected
3315 July 2005 ICC 2005 A Coruña 33
Product 7: European changes database
3415 July 2005 ICC 2005 A Coruña 34
18
Products 8 and 9: Raster format CLC at 250 mand 100 m
Product 10: Estadísticas CLC por km2
3515 July 2005 ICC 2005 A Coruña 35
Product 11: National Metadata
Product 12: European Metadata
3615 July 2005 ICC 2005 A Coruña 36
19
1. Background of I&CLC2000 in Spain
Summary
. Bac g ou d o &C C 000 Spa
2. Quality control
3. Results
4. Lessons learnt
5 Future actions: SIOSE Project
3715 July 2005 ICC 2005 A Coruña 37
5. Future actions: SIOSE Project
6. Conclusions
Radiometric and geometric control of Landsat 7 orthorectified images
Quality control
Landsat 7 orthorectified images
Verification of vector database by each production team
Verification of vector database by European Technical Team (2 visits)
3815 July 2005 ICC 2005 A Coruña 38
( )
Final validation: determination of global database quality level
1) Land use and Land cover information is a critical issue in Spain:Water supply (quantity and quality)Water supply (quantity and quality)Urban sprawl (especially in the coast)Tourism expansion/sustainabilityInfrastructure planning & ecological impactDesertificationR l E t t b (“b bbl ”?)
7315 July 2005 ICC 2005 A Coruña 73
Real Estate boom (“bubble”?)Housing problem (huge price increase)Climate change (Kyoto protocol)Very high rate of Land Cover change in Spain
Lessons learnt
2) V bi i i2) Very big impact in:
Communication media (Journals, radio, TV)
Political life
7415 July 2005 ICC 2005 A Coruña 74
38
Lessons learnt
3) Politicians “sometimes” make promises or
take decisions based on what is published ontake decisions based on what is published on
journals, not on Technical Reports →
We should be careful with the information we
send to journals and other media
7515 July 2005 ICC 2005 A Coruña 75
Databases should be easy to understand by
non expert people
Lessons learnt
4) Corine Land Cover addresses this information needs but … doesn’t solve them completely, because:
Scale too little
Minimum polygon too big
Updating period too long
7615 July 2005 ICC 2005 A Coruña 76
Nomenclature not detailed enough
Nomenclature not good enough
39
Lessons learnt
Nomenclature not good enough (1):
It is actually a “thematic map” (classification), not a complete GIS database
It mixes cover and use (→ semantic
7715 July 2005 ICC 2005 A Coruña 77
problems; erroneous conclusions,…)
Lessons learnt
Nomenclature not good enough (2):
Mixed classes → information lost
It produces incoherencies when compared with other Land Use and land Cover databases (especially when
7815 July 2005 ICC 2005 A Coruña 78
analyzed by non expert people)→ political “wars”.
40
Lessons learnt
Nomenclature not good enough (3):
Only one parameter per polygon (class label). Reality is much more complex !!:
e.g: percentage of trees in a forest, building density,…
7915 July 2005 ICC 2005 A Coruña 79
e.g. Class 244: Agroforestry areas
The “DEHESA” is an old land-use
i i fsystem consisting of scattered oak trees on pasture or crop land used by livestock.
This system is
8015 July 2005 ICC 2005 A Coruña 80
yfrequent in the west, south-west and central parts of the Iberian Peninsula
41
1. Background of I&CLC2000 in Spain
Summary
. Bac g ou d o &C C 000 Spa
2. Quality control
3. Results
4. Lessons learnt
5 Future actions: SIOSE Project
8115 July 2005 ICC 2005 A Coruña 81
5. Future actions: SIOSE Project
6. Conclusions
Present situation of Land Use/LandCover information in Spain
Various institutions feel a strong need for more big scale / high detail information
Some institutions of National and regional (Autonomous Communities) levels, keep geographic information databases related with land use/cover at higher level of detail (geometrically –bigger scale- and semantically) than CLC2000.
8215 July 2005 ICC 2005 A Coruña 82
E.g.: Minister of Agriculture, Minister of Environment, various Autonomous Communities,…
42
8315 July 2005 ICC 2005 A Coruña 83
8415 July 2005 ICC 2005 A Coruña 84
43
8515 July 2005 ICC 2005 A Coruña 85
But this databases are not integrated:But this databases are not integrated:Different scales
Incomplete cover of Spanish territory (regional databases)
Different updating periods
8615 July 2005 ICC 2005 A Coruña 86
Incompatible nomenclatures (very difficult translations between them)
44
INSPIRE visionData should be collected once and maintained at the level where this can be done most effectivelywhere this can be done most effectivelyit should be possible to combine seamlessly spatial information from different sources across Europe and share it between many users and applicationit should be possible for information collected at one level to be shared between all the different levels, detailed for detailed investigations, general for strategic purposes
8715 July 2005 ICC 2005 A Coruña 87
geographic information needed for good governance at all levels should be abundant under conditions that do not refrain its extensive use
SIOSE Project (Spanish Land Cover information system)
Nominal Scale: 1:25.000Minimum polygon:
2 Ha general1 Ha artificial coverages
8815 July 2005 ICC 2005 A Coruña 88
1 Ha artificial coveragesPeriodicity: 5 years
45
SIOSE Project
Base information:SPOT 5 images XS+P 2.5m (2005)Orthophotos 0.5 m (PNOA)
Decentralized production (regional governments)
8915 July 2005 ICC 2005 A Coruña 89
SIOSE Project
Terrestrial geo-referenced (and oriented) digital h d bphoto database
9015 July 2005 ICC 2005 A Coruña 90
46
SIOSE Project
A new Land Use / Land Cover Data Model:Multi-criteria (cover + use)M l i ( l i l ibMulti-parameter (multiple attributes possible for 1 polygon)Object oriented (UML description, GML implementation)Extensible (for future and/or specific
9115 July 2005 ICC 2005 A Coruña 91
needs)Designed in a cooperative effort to fulfill the requirements of all institutions/social agents
SIOSE Conceptual Data Model: Design Guidelines
Separation between Land Cover (biophysical criteria) and Land Use (socio-economic criteria)
There is only one geometric entity class in SIOSE (POLYGON)
Mixed Classes in SIOSE: created by association of
9215 July 2005 ICC 2005 A Coruña 92
ysingles classes
47
Thematic Working groups are using the first draft version of the SIOSE Conceptual Data model it describes abstract classes and relations, and some characteristic attributes, linked to the existing land cover / uses nomenclatures. So it is not the
SIOSE Conceptual Data Model: Design Guidelines
g /future SIOSE physical data model.SIOSE Conceptual Data model: in UML notation (Unified Modeling Language). It provides normalized notation about classes and relations between them, according to ISO TC211 and OGC recommendations.This standardized notation provides flexibility to the model, so that Thematic Working Groups and future users can
9315 July 2005 ICC 2005 A Coruña 93
so that Thematic Working Groups and future users can modify and extend it easily.
SIOSE Conceptual Data model has considered previous Spanish Land Cover and Use Nomenclatures and Databases, from national and regional institutions.
SIOSE Data Model: Conceptual Model UML Diagram
ABSTRACT CLASS
InheritanceAssociation
Olive grove Class inheritsproperties and attributes of:
9415 July 2005 ICC 2005 A Coruña 94
properties and attributes of:•Cover•Agricultural Surfaces•Woody Crops
48
4. SIOSE Conceptual Data Model: UML
9515 July 2005 ICC 2005 A Coruña 95
SIOSE Conceptual Data Model: Relation Polygon – Land Cover – Use
9615 July 2005 ICC 2005 A Coruña 96
49
Agricultural Coverages
9715 July 2005 ICC 2005 A Coruña 97
Land Uses
9815 July 2005 ICC 2005 A Coruña 98
50
Example: Translation from CLC Nomenclature to SIOSE
SIOSE: Polygon associated to the following cover:Cover 1: 1.4 Infrastructures
Spanish National Plan for Observation of Territory y
(PNOT)
15 July 2005 ICC 2005 A Coruña 107
Planes Autonómicos de Ortofotografía
Aérea
0,10 a 0,25 mAlta:
Pancro: 1 a 10 mMultiesp: 4 a 30 m
Media:Pancro: 10 a 15 mMultiesp: 20 a 50 m
5 a 2 años 12 a 3 meses 4 a 1 meses
Baja:Multiespectral:100 a 1.000 m
30 a 1 día
Plan Nacional de Ortofotografía Aérea
(PNOA)
Plan Nacional de Teledetección(PNT)
Frecuencia Temporal
PLAN NACIONAL DE OBSERVACIÓN DEL TERRITORIO (PNOT)
0,5 m
Nombre del Plan
2 años
Resolución espacial
1ª fase:Obt ió t t i t d i á 5 a 2 años 12 a 3 meses 4 a 1 meses
SPOT (HRVIR) Landsat 7 (ETM+)
IRS Landsat 5 (TM)
Futuro sistema Pleiades TERRA (ASTER)
Futuro sistema español deobservación de la Tierra
IRS
Nombre del Sistema Bases de Datos Organismos 2.000 a 5.000 10.000 25.000 25.000 a 50.000 100.000 a200.000 500.000 1.000.000
AGE BCN 25 BCN 200
30 a 1 díaFrecuencia Temporal
Bases Cartográficas (BC)
Escala de la Base de DatosSistema de Información
(Armonización, síntesis y diseminación según los principios INSPIRE entre los distintos organismos que recogen o utilizan esta
información)
Terra (MODIS)
NOAA
Fotografía aérea1: 30.000
Fotografía aérea escalas
1: 10.000 a 1: 18.000
Envisat (MERIS)
SPOT (Vegetation)
2 años
Sensores
2ª Fase: Extracción y
Obtención y tratamiento de imágenes aeroespaciales
Área temática
Sistema CartográficoNacionalCartografía básica
10815 July 2005 ICC 2005 A Coruña 108
CC.AA. ♦ ♦
Ocupación del Suelo
Sistema de Informaciónde Ocupación del Sueloen España (SIOSE)
Bases de Datos de Cobertura y Usos del Suelo
AGE /CC.AA. Corine Land Cover
Otros tipos de información
AGE / CC.AA /Universidades
Parámetros Bio-físicos(NDVI, Temp. Suelo, etc…)
SIOSEMCAMFE
t acc ó ydiseminación de la información
Nacional
55
Validación estadística
Método escogido:Aleatorio estratificado se trata de determinar laAleatorio estratificado se trata de determinar la probabilidad p de acierto de cada estrato desde el punto de vista temático
Dos productos a validar:CLC2000
CLCh
10915 July 2005 ICC 2005 A Coruña 109
CLChange
Obtención del tamaño de la muestra n
El nivel de confianza ( )2 1 ppzEl nivel de confianza determina el parámetro z
Intervalo de confianza d
Estimación de la probabilidad p de acierto para cada estrato
C ió fi i d > N
( )20
1d
ppzn −=
nnn 11 0
0
−+=
11015 July 2005 ICC 2005 A Coruña 110
Corrección por finitud-> N número total de polígonos en cada estrato
N1+
56
Validación CLC2000
Un estrato por cada clasePara que cada clase tenga representaciónPara que cada clase tenga representación estadística
Determinación del tamaño de la muestra en cada CCAA
Para que en cada CCAA se puedan obtener datos dí i fi bl
11115 July 2005 ICC 2005 A Coruña 111
estadísticos fiables
Validación del CLChanges
Dos estratos:Polígonos de cambioPolígonos de cambio
Validación del tipo de cambios
Polígonos sin cambioValidación de que no se han omitido cambios
11215 July 2005 ICC 2005 A Coruña 112
57
Información de referencia:
Para CLC2000Imágenes Landsat IMAGE2000Imágenes Landsat IMAGE2000
Ortofotografía aérea
Mapa de Cultivos y Aprovechamientos
Para CLChangeImágenes Landsat IMAGE2000
11315 July 2005 ICC 2005 A Coruña 113
Imágenes Landsat CORINE90
Tamaño de Muestra CLC2000
Nivel de confianza del 95% z=1.96
Intervalo de error del 5% d=0 05Intervalo de error del 5% d=0.05
Probabilidad a priori del 90% p=0.9
Tamaño de la muestra para una población infinita n0=138
11415 July 2005 ICC 2005 A Coruña 114
58
Tamaño Muestra CLC2000
Corrección por finitud
Cálculo de n teniendo en cuenta que la población q pno es infinita
Frequency de cada clase en ámbito nacional N
nn 0=
11515 July 2005 ICC 2005 A Coruña 115
Nnn 11 0 −+
=
Tamaño de la muestra por CCAA
Para determinar el número de muestras en cada CCAA se necesita un Frequency de q ytodas las clases al nivel 3 para cada CCAA
Reparto proporcional del tamaño de la muestra por CCAA
11615 July 2005 ICC 2005 A Coruña 116
59
Selección de muestras por CCAA
Una vez conocido el tamaño de la muestra en cada clase se seleccionan aleatoriamente los n polígonos de entre todos los existentes en esa clase
Con Excel se pueden realizar los siguientes pasos:
=ALEATORIO()*tamaño de la muestra
Y a continuación redondear hacia arriba =REDONDEAR.MAS()
11715 July 2005 ICC 2005 A Coruña 117
()
Selección de muestras
Sería necesario comprobar que no existen muestras repetidasp