CORINE Title Land Cover. First name SURNAME György BÜTTNER Position Senior advisor

Transcription

1 CORINE Title Land Cover First name SURNAME György BÜTTNER Position Senior advisor Name EEA/ IDM4 of the entity Place, Copenhagen, date 19/10/2015

2 Outline CLC basics Geographic coverage of CLC2012 Methodology of updating Mapping examples Technical follow up European products Outlook to CLC2018 Conclusions 2

3 CORINE Land Cover basics Mapping ~permanent surface features at scale 1: based on physical characteristics (changes > 1 year) MMU: 25 ha (5 ha for changes); MMW: 100 m Nomenclature: 5 main groups, three levels, 44 level-3 LU/LC classes (representing Europe) Basic data support: satellite imagery Ancillary (in-situ) data: national orthophotos, topographic maps, VHR imagery Implemented by national teams Inventories: 1990, 2000, 2006, 2012 MMU, MMW and nomenclature have not changed since the beginning! 3

4 Geographical coverage: EEA39 EU28 countries 5 additional EEA members 6 cooperating countries Total area = ~6 Mkm 2 AL, XK: first time CLC implemented by NT DE, ES and IE: new semiautomatic methodology GR: parallel implementation of CLC2012 and CLC2006 W-Balkan countries and TR: difficult start TR: costs covered by national resources 4

5 Updating methodology In majority of countries: photointerpretation (CAPI) uniform methodology (Guidelines) In rest of countries: semi-automatic methodology highly country / data dependent Combined image processing, in-situ data integration and cartographic generalisation (FI, IE, IS, NL, NO, SE) Full bottom-up solution: generalisation of high resolution national land monitoring data (DE, ES) 5

6 Strategies for updating CLC Revision of old CLC Interpretation of new CLC Revision of old CLC Interpretation of CLC Changes Derivation of CLC Changes Updating first Derivation of new CLC Change mapping first Change mapping first serves better the different MMUs (5 ha/25 ha) Mapping changes is under expert s control Change mapping first methodology is described in EEA Guidelines 6

7 Different MMUs - what does it mean? MMU for CLC-Change = 5 ha ( ~ 224m x 224m) Demand set by policy MMU for CLC = 25 ha (= 500m x 500m) CLC and CLC-Change have different scales! consequences 7

8 CLC change mapping: principles 1: All changes > 5 ha should be mapped IMAGE / CLC2006 IMAGE2012 CLC-CHANGE CLC ha IMAGE / CLC2006 IMAGE2012 CLC-CHANGE CLC ha CLC2012 CLC2006 CLC-change!!! 8

9 CLC change mapping: principles The real evolution process should be described Changes are interpreted directly, by comparing old and new satellite images. IMAGE / CLC2006 IMAGE 2012 CLC-CHANGE CLC ha 35 ha 17 ha If we compared CLC2006 and CLC2012, false changes would be obtained. CLC-Change CLC2012-CLC2006!!! 9

10 CLC change mapping example Burnt forest (Kosovo) Changes are interpreted directly on satellite images, and not derived by intersection of two databases. EXPERTS INVOLVEMENT IS ONE OF THE STRENGTHS OF CLC. 10

11 Producing the new CLC CLC new is produced by an automatic GIS process: CLC new = revised CLC old (+) CLC-Change old,new Revision of CLC2006 is mandatory in order to map real changes! < 25 ha polygons are generalised (using priority table) Consequence: CLC new CLC old CLC-Change old,new 11

12 Technical support and follow up Training of national teams: 14 in-country trainings / technical discussions, where the CLC team and/or the methodology was new (sometimes with a joint HRL training). (EEA - ETC/SIA FÖMI) Verification: Samples of CLC data were checked usually twice in each country. (EEA - ETC/SIA FÖMI) 1 st verification: soon after start, all photointerpreters checked. 2 nd verification: close to finish, checking randomly selected areas. Regional level verification in ES and IT. Dominantly remote verifications, a few missions only. Main aim is harmonisation, not an assessment of accuracy. Altogether 128 verification reports. On-line technical self checking of final datasets (new, ETC/SIA-GISAT) Final technical control (DBTA Report, ETC/SIA-GISAT) 12

13 Technical support and follow up 13

14 Ancillary (in-situ) data in verification VHR 2004 Example: (in Belgium) VHR imagery was used. Time series feature is very useful for verification of changes VHR

15 Status of CLC2012 Content of V18_4 (published recently): 37 countries have finished CLC2012 Missing from full coverage: Andalusia (ES) Turkey (except Eastern part) Special features: CLC2006_GR Revised CLC2000 (26 countries) Revised CLC2006 (37 countries) blue: CLC2012 yellow: others (incl. earlier CLCs) green: Urban Atlas red: CLC coverage document (pdf) 4610 download events between July 1 st to October 8 th

16 CLC V18_4 16

17 Validation of CLC2012 layer Preliminary results from SIRS Validation by zones exceeding km 2 30 countries validated already in 16 zones Countries not yet validated: CY, DE, DK, ES, FR, GR, SE, TR Country/group accuracy range: 73.1 % 95.8 % Overall accuracy (30 countries) = 83,6% tbc 17

18 European CLC products on Layer Type Main characteristic CLC status layers; Revised CLC status layers (2000, 2006) vector (primary) geodatabase and SQLite Standard level-3 nomenclature; MMU=25 ha CLC Change layers raster (derived) geotiff vector (primary) geodatabase and SQLite raster (derived) geotiff 100 m grid 250 m grid Standard level-3 nomenclature; change polygons have two attributes; MMU=5 ha Two 100 m grids (formation code and consumption code) 18

19 Outlook to CLC2018 Number of countries applying semi-automatic methodology is likely to slowly further increase. need for harmonization of methodologies (EAGLE experience) and adaption of guidelines. Photointerpretation methodology is likely to dominate. Training will be needed in some countries (new team, new methodology or both). More teams will have access to LPIS (or equivalent) in-situ data, so agriculture classes will improve ( revision of CLC2012). Similar verification procedure and final technical control are proposed as implemented in CLC2012 to maintain good data quality. 19

20 Outlook to CLC2018 ideas for enhancement Introduce attributing in some of the classes instead of using a separate CLC class (maintaining compatibility with previous surveys is not difficult): Construction sites (133) construction attribute in 11x, 12x, 132, 14x Irrigated arable land (212) irrigation attribute in 211, 22x (more general meaning: irrigated agriculture land). Burnt areas (334) burnt status attribute in 31x, 322, 323, 333, 412 but also in 22x (more general meaning: burnt vegetation). Considering to increase the details of CLC: MMU change from 25 ha to 10 ha or 5 ha Involves replacing 100m MMW with 50m Significant increase in workload increased costs! 20

21 Conclusions CLC is the most popular dataset of EEA; Four inventories already, spanning >25 years with the same nomenclature and resolution; Highly cooperative participation in the EEA39 countries, skilled national teams in place; Strong technical support (training, verification, DB acceptance) are key elements to maintain good data quality; Users should be aware that status layers and change layers have different resolutions; Copernicus programme will provide a long-term background to CLC (6 years repetition); Future: gradual shift towards bottom-up approach: national land monitoring data will feed the European CLC by applying the EAGLE concept. 21

22 Acknowledgement National project teams in 38 countries Ana Sousa and Ilona Schiøler (EEA) - Grant Agreements with countries Kees Van Duijvendijk (EEA) database management Barbara Kosztra (ETC-SIA/FÖMI) - training, verification Márta Gunawan (ETC-SIA/FÖMI) - verification Gábor Taracsák (ETC-SIA/FÖMI) - InterChange development Róbert Lehoczki (ETC-SIA/FÖMI) - InterChange support Gergely Maucha (ETC-SIA/FÖMI) projection issues Tomas Soukup and Miroslav Kopecky (ETC-SIA/GISAT) - final technical control, European mosaic 22