Urban Atlas

27 rue du Carrousel - Parc de la Cimaise – Immeuble I – 59650 VILLENEUVE D’ASCQ 03.20.72.53.64 - 03.20.98.05.78 - E-Mail : [email protected] - Site Internet : www.sirs-fr.com

S.A.S. au capital de 312.025 € - RCS LILLE 444 654 271 – APE 6311 Z - N° d’identification FR 07 444 654 271 - SIRET 444 654 271 00022

Urban Atlas DELIVERY OF LAND USE/COVER MAPS OF MAJOR EUROPEAN URBAN

AGGLOMERATIONS.

FINAL REPORT (V 2.0)

~ November 2011 ~

Urban Atlas FINAL REPORT

European Commission

SIRS sas SYSTEMES d’INFORMATION A REFERENCE SPATIALE

27 rue du Carrousel – Parc de la Cimaise – 59 650 VILLENEUVE D’ASCQ Tél : 03 20 72 53 64 – Fax : 03 20 98 05 78

Site Internet :http://www.sirs-fr.com Mail : [email protected]

08/01/2013

Page 2/96

SUMMARY

Document identification ___________________________________________________________ 6

Document release sheet _________________________________________________________ 6

Approval network ______________________________________________________________ 6

List of abreviations _______________________________________________________________ 7

Executive summary _______________________________________________________________ 8

1. Background __________________________________________________________________ 10

1.1. Nature and purposes _______________________________________________________ 10

1.2. Related Documents ________________________________________________________ 11 1.2.1. Input _________________________________________________________________ 11 1.2.2. Output ________________________________________________________________ 11

2. List of LUZs produced __________________________________________________________ 11

3. Urban Atlas Methodology (extract of Mapping guide v1.1) ____________________________ 12

3.1. Pre-processing and geometric adaptation of COTS navigation data _________________ 12

3.2. Pre-processing of Topographic Maps __________________________________________ 12

3.3. Classification and Interpretation ______________________________________________ 12

3.4. Application of FTS Sealing layer ______________________________________________ 12

3.5. Interpretation rules _________________________________________________________ 13

3.6. Minimum mapping unit ______________________________________________________ 14

3.7. Priority rules ______________________________________________________________ 14

3.8. Legend Table ______________________________________________________________ 15

4. Encountered difficulties and solutions ____________________________________________ 15

4.1 Adjustments to the "Mapping Guide" __________________________________________ 15 4.1.1. Interpretation rules_______________________________________________________ 15 4.1.2. Minimal Mapping Unit (MMU) ______________________________________________ 16 4.1.3. Typology : post "Urban Fabric" _____________________________________________ 17 4.1.4. Classification of scrubs and forestry regeneration. ______________________________ 18

4.2. Supply of images __________________________________________________________ 19 4.2.1. Amount of data and image processing _______________________________________ 19 4.2.2. Data types by LUZs ______________________________________________________ 20 4.2.3. Availability of archive images and Image quality ________________________________ 21 4.2.4. Image data format _______________________________________________________ 25

4.3. Use of "Soil-Sealing" _______________________________________________________ 25 4.3.1. Abstract _______________________________________________________________ 25 4.3.2. Assessment of users requirements __________________________________________ 26

4.3.2.1. Evaluation of the soils sealing layer as an input to the Urban Atlas ______________ 26


European Commission




08/01/2013

Page 3/96

4.3.2.2. Distribution of unfilled urban fabric polygons _______________________________ 29 4.3.2.3. Conclusions ________________________________________________________ 29

4.3.3. Detailed description of the methodological approach ____________________________ 30 4.3.3.1. NDVI calculation _____________________________________________________ 30 4.3.3.2. Image processing methodology _________________________________________ 30 4.3.3.3. Production flow ______________________________________________________ 33

4.3.4. Results _______________________________________________________________ 33 4.3.5. Recommendations for further UA updates ____________________________________ 35

4.4. LUZs limits _______________________________________________________________ 35

4.5. Road splitting _____________________________________________________________ 36

4.6. Recommandations for future use and updates __________________________________ 36 4.6.1. Mapping guide __________________________________________________________ 36 4.6.2. LUZs limits _____________________________________________________________ 36 4.6.3. Soil Sealing ____________________________________________________________ 37 4.6.4. Image management ______________________________________________________ 37

5. Quality control ________________________________________________________________ 37

5.1. Internal quality controls _____________________________________________________ 38 5.1.1. Attribute table controls ____________________________________________________ 38 5.1.2. Geometric control _______________________________________________________ 38 5.1.3. Continuous control_______________________________________________________ 38 5.1.4. Photo-interpreters’ cross-controls ___________________________________________ 38

5.2. External quality controls ____________________________________________________ 38 5.2.1. Pre-defined grid sampling _________________________________________________ 38 5.2.2. Determining the number of control points _____________________________________ 40 5.2.3. Determining the distribution of control points (Secondary sampling Unit) _____________ 41 5.2.4. Blind test by three European experts _________________________________________ 44 5.2.5. Quality improvement _____________________________________________________ 45 5.2.6. Advantages of the method _________________________________________________ 46 5.2.7. Intermediate delivery control _______________________________________________ 46 5.2.8. Final delivery control _____________________________________________________ 46 5.2.9. Quantifying accuracy : the confusion matrix ___________________________________ 46 5.2.10. Global matrices ________________________________________________________ 49

5.3. Systematic verification by EEA _______________________________________________ 51 5.3.1. Content of controls ______________________________________________________ 51 5.3.2. Results _______________________________________________________________ 52

5.4. Experts’ check and validation ________________________________________________ 52 5.4.1. Progress of the checks ___________________________________________________ 52 5.4.2. Results of the expert’ checks _______________________________________________ 53

5.4.2.1. Confusion matrices __________________________________________________ 53 5.4.2.2. QA-SIRS/ Experts’ validation results comparison ___________________________ 54 5.4.2.3. Main conclusions and experts’ remarks ___________________________________ 57

5.5. Land Cover areas statistics __________________________________________________ 59

6. Data delivered_________________________________________________________________ 60


European Commission




08/01/2013

Page 4/96

6.1. For each LUZ ______________________________________________________________ 60 6.1.1. Vector data ____________________________________________________________ 60 6.1.2. Delivery report __________________________________________________________ 65 6.1.3. Land Use/cover Maps ____________________________________________________ 66

6.2. LUZs Deliveries Report _____________________________________________________ 66

Conclusion _____________________________________________________________________ 67

Appendix 1 : List of LUZs produced _________________________________________________ 70

Appendix 2 : Results of confusion matrices by LUZ ___________________________________ 78

Appendix 3 : Land/Use Cover statistics ______________________________________________ 85

Area percentage by Country level 3 _______________________________________________ 85



Appendix 4 : LUZ Delivery Report - Hamburg example _________________________________ 88

Appendix 5 : LUZ Map - Manchester (UK) example _____________________________________ 95


European Commission




08/01/2013

Page 5/96

ILLUSTRATION TABLE Table 1 : Product accuracies__________________________________________________________13 Table 2 : Legend table______________________________________________________________ 15 Figure 1 : Application of MMU “2500 m²” in BRUXELLES center _____________________________ 17 Table 3 : « URBAN FABRIC » Typology ________________________________________________17 Table 4 : Images files received ( type and number)________________________________________19 Table 5 : Data type by LUZs__________________________________________________________20 Figure 2 : BRISTOL, First set of datas __________________________________________________22 Figure 3 : BRISTOL, Second set of datas________________________________________________23 Figure 4 : BRISTOL, Final set of datas__________________________________________________23 Figure 5 : WESTON UPON MARE Pansharpening ________________________________________24 Figure 6 : Kobenhaven, « Urban Fabric » existing and new soil sealing_________________________20 Figure 7 : Urban mask results: Copenhagen (DK) _________________________________________28 Figure 8 : Images used in the original Soil Sealing production________________________________29 Figure 9 : Scatter plot(a) and retrieval of original Sealing/NDVI relation ship(b) __________________32 Figure 10 : Kobenhaven, «Urban Fabric» classification with existing and new soil sealing __________34 Figure 11 : Stockholm, CAPI adjustments of LUZs limits ____________________________________35 Figure 12 : “LUZ Delivery report – Topologycal quality check” extract __________________________36 Figure 13 : CLC Urban mask creation___________________________________________________39 Figure 14: Urban/Non Urban sample ___________________________________________________40 Figure 15 : Example : control point of Göttingen___________________________________________42 Figure 16 : Example : control point of Cambridge__________________________________________43 Figure 17 : Example : control point of Lyon ______________________________________________42 Figure 18 : Experts Influence Areas ____________________________________________________45 Figure 19: Brussels matrices _________________________________________________________48 Figure 20 : Urban Atlas global matrices _________________________________________________50 Figure 21: Detail check list data _______________________________________________________51 Figure 22: Detail check list metadata ___________________________________________________52 Table 6: Expert matrices results _______________________________________________________54 Table 7 : QA-SIRS/Validation « Experts » results comparison________________________________55 Figure 23: Overall Accuracy __________________________________________________________56 Figure 24 : Rural Accuracy ___________________________________________________________56 Figure 25: Urban Accuracy __________ ________________________________________________57 Figure 26 : Land use/cover repartition types by countries on level 2 of the typology _______________59 Figure 27 : Land use/cover repartition types by countries on level 1 of the typology_______________60


European Commission




08/01/2013

Page 6/96

Document identification

Document release sheet

Project Urban Atlas

Title Final Report - Urban Atlas

Reference Final_Report_Urban_Atlas _v1.0_20110930

Version 2.0

Date 2011-11-22

Author SIRS – Lionel MEQUIGNON

Distribution European Commission, Enterprise and industry directorate general, SIRS

Approval network

Version Approval

2.0

Name : Jean-Paul GACHELIN C.E.O. SIRS. Date : 2011-11-22


European Commission




08/01/2013

Page 7/96

List of abreviations

ALOS Advanced Land Observation Satellite

AOI Area Of Interest

CAPI Computer Assist Photo-interpretation

CLC Corine Land Cover

CORINE COordination des INformations sur l’Environnement

EC European Commission

EEA European Environment Agency

EO Earth Observation

ESA European Space Agency

ESRI Environmental Systems Research Institute

EU European Union

FTS Fast Track Service

FTSP Fast Track Service Precursor

GIS Geographic Information System

GMES Global Monitoring for Environment and Security

IFEN Institut Français de l'Environnement

IGN-FI Institut Géographique National - France International

LUZ Large Urban Zone

MMU Minimal Mapping Unit

MMW Minimal Mapping Width

NDVI Normalized Difference Vegetation Index

QA Quality Assessment

SIRS Systèmes d'Information à Référence Spatiale

SL Sealing Layer

SPOT Système Pour l'Observation de la Terre

TB Terabyte

UA Urban Atlas

VHR Very High Resolution


European Commission




08/01/2013

Page 8/96

Executive summary

This document presents the final report on the production of the Urban Atlas. The aim is to present the progress made by the project, its background and the difficulties encountered. This document will also provide information which will allow the product to be improved for future updates. The main difficulties encountered and the solutions/ improvements implemented fall into three categories, detailed below.

- Technicals: Level of detail for taking roads into account: Should only the most important routes be considered or should a maximum number of routes be taken into account so as not to neglect the surface area covered by the road network? The second option was chosen. Complexity of the “routes and associated spaces” polygons: The significant geometric complexity of these objects required the polygons to be cut into a regular 10km sided grid for the largest LUZs. This complexity prevented the geometrical and topological controls from being completed (technological limits). Minimum Mapping Unit : The MMU planned for in the “Mapping Guide” had to be decreased. In fact, the consideration of the entire road network consequently resulted in a larger segmentation of the land use. Therefore, certain homogenous objects were cut across by the network of routes and divided into several distinct entities which were smaller than the MMU. In order to preserve them and so as not to introduce a bias into the map, it was decided that the MMU should be decreased to 500 m² instead of 2500 m² for the posts relating to artificial surfaces and to 10 000 m² for the other surfaces. Quality of the “Soil Sealing” data: The “Urban Fabric” posts were classified automatically using the Soil Sealing raster data. Problems were identified regarding this data being incomplete (pixels with a 0 value while the corresponding surface had buildings on it) at the beginning of the project. This data therefore had to be re-calculated using the interpreted “Urban Fabric” polygons.

- Thematicals: “Urban Fabric” land use posts. Habitat zones are mapped and classified according to the proportion of artificial surface within these zones. In the typology used at the start, the habitat zones where the proportion of the artificial surface was less than 10% were


European Commission




08/01/2013

Page 9/96

obscured. An additional post was therefore created in order to take them into account: “11240: Discontinuous Very Low Density Urban Fabric (S.L.: <10 %)”. Classification of Scrub and Forest Regeneration: These types of land use were initially classified in the “20000: Agricultural, Semi natural and Wetland areas” post. Following discussions, classification in the “30000: Forest” post was judged to be more suitable.

- Logistics:

Three main difficulties were encountered. They relate exclusively to the provision of satellite images. Availability of archive images: The production of the “Urban Atlas” planned for images acquired in 2006 (+/- 1 year) to be used. However, it turned out that this data was either not available or it was incomplete for some of the LUZs. The acquisition of new images therefore had to be programmed for 2009 and 2010. Unfortunately, this introduced heterogeneity in processing the LUZs in terms of the dates the images were taken. Image processing: As part of the contract, the plan was for the images to be delivered ‘ready for use’. For the SPOT data this aspect of the contract was complied with; however this was not the case for the ALOS data. With regard to the latter, production team had to take charge of fusing the panchromatic data with the multispectral data. This involved a significant additional investment in terms of time on the part of production team. Images quality (e.g. clouds, snow and gaps): The technical specifications detailed that images were acceptable if the presence of clouds, snow or the absence of data was below 5% of the total surface of the LUZ. In practice, images had to be rejected even when they satisfied this condition. In actuality, in the example of mapping urban environments, it was not acceptable to use images with cloud cover over urban areas. This ratio should only apply to rural or natural areas. The consequence was numerous exchanges, research into alternative images and therefore a significant amount of time spent. Lastly, all the difficulties encountered were resolved. Lessons should be taken from this experience so that future work of a similar nature, specifically the future update of the “Urban Atlas”, can be carried out under the best conditions.


European Commission




08/01/2013

Page 10/96

1. Background

1.1. Nature and purposes

The Urban Atlas contract is a service contract for the delivery of land use/cover maps of major European urban agglomerations. The work consists of the production of a series of land use/cover maps of larger urban zone (LUZ) as defined in the Urban Audit identified by a list of coordinate polygons. The work shall be based on satellite images of the reference year 2006+/-1 wich will be made available by the EC through ESA. The output of the work shall be in the form of GIS compatible vector maps of the urban agglomerations mentioned above.

The Urban Atlas is a joint project of the Regional Policy DG and the Enterprise and Industry DG. The project will complement the Urban Audit with harmonised data on land use and land cover. This project will provide a European overview of urban land use patterns and the possibility to track changes in the future, which so far has not been available. Currently, the only source of EU-wideland use or land cover information is CORINE which was primarily designed to measure natural areas and agricultural land. Due to the coarse resolution and high density thresholds, CORINE failed to capture large parts of urban areas. The land use maps will also be of interest to the cities concerned as they will be able to compare their land use patterns with that of other European cities. Furthermore, the Urban Atlas offers cities the opportunity to build more detail on top of these maps, the so-called "downstream services".

The Urban Audit (http://www.urbanaudit.org/) is a database allowing pan-European comparison of over 500 cities with a set of over 300 indicators across the domains Demography, Social Aspects, Economic Aspects, Civic Involvement, Training and Education, Environment, Travel end transport, In-formation Society, Culture and Leisure. The Urban Atlas aims to become part of a sustainable GMES Core Service that is continuously updated to find acceptance and ownership by local users, both in the public and private domain.

(http:/www.urbanaudit.org/)


European Commission




08/01/2013

Page 11/96

1.2. Related Documents

1.2.1. Input

Document ID Description

CALL FOR TENDERS N° ENTR/08/029 Technical specifications

ITD-0421-RP-0003-C5 C5 – Service Validation Protocole

ITD0-0421-GSELand-TN-01 Mapping Guide for a European Urban Atlas V1.1

Atlas_Urbain_v2.0_20080710 SIRS/IGN FI/EUROTOPO-IGC proposal

Offre_UA_V1.1-20090803 Verification and correction of the sealing layer before its usage in the production ofLarger Urban Zone Maps

Atlas_Urbain_contrat2_v1.1_20091214 Offre renouvellement UA V1.1. 20091214

QAP_V3.0_14-12-2009 QAP - SIRS/IGN FI/EUROTOPO IGC

1.2.2. Output

Document ID Description

UA_DR_2010-10-21 CE - Urban Atlas - Contract 1 - Final Luz Delivery Report - 21-10-2010 – 239 Luz

UA_DRM_2010-10-21 CE - Urban Atlas - Contract 1 - Final Luz Delivery Report -Maps 21-10-2010 - 239 Luz

UA_DR_2011-10-04 CE - Urban Atlas - Contract 2 - Final Luz Delivery Report - 04-10-2011 - 66 Luz

UA_DRM_2011-10-04 CE - Urban Atlas - Contract 2 - Final Luz Delivery Report -Maps 04-10-2011 - 66 Luz

UA_Final-Report_2011-11-22 Urban Atlas –Delivery of Land/Use maps of major European urban agglomerations_V 2.0

2. List of LUZs produced

The table of delivered LUZs is presented in appendix N°1. LUZs are classified by date of delivery.

305 LUZs have been delivered for a global area of 586 340 km² and 6 318 965 polygons created. 239 LUZs have been delivered according to the priority list of the first phase of the contract and 66 in the second phase. Nevertheless, for technical reasons related to the supply of satellite images, some LUZs first planned within contract 1 could not be produced. These LUZs have been replaced by LUZs planned within contract 2. These are LUZs of Rzeszow, Nowy Sacz, Derry and


European Commission




08/01/2013

Page 12/96

Opole, replaced respectively by Jelenia Gora, Koblenz, Perugia and Besançon. These replacements have been made so as to respect the first planned areas to be produced.

3. Urban Atlas Methodology (extract of Mapping guide v1.1)

3.1. Pre-processing and geometric adaptation of COTS navigation data

The EO data are the basis for interpretation. The geometrical differences between EO data and COTS navigation data, the COTS navigation data has to be corrected towards the EO data.

3.2. Pre-processing of Topographic Maps

Topographic maps are used for interpretation of objects. Topographic maps should be used in digital form with precise geo-coding. The usage of printed (analogue) maps is not recommended. In case of geometrical differences between EO data and topographic maps, the erroneous data (either RS-data or topo-maps) needs to be identified using reliable datasets providing spatial reference information. Then the geometry of the mapping product shall be congruent with the correct dataset.

3.3. Classification and Interpretation

Application of automatic classification routines, as segmentation and clustering be applied whenever appropriate: - Automated segmentation and classification to achieve a first differentiation into basic land cover classes (urban vs. forest vs. water vs. other land cover) is possible and upon decision of the service providers. - As backbone for the object geometry the COTS navigation data network is recommended but only with the method defined in the Annex of the Mapping guide.

3.4. Application of FTS Sealing layer

The FTS - sealing layer is used for classification of the sealing densities of class 1.1 Urban fabric in level 3 and level 4.


European Commission




08/01/2013

Page 13/96

3.5. Interpretation rules

– The delineation is to be done on the EO Data. EO data should be considered as primary (guiding) data source. – The interpretation of the object is done using:

– the EO Data, the topographic maps and COTS navigation data; – auxiliary information including local expertise.

– The interpreted area should be interpreted with at least 100m extension (100 buffer) to ensure accuracy and continuity of polygons. During the postprocessing phase a subset with the spatial extent of the final product will be generated. At the border of this subset (final product) polygons smaller than the MMU may be present. – In areas where two or more scenes overlap, it must be assured, that the most recent data are used for delineation and interpretation. – In case of cloud-coverage on the most recent scene the affected part (only this part!) shall be interpreted using a cloudless older scene. – If two or more objects overlapping at different levels, the top level is mapped continuously, e.g. Road Bridge over Railway is mapped as seen, the railway polygon is broken and the road is mapped as a contiguous feature. – In case of two or more objects overlapping on the same height level, the visually dominant and complete object (in use and shape) is mapped continuously. For example a road / railway crossing viewed at the same height level: the railway shall be mapped continuously to maintain the network. The road shall be broken.

Table 1: Product accuracies


European Commission




08/01/2013

Page 14/96

3.6. Minimum mapping unit

– Minimum mapping unit (MMU): - Class 1 : 0,25 ha - Class 2 – 5 : 1 ha

Exception of MMU 0.25 / 1ha : in case of an homogeneous area > MinMU but divided in 2 or more polygons by the road network, each part can be smaller to preserve the land cover information. However, no polygon can be smaller than 500 m² (e.g.: a 1 ha forest divided in 4 polygons by the road network has to be mapped).

– Minimum mapping width (MMW) between 2 Objects for distinct mapping: 10m – Maximum mapping width (MMW) between 2 Objects for mapping together: 10m Exception of minimum width 10 m of a mapping unit: to maintain continuity of linear structures, they can be mapped smaller than 10 m over a distance of up to 50m.

3.7. Priority rules

Priority mapping rules for areas smaller than the MMU:

– Smaller areas are added to the adjacent unit with the next lesser number of the same sub-class,

– Smaller areas are added to the adjacent unit of the same upper class, – Smaller areas are added to the adjacent unit with the longest common border line,

except to railways or roads (exception here: if an object is below the MMU size and completely surrounded by e.g. road or railway network it shall be aggregated with that surrounding traffic line).


European Commission




08/01/2013

Page 15/96

3.8. Legend Table

Table 2: Legend Table

1 Artificial surfaces

11 Urban Fabric

11100 Continuous Urban Fabric (S.L.: > 80%)

11210 Discontinuous Dense Urban Fabric (S.L.: 50% - 80%)

11220 Discontinuous Medium Density Urban Fabric (S.L.: 30% - 50)

11230 Discontinuous Low Density Urban Fabric (S.L.: 10% - 30%)

11240 Discontinuous Very Low Density Urban Fabric (S.L. < 10%)

11300 Isolated Structures

12 Industrial, commercial, public, military and privates units

12100 Industrial, commercial, public, military and private unit

12210 Fast transit roads and associated land

12220 Other roads and associated land

12230 Railways and associated land

12300 Port areas

12400 Airports

13 Mine, dump and constructions sites

13100 Mineral extraction and dump sites

13300 Construction sites

13400 Land without current use

14 Artificial non-agricultural vegetated areas

14100 Green urban areas

14200 Sports and leisure facilities

2 Agricultural + Semi-natural + Wetlands areas

3 Forest (natural and plantation)

5 Water (description of mapping units can be found in the mapping guide v1.1)

4. Encountered difficulties and solutions

4.1 Adjustments to the "Mapping Guide"

4.1.1. Interpretation rules

The objects were interpretated using :

- the Earth Observation data, - the Navigation data,


European Commission




08/01/2013

Page 16/96

- ancillary information. The interpretation rules are described in the Mapping Guide (Tender Specifications v1.1) However the Mapping Guide V1.02 had to be amended early on in of the project, following the production of the first five LUZs.

4.1.2. Minimal Mapping Unit (MMU)

The contractual Minimal Mapping Units of thematic extractions were 2500 m2 for nomenclature posts relating to artificial areas (classes 1…) and 10 000 m2 for the others (2…, 3…., 4….. and 5….).

As a result of the technical choice aimed at respecting the roads in their entirety, the problem of polygon split arose. Actually, in certain cases, the density of the network caused a split of the homogeneous thematic entity into several entities whose area was under the cartographic thresholds specified by the project. Compliance with these specifications thus required to union the entities below the thresholds to be combined with the included "road" entity. This would have introduced thematic anomalies within the mapping. In order to avoid these thematic skews as much as possible, the decision was required to lower the cartographic threshold. Relative to the results from the different tests, the minimum cartographic value kept was 500m2. Thus, no polygon (except borders) must be under this threshold in the maps produced. The following presentation on a BRUSSELS extract shows the result of a dissolving polygon under 2500 m².


European Commission




08/01/2013

Page 17/96

Figure 1 : Application of MMU “2500 m²” in BRUSSELS centre

The application of the rule of 2500 m² in the centre of Brussels induces the reclassification of multiple polygons "Urban Fabric" in "Roads and associated lands”. The mapping is biased

Mapping Guide updates : This change has been specified in the "Mapping Guide" by the addition to the section « 4.2.9. Minimum mapping units » of the following sentences:

“Exception of MinMU 0.25 / 1ha : in case of an homogeneous area > MinMU but divided in 2 or more polygons by the road network, each part can be smaller to preserve the land cover information. However, no polygon can be smaller than 500 m² (e.g.: a 1 ha forest divided in 4 polygons by the road network has to be mapped).”

4.1.3. Typology : post "Urban Fabric"

The typology relating to the "Urban fabric" posts, first decided upon at the start of the Urban Atlas project, was as follows:

Table 4 : « URBAN FABRIC » Typology 11100 Continuous Urban fabric (S.L. > 80 %)

11210 Discontinuous Dense Urban fabric (S.L.: 50 - 80 %)

11220 Discontinuous Medium Density Urban fabric (S.L.: 30 - 50 %)

11230 Discontinuous Low Density Urban fabric (S.L.: 10 - 30 %)

This classification is based on the proportion of artificial area compared to the vegetated areas within a group of residential housing. It is automatically calculated by crossing it with "soil sealing" data.


European Commission




08/01/2013

Page 18/96

This typology appeared to be incomprehensible as it did not take into consideration the urban fabric which has vegetated surface proportions of over 90%, but which cannot be classified as forests or agricultural areas.

It was thus decided to create an additional post aimed at characterising these areas with a very low urban fabric density: 11240 – Discontinuous Very Low Density Urban fabric (S.L.: < 10 %). Mapping Guide updates : The changes brought to the "Mapping Guide" consisted in

- Integrate the post “11240 – Discontinuous Very Low Density Urban fabric (S.L.: < 10 %)” in the section “4.4: Legend Table”,

- Update the section « 4.6: Description of mapping Units Urban atlas » by the

addition of the following text : “1.1.2.4 Discontinuous Very Low Density Urban Fabric

MMU: 0,25 ha, MMW: 10 m Average degree of soil sealing: < 10%

Predominant residential buildings, roads and other artificially surfaced areas. The vegetated areas are predominant, but the land is not dedicated to forestry or agriculture. Example: exclusive residential areas with large gardens”

4.1.4. Classification of scrubs and forestry regeneration.

Firstly, the "mapping guide" specification was to class scrubs and forestry regeneration in the post “20000: Agricultural, Semi-natural and Wetlands areas”. This choice, even if it can be justified, has been discussed by the European Experts in charge of the thematic quality control of the data produced. It was then decided that the post “30000: Forest” was preferable as it was more relevant. Mapping Guide updates : The changes brought to the "Mapping Guide" consisted in:


European Commission




08/01/2013

Page 19/96

- Remove in the section« 2 d), Shrubs and / or herbaceous Vegetation incl. transitional woodland », the sentence « Forest regeneration / re-colonisation: clear cuts, new forest plantations » and add it to the section « 3. Forest (natural and plantation) ».

4.2. Supply of images The main problems encountered during the work concerned the supply of satellite images. These problems were availability of archive images, images quality or images data format. The tables below present the number of files received and their distribution by data type. The problems encountered in terms of supply and the solutions found are then presented.

4.2.1. Amount of data and image processing

Table 4 : Image files received ( type and number)

Product type Number of files

PRODUCTS SPOT 5 PXS (2,5 meters) 608

PRODUCTS SPOT 5 P (2,5 meters) 127

PRODUCTS SPOT 5 XS (10 meters) 115

PRODUCTS ALOS P (2,5 meters) 350

PRODUCTS ALOS XS (10 meters) 167

PRODUCTS RAPIEYE XS (5 meters) 304

PRODUCTS QUICKBIRD (2,6 meters colour) 3

1674

At the end of the project, the total number of files received was 1674. These data represent a volume of more than 2TB.


European Commission




08/01/2013

Page 20/96

One of the problems was the necessity to merge panchromatic and multispectral images. In terms of the project, this operation had to be done first and images delivered to the producer should have been already merged. In practice, all the ALOS images and some of SPOT images have been delivered unmerged. Thus, the production Team had to assume this task. This work appeared to demand a lot of extra time as the volume of images was particularly important (242 files for SPOT and 517 files for ALOS). Recommandation: As for updates to come, this operation should be either actually done first or clearly identified as an additional task in the specifications.

4.2.2. Data types by LUZs

To ensure a full coverage of the 305 LUZs, a multi-sensor strategy had to be adopted. LUZs were mostly (73%) covered by SPOT 5 data (2.5 meters resolution). 13% of LUZs are fully covered by ALOS data. As for the rest of the LUZs, it was necessary to obtain heterogeneous data because the coverage was incomplete and clouds were present, sometimes both. In the case of these LUZs, problems were major and required more time.

Table 5 : Data type by LUZs

Data types Number of LUZs %

SPOT 5 222 72,79

ALOS 39 12,79

ALOS + SPOT 5 21 6,89

QuickBird 1 0,33

SPOT+RapidEye 16 5,25

ALOS+RapidEye 3 0,98

SPOT+ALOS+RapidEye 3 0,98

305


European Commission




08/01/2013

Page 21/96

4.2.3. Availability of archive images and Image quality

- Availability of archive images:

As for some LUZs planned in contract 1, the availability of archive images dated back to 2006 (+/-1 year), the reference year for mapping, was insufficient in terms of coverage. In order to find a solution, new image acquisitions were made in 2009 and 2010. This caused discrepancies in the homogeneity of the work over the whole LUZs in terms of reference dates. On the other hand and in spite of this, the LUZs first planned in contract 1 could not be supplied and some were replaced by LUZs planned in contract 2.

- Image quality:

The problem of image quality was principally linked to cloud coverage.

The contract specified that images where the cloud coverage did not exceed 5% of the LUZ surface would be acceptable. However, some images had to be rejected even if they complied with this threshold. Indeed, in many cases, clouds covered the urban areas making their interpretation impossible, which was not acceptable in a mapping project principally dedicated to identifying urban areas.

In some cases, the delivered images did not completely cover the LUZ area and thus could not be accepted. This concerned both the Panchromatic as well as Multispectral images. Moreover, images were mainly accepted or rejected through "quicklooks" (very under-sampled images). Yet, this type of pre-visualisation only allows an a priori quality judgement to be made. Only the analysis of images in true resolution allows a reliable decision to be made. This resulted in an increase of exchanges which of course then slowed down the project’s progress.

Example: BRISTOL LUZ case

The example of the City of Bristol is a typical case which demonstrates the difficulty in obtaining images due to a non-exhaustive coverage or presence of clouds. The following illustrations show the different steps which were required in order to acquire images showing complete cloud-less coverage.


European Commission




08/01/2013

Page 22/96

Figure 2: BRISTOL - first dataset (ALOS panchromatic acceptable quality but ALOS XS coverage incomplete and very

cloudy)


European Commission




08/01/2013

Page 23/96

Figure 3: BRISTOL -second dataset (SPOT 5 PXS Additional data, significant cloud coverage in the SW)

Figure 4: BRISTOL - final dataset

(Acquisition of RapidEye complete and cloud-less cover)


European Commission




08/01/2013

Page 24/96

The following illustration shows that in Weston-Super-Mare (Bristol LUZ) area, image processing has been used whenever possible. In this precise case, we could merge a panchromatic ALOS image (2.5 meters resolution) with a RapidEye image (5 meters resolution) so as to take advantage of both the radiometric richness of RapidEye (XS) and the geometric precision of panchromatic (P) ALOS.

Figure 5: WESTON-SUPER-MARE Pansharpening WESTON: first dataset WESTON: second dataset

WESTON: Pansharpening Alos Panchro + RapidEye

In consideration of the problem of the images acquired being incomplete and cloudy, the following alternative solutions were found (in order):

a) new research into the same satellite sensor data archives or recent acquisitions (SPOT 5 / ALOS); b) research into other satellite sensors (e.g. QuickBird, Formosat, etc.); c) research into the RapidEye data available;


European Commission




08/01/2013

Page 25/96

d) research into IMAGE 2006 CORINE LAND COVER. The RapidEye and image2006 resolution is not compliant with the technical specifications of the Urban Atlas. Thus, the following decisions were made: - RapidEye data can only be used in addition in the case of clouds, snow or gaps being present.

- Image 2006 CORINE can be used in addition in the case of clouds, snow or gaps being present only in rural or natural areas.

- Recommandations for future updates:

In the case of cloudy archive, the mitigation below could be proposed :

- Define a area of interest for each Luz, with preferably no cloud cover. This area of interest must include build-up areas. This could be automatically generated by the extraction of "build-up" polygons out of the existing and bufferized Urban Atlas so as to take in consideration the fact that urban mutations will be notably localized in immediate suburbs. This data would thus be delivered to the image supplier so as to help him in his research.

- Accept 5 or 10 meters resolution images outside the area of interest, to complete the Land/Use cover map on natural or agricultural areas.

- Accept that use multisensor images are necessary.

4.2.4. Image data format

In some cases, the cartographic projection system of delivered images was not compliant. This required the data to be re-processed.

Finally, the files delivered were sometimes corrupted and thus unusable (archive files ".zip") The coordination tasks involved in supplying the images were significantly increased because of an inadequate quality check of the images provided by the image supplier

4.3. Use of "Soil-Sealing"

4.3.1. Abstract

An initial Soil Sealing layer was produced in 2008 based on Image 2006 data for 38 countries including the EU 27. The production of this initial 2006 GMES Fast Track


European Commission




08/01/2013

Page 26/96

Service Precursor (FTSP) on Land Monitoring, specifically the degree of soil sealing, was based on the derivation of a built-up mask, as defined above, based on a hybrid ISODATA/Maximum Likelihood classification with manual corrections. The derivation of a sealing level within the built-up mask was based on NDVI derived from Image 2006 and visually interpreted calibration sites. This was used to derive a linear regression model of the degree of sealing ensuring that the highest NDVI value for unvegetated surfaces was set to 100 and the smallest NDVI value for an area fully covered by vegetation was set to 0. Therefore, the FTSP2006 degree of the sealing layer was based on vegetation activity within a land cover based built-up mask (i.e. an airport runway may be included, but not the grassland around it) excluding mines, quarries or construction sites that are unsealed and generally un-vegetated.

It was shown very early on in the project the fact that the FTSP2006 built-up mask was based on land cover as opposed to land use would pose a problem for the UA production. Production Team has made an evaluation of the sealing layer as an input to the Urban Atlas: 40 to 60% of the artificial areas as mapped in the Urban Atlas (UA) were not classified as artificial in the Sealing layer (SL). For an accurate characterization of the UA urban fabric classes, Production Team has to fill the missing SL data for the area covered by urban fabric polygons. The proposed production methodology to update the soil sealing layer in the Urban Fabric interpreted polygons makes it possible to retrieve the original NDVI/SL relationship thus ensuring full compatibility with the original SL layer.

4.3.2. Assessment of users requirements

Based on the Urban Atlas production experience, it was rapidly identified that errors of omission were present in the soil sealing layer. As a result, a quantitative assessment of the soil sealing layer as in input to the Urban Atlas was undertaken by Production Team. This detailed study provides a good understanding of omission / commission errors to be corrected.

4.3.2.1. Evaluation of the soils sealing layer as an input to the Urban Atlas

Production Team is committed to carry out the production of the Urban Atlas. This section describes the evaluation of the soils sealing layers as an input to the Urban Atlas.


European Commission




08/01/2013

Page 27/96

4.3.2.1.1. Objectives

The objectives were to assess the quality of the Soil sealing on the 5 first LUZs produced: Graz (AT), Copenhagen (DK), Patra, (GR), Miskolc (HU) and Rotterdam (NL), and more specifically:

Assess the potential of using the Soil Sealing layer to determine the Urban Atlas urban mask

Identify any issues related to the use of the Soil Sealing layer as an input to determining the urban fabric classes

4.3.2.1.2. Urban Mask Comparison: Soil Sealing and Urban Atlas

To support Computer Assisted Photo-Interpretation (CAPI), the soil sealing layer could be used to create the UA urban mask. However, for the first 5 LUZ, the UA urban mask was deliberately created independently.

The Soil Sealing layer was reclassified and the UA vector layer rasterised based on the urban vs urban codes

The two masks were compared based on a sample of 1000 points selected randomly with a minimum distance of 100m apart to avoid spatial autocorrelation

Results are presented in the tables below for Copenhagen (DK).


European Commission




08/01/2013

Page 28/96

Figure 6: Urban Mask Results: Copenhagen (DK)

Sample Sealing Urban atlas

Sample Urban AtlasSealing Natural

Artificial

Sample point

Urban Atlas

FTSP Natural Artificial Total User Commission

Natural 502 205 707 71,0% 29,0%

Artificial 6 287 293 98,0% 2,0%

Total 508 492Grand Total 1000

Producer 98,8% 58,3% Overall 78,9%

Omission 1,2% 41,7%

4.3.2.1.3. Initial conclusions

High level of omission error: 40 to 60% of the artificial areas as mapped in the Urban Atlas were not classified as artificial in the Soil Sealing layer,

This makes it unpractical to use the soil sealing to automatically extract the urban mask as part of the Urban Atlas production,


European Commission




08/01/2013

Page 29/96

4.3.2.2. Distribution of unfilled urban fabric polygons

Figure 7 : Distribution of unfilled Urban Fabric polygons: Copenhagen (DK)

4.3.2.3. Conclusions

From a land use point of view, artificial areas can include large amounts of vegetated areas. This mean that a significant proportion of the areas classified in the “Urban Fabric” category were not covered by the FTSP2006 Sealing layer, making the derivation of level 3 “Urban Fabric” classes impossible. Tests were carried out over the 5 first LUZs produced and brought the following conclusions:

• The soils sealing layer cannot be used to determine the Urban Mask (omission error from 40 to 60%). The settlement extraction for the Urban Atlas production has to be done independently,

• 30 to 50 % of urban fabric areas not covered by soil sealing, • Only 10 to 20 % of urban fabric polygons are fully covered, • This could result in a shift down to one class level in the UA, e.g. polygon labelled

as 11220 when in fact it should be 11210.

For an accurate classification of the urban fabric classes, Production Team has to fill the gaps of the soil sealing layer on the areas covered by urban fabric polygons.


European Commission




08/01/2013

Page 30/96

As a result of these observations, the following workarounds were:

- Continue to characterise the urban mask independently from the soil sealing layer; - Assume that 0 in the soil sealing layer corresponds to a 0 sealing level (as

done for the first 5 LUZ); - Include a new UA field indicating the level of soil sealing layer coverage for

each polygon; - Alternatively, calculate the degree of sealing for the areas classified as “Urban

Fabric” as part of the UA production based on Image2006 used to produce the initial soil sealing.

This last solution was selected and it was thus necessary to re-calculate the soil sealing in the urban mask provided by the interpretation.

4.3.3. Detailed description of the methodological approach

4.3.3.1. NDVI calculation

In the existing sealing layer, it is understood that the sealing level was based on the Normalised Difference Vegetation Index (NDVI) combined with an accurate delineation of the built-up area. The relationship NDVI / sealing level is linear.

NDVI = (XS3-XS2) / (XS3+XS2)

It has been demonstrated in numerous past studies that the NDVI can be used as a reliable indicator of soil sealing as it assumes that any soil that is not covered by vegetation in built-up is likely to be sealed.

4.3.3.2. Image processing methodology

The image processing methodology was tested on the Copenhagen LUZ (below).

4.3.3.2.1. Sealing calculation

The Urban Atlas vector file was used to create an Urban Mask making sure that the sealing level is derived for all artificial areas and avoid commission errors The soil sealing metadata vector file on image deviations (see figure 8 below) was used to identify the images used for the original production of the sealing layer. Care was taken to identify the right images to maximize the level of compatibility.


European Commission




08/01/2013

Page 31/96

NDVI images were computed from raw images as a basis for the sealing level extraction.

Figure 8: Images used in the original Soil Sealing production

A systematic grid of points was then used to extract pixel values from both the NDVI images and the original sealing layer. NDVI and sealing observations were paired for each image to identify the images actually used in the sealing level calculation. A scatter plot was produced for each NDVI image as shown in Figure 9a below. There is a considerable scatter of points as the whole image was used to ensure no data was left out. However, the linear relationship between the sealing level and the NDVI can be clearly seen. This was used to retrieve the linear regression model used in the original soil sealing layer production as shown as Figure 9b.


European Commission




08/01/2013

Page 32/96

Figure 9: Scatter plot (a) and retrieval of original Sealing/NDVI relationship (b)

(a)

(b)

The linear regression model retrieved makes it then possible to recalculate the sealing level for the NDVI images used to calculate the original sealing layer and expand it to artificial areas omitted from the original sealing layer.


European Commission




08/01/2013

Page 33/96

4.3.3.2.2. Updating of the existing sealing layer

Only the pixels recalculated with new values are corrected in the existing national mosaics:

- Original pixel value = 0 or pixel is selected as CLC 1.3.x classes - New pixel value = 1-100 - A conditional statement based algorithm is used to integrate the new

pixels from the processed scene into the national mosaic. This will ensure that consistency problems across images are kept to a minimum.

4.3.3.3. Production flow

4.3.3.3.1. Images : download and NDVI processing

The image 2006 layer has to be downloaded in our Urban Atlas production chain to integrate new imperviousness calculation on the urban fabric polygons. NDVI are simultaneously calculated on the images.

4.3.3.3.2. Processing of the LUZ

The production is organized on a LUZ per LUZ basis.

4.3.4. Results

The following presentation shows the results on a Copenhagen LUZ extract.


European Commission




08/01/2013

Page 34/96

Figure 10: Copenhagen - existing “Urban Fabric” and new soil sealing Existing Soil Sealing New soil sealing

Figure 7: Copenhagen - ”Urban Fabric” classification with existing and new soil sealing

The dark red colour corresponds to the zones where the artificial surface reaches its peak; the orange and yellow correspond to the low-artificial zones. We can easily see the changes of classes after soil sealing re-calculation.


European Commission




08/01/2013

Page 35/96

4.3.5. Recommendations for further UA updates

In any case, the issue regarding the definition of land cover versus land use of built-up areas remains and it is strongly advisable that the sealing layer should be recalculated following the methodology developed in future updates of the Urban Atlas for the area classified as urban fabric to ensure that there is no gap in the coverage. The method is now fully automated and provides consistent results which are compatible with the original sealing layer data, as the same regression relationships are used.

4.4. LUZs limits

In some cases, LUZs geometric limits were particularly inaccurate. This was certainly due to an inadequate choice of scale during the capture. It was thus necessary to adjust these limits in CAPI. The following image gives an example of limit adjustments made to the Stockholm LUZ (yellow: the original limits; green: adjusted limits)

Figure 11 : Stockholm, CAPI adjustments of LUZs limits

The new LUZ limits are available as a shape file in each Geodatabase (see 6.1.1. section page 60). In future updates, these files of new contours to be used.


European Commission




08/01/2013

Page 36/96

4.5. Road splitting

However, in some cases, especially in huge LUZ, the geometric check and arcgis’s topology failed. The main cause was the size and shape of a very large and complex polygon composed of roads (code 12220). So, to avoid this problem, the Production Team created a 10 km size grid for those LUZs, which split this huge polygon into smaller ones. Only the 12220 code roads are split using this method. After this operation, the geometric and topology checks passed successfully. When roads are split on an LUZ this is specified in the "LUZ Delivery Report" under "Topological QUALITY CHECK" section.

Figure 12 “LUZ Delivery report – Topological quality check” extract Internal quality

check Date of control

(y/m/d) Result Remark (errors, corrections, etc.)

SIRS

11/10/04

ok

Roads had been split

4.6. Recommandations for future use and updates

4.6.1. Mapping guide

The main adjustments required were made at the completion of the Urban Atlas. These have been identified early in the project. In the next update, it is recomanded to implement an operational test phase of production. During this testing phase the necessary other changes will eventually be identified.

4.6.2. LUZs limits

During the production some imperfections on the limits of LUZs were identified. They were corrected especially if some surfaces were forgotten (especially on the coasts). The new limits LUZs are provided with the data produced. The scale of creation of these contours seems inappropriately and too imprecise. It may be helpful to correct them exhaustively to correspond better to the administrative reality, bio-physical, ... of the territories.


European Commission




08/01/2013

Page 37/96

4.6.3. Soil Sealing

It was shown very early on in the project that the FTSP2006 built-up mask was based on land cover as opposed to land use would pose a problem for the UA production. The production team has made an evaluation of the sealing layer as an input to the Urban Atlas: 40 to 60% of the artificial areas as mapped in the Urban Atlas (UA) were not classified as artificial in the Sealing layer (SL). In any case, the issue regarding the definition of land cover versus land use of built-up areas remains and it is strongly advisable that the sealing layer should be recalculated following the methodology developed in future updates of the Urban Atlas for the area classified as urban fabric to ensure that there is no gap in the coverage. The method is now fully automated and provides consistent results which are compatible with the original sealing layer data, as the same regression relationships are used.

4.6.4. Image management

The main problems encountered during the work concerned the supply of satellite images. These problems were : availability of archive images, images quality or images data format. In the case of cloudy archive, the mitigation below could be proposed :

- Define an area of interest for each Luz, with preferably no cloud cover. This area of interest must include build-up areas. This could be automatically generated by the extraction of "build-up" polygons out of the existing and bufferized Urban Atlas so as to take in consideration the fact that urban mutations will be notably localized in immediate suburbs. This data would thus be delivered to the image supplier so as to help him in his research.

- Accept 5 or 10 meters resolution images outside the area of interest, to complete the Land/Use cover map on natural or agricultural areas.

- Accept that use multisensor images are necessary.

5. Quality control

Four different levels of quality controls have been carried before delivery of the products to the user:

- Internal quality control during production, - Independent thematic validation before delivery,


European Commission




08/01/2013

Page 38/96

- Systematic vérification by EEA, - Sample thematic validation by external experts.

5.1. Internal quality controls

5.1.1. Attribute table controls

The consistency of attributes tables is automatically checked.

5.1.2. Geometric control

The database geometric precision has been checked at regular intervals before each delivery.

5.1.3. Continuous control

Technical details and the accuracy of the image capture are described in the Mapping Guide and its appendix. The person in charge of quality checked the accuracy of the capture. The sectorization must respect the specifications. Whenever a gap was notified, the operator had to re-do the work.

5.1.4. Photo-interpreters’ cross-controls

The visual assessment consisted of a "cross control". At regular intervals and after a significant are had been processed, each operator submitted his work to another for control.

In the end, this process ensured high quality levels and a homogeneous analysis. It also prevented results differing from one operator to another.

5.2. External quality controls

External quality control was carried-out by IGN-FI. It consisted of a visual interpretation of a sample of points.

5.2.1. Pre-defined grid sampling

The method used was a random oriented sampling according to a European grid and a variable sample step, depending on the LUZ size. The sampling was obtained as followed:

- The "European" grid is the one the European Environment Agency (EEA) use for the assessment of landscapes and ecosystems. This grid is made up of a geographical


European Commission




08/01/2013

Page 39/96

reference, orthogonal references with regular areas of one square kilometer and is refined according to the LUZ size which intersects it (whole, half, quarter) so as to control +/-10 of produced area.

- A first classification in (U-UN) Urban-Non Urban is established from FTS data and CORINE Land cover level 1 - class 1 for Urban, and the grouping of the other classes level 1 for Non-Urban.

Figure 13 : CLC Urban mask creation

LUZ Lyon CORINE Land Cover 2000 (Level 1) CLC Urban mask

This first classification is interesting as it determines the choice of a larger number of samples in urbanized areas than in non-urbanized areas.

By crossing the European grid above defined with Urban-Non Urban classification, it is suggested that 80% samples in the Urban zone and 20% in the Non-Urban zone (0-25% urban) are chosen.


European Commission




08/01/2013

Page 40/96

Figure 14 : Urban/Non Urban sample

Lyon LUZ Classification Urban – Non Urban + European Grid

CLC Urban mask + European Grid + Sample (blue)

It is proposed to consider the threshold of 33% urban per cell, the urban percentage in the meaning of CLC to define a cell as “urban cells”. The remaining cells will be considered by rural cells.

The samples are set by the experts responsible for evaluating the quality and are not communicated to the team in charge of the production. That way, the independence of the method is guaranteed. In the overall LUZ and using the previous hypothesis (33% urban per cell), approximately 72,000 cells are defined as "urban cells" of a total of 636,000 cells intersected by LUZ and the kilometer grid. Then we obtain a ratio of 11.3% of urban cells but very close to 13% corresponding to the class 1 in CLC2000 in Europe.

5.2.2. Determining the number of control points

Based on the distribution of control points required (aim at 80% of the control points for the urban cells and 20% for rural cells), it is possible to determine the number of control points to consider only the ratio determined above, that is to say11.3% of urban cells (≈ 1 urban cells to 8 rural cells on average). The LUZ have been classified into 3 categories according to the percentage of urban cells:


European Commission




08/01/2013

Page 41/96

LUZ Percentage of « urbain cells» / LUZ

Number of LUZ concerned

weakly urbanized between 0 to 6% 81

moderately urbanized between 6 to 17% 146

Highly urbanized beyond 17% 78

From there, the average ratio of 1 urban cell to 8 rural cells can be adapted

LUZ adapted Ratio

To 10 rural cells Per urban cell

weakly urbanized 1/100 1 point 10 points

moderately urbanized 1/30 1 point 3 points

Highly urbanized 1/20 1 point 2 points

Whether

LUZ weakly urbanized: 1 point every 12 « rural cells », 8 points by « urban cells »

LUZ moderately urbanized: 1 point every 12 « rural cells », 3 points by « urban cells »

LUZ Highly urbanized: 1 point every 10 « rural cells », 1,5 points by « urban cells ».

5.2.3. Determining the distribution of control points (Secondary sampling Unit)

A second grid level has been defined with a smaller size 200m * 200m.

The control points are considered as the centroids of the cells of level 2. For each of the urban categories specified in the LUZ (weakly, moderately or highly) the proposed distribution is shown in the following diagrams:

200m

Cell

1 km x1 km


European Commission




08/01/2013

Page 42/96

LUZ weakly urbanized:

Figure 15: Example: Control point of Göttingen (red color: Urban point / Green color: Rural point)

Urban Cells: Rural cells :

every 10 cells of level 1


European Commission




08/01/2013

Page 43/96

LUZ moderately urbanized :

Figure 16: Example: Control point of Cambridge (red color: Urban point / Green color: Rural point)

Urban cells: Rural cells :



European Commission




08/01/2013

Page 44/96

LUZ highly urbanized :

Figure 17: Example: Control point of Lyon (red color: Urban point / Green color: Rural point)

Urban cells: Rural cells :


5.2.4. Blind test by three European experts

Whenever the Urban Atlas is completed for one LUZ, samples are set and three experts share samples of produced data. They operate over geographic zones where they have significant experience.


European Commission




08/01/2013

Page 45/96

That way, the three appointed experts share their knowledge of land use across Europe. The quality of each LUZ would then be assessed by at least two experts.

Figure 18 : Experts Influence Areas

Expert Influence Area Quality evaluation 1 Expert Influence Area Quality evaluation 2 Expert Influence Area Quality evaluation 3

5.2.5. Quality improvement

The quality is estimated for each LUZ. Metadata over quality are generated and a return to production for corrections is made if the level of quality does not answer the specifications. Each class which quality is below the required thresholds is being verified by the person in charge for the thematic quality production team and the data is subjected again to external quality control. 58 of 305 LUZs have required corrections on some classes after quality control. In practice, no LUZ needed a second correction procedure.


European Commission




08/01/2013

Page 46/96

5.2.6. Advantages of the method

The quality assessment is independently carried out and is not subject to any interference from production teams. It is thus an external assessment which approves/rejects the data produced. This method guarantees the respect of the criteria enacted by referenced terms.

The grid used by the EEA comes from the sharing of existing tools which have been tried-and-tested in control processes and in the validation of the European land use database. This grid provides a unique identifier and thus makes it easy to track samples in accordance with EEA standards.

To sum up, this method is easily applicable and highly automated in terms of sample generation and the quality assessment by independent experts who have regional experience and can ensure the criteria are met.

5.2.7. Intermediate delivery control

Whenever an intermediate delivery was about to be made, a complete structural and attributary quality control of the data was carried out.

5.2.8. Final delivery control

Quality indicators are:

• intrinsic quality of data (e.g. topology, structuration, etc.); • exhaustive data (i.e. no missed areas); • thematic (i.e. the analysis and interpretation are ok). 5.2.9. Quantifying accuracy : the confusion matrix

According to the specifications, the statistical accuracy of the database was assessed through the calculation of a confusion matrix relating to the quality control points created by European experts from IGN-FI.

The size of this sample was decided upon in collaboration with the client.

Each cell of this dual-entry chart gives the classified number of points with the:

• line: the point attributed by the interpreter when mapping;


European Commission




08/01/2013

Page 47/96

• column: the attribution as it was recognised using exogenous documents (i.e. aerial photographs, site visits, other databases, etc.) during the checks.

The marginal and diagonal totals of this matrix are used to assess two types of accuracy, commonly called ‘user accuracy’ and ‘producer accuracy’. Measurements are given by attribution categories. As for each of the LUZ produced, three confusion matrixes have been calculated. The first one focused on urbanized areas, the second on non urbanized areas and the third matrix gives an overall accuracy. As an example, please see the matrix relating to the Brussels LUZ below.


European Commission




08/01/2013

Page 48/96

Figure 19 : Brussels matrices

Urban confusion matrix (Brussels)

Rural confusion matrix (Brussels)

Overall confusion matrix (Brussels)


European Commission




08/01/2013

Page 49/96

On no account could the global accuracy of each matrix be under 85%. Otherwise, the quality controller in charge of the LUZ specified the thematic problems to the operator who then had to correct the thematic discrepancies over the whole LUZ. In the end, the matrices were recalculated until the LUZ became compliant.

5.2.10. Global matrices

The global matrices calculating all of the 305 delivered LUZs are presented below. In addition, a summary chart of the results obtained for each LUZ is available in the appendix.


European Commission




08/01/2013

Page 50/96

Figure 20 : Urban Atlas global matrices

Urban global confusion matrix

Rural global confusion matrix

Overall global confusion matrix


European Commission




08/01/2013

Page 51/96

5.3. Systematic verification by EEA

5.3.1. Content of controls

Each delivered LUZ was systematically controlled by the EEA and the results were transcribed through a report. This Database Technical Acceptance (DBTA) report summarizes the results of the technical acceptance procedure done for a final Urban Atlas database according to the standard check list defined. Technical quality has been checked for data and metadata provided in the delivery. If errors are detected, the report is transmitted to the producer who corrects and makes a new delivery. The following figures shows the various controls.

Figure 21: Detail check list - Data


European Commission




08/01/2013

Page 52/96

Figure 22: Detail check list - Metadata

5.3.2. Results

Generally, the percentage of LUZs for which the control showed errors is low (- 5%). Most of them concerned LUZs delivered earlier in the project. The main reason is that initially the LUZs were delivered to the shape file format. When deliveries have been made in Geodatabase, format releases for non-compliance were exceptional. In addition, a rigorous topological and structural quality control procedure has been established (see Appendix 4 page 94).

5.4. Experts’ check and validation

In the first and second contracts, a validation session was organised by the European Commission.

5.4.1. Progress of the checks

The control took place in the producer premises in October 2009 and February 2011 and was globally done according to the same procedure as the external control described above.


European Commission




08/01/2013

Page 53/96

Eight approved experts in Land Cover Mapping were appointed by the client (the European Commission) and the 21 following LUZs were assessed, for a total area of around 74000 km2 (13 % of the total LUZ area):

- Lyon (FR) - Kaunas (LT) - Iraklion (GR) - Warzawa (PL) - Lisbon (PT) - Tampere (FI) - Bruxelles (BE) - Craiova (RO) - Kopenhavn (DK) - Valetta (MT) - Berlin (DE) - London (UK) - Tartu (EE) - Burgas (BG) - Gyor(HG) - Roma (IT) - Wien (AT) - Zilina (SK) - Madrid (ES) - Arnhem (NL) - Linkoping (SE)

For each LUZ, a grid of points was generated in accordance with the same principle adopted for the external quality controls made up by IGN-FI. The only difference was that the controls had to be carried out using two procedures: a blind control (the experts did not have the producer’s interpretation) and a non-blind control (the experts had the producer’s interpretation). As a result of these control procedures, the confusion matrices were calculated in accordance with the same principle as described above (3 matrices per LUZ).

5.4.2. Results of the expert’ checks

5.4.2.1. Confusion matrices

The overall accuracies resulting from the experts’ controls are presented below.


European Commission




08/01/2013

Page 54/96

We can see that overall the results meet the specifications of the "validation protocol".

Table 6: Expert matrices results (blind re-interpretation)

LU

Z

Sam

plin

g po

ints

Ove

rall

accu

racy

Rur

al a

ccur

acy

Urb

an a

ccur

acy

Bruxelles 802 90,77% 94,76% 91,27%

Kobenhavn 816 86,76% 95,71% 87,01%

Valetta 805 85,96% 94,78% 85,96%

Warszawa 815 91,29% 95,21% 92,39%

Lisboa 760 90,92% 96,45% 92,76%

Kaunas 762 92,13% 96,06% 93,18%

Tempere 772 91,32% 95,21% 92,62%

Lyon 730 93,97% 96,99% 94,79%

Iraklion 830 90,24% 96,75% 90,36%

Craiova 834 89,57% 94,96% 90,29%

Berlin 1597 91,04% 87,41% 87,41%

Burgas 800 98,13% 99,79% 95,54%

Gyor 800 98,88% 98,43% 99,43%

London 1539 94,20% 93,40% 95,00%

Madrid 1580 90,00% 86,65% 94,09%

Roma 800 97,88% 97,76% 97,99%

Tartu 800 96,75% 98,20% 94,35%

Wien 800 97,10% 96,65% 97,67%

Zilina 800 94,92% 95,04% 94,75%

Arnhem 800 92,98% 92,25% 93,75%

Linkoping 800 93,64% 91,52% 96,78%

5.4.2.2. QA-SIRS/ Experts’ validation results comparison

The following chart and illustration shows a comparison of the results from both of the quality controls carried out on data of the 21 produced LUZs.


European Commission




08/01/2013

Page 55/96

Table 7: QA-SIRS/Experts validation results comparison (blind re-interpretation)

LU

Z

Ove

rall

accu

racy

"Val

idat

ion"

Ove

rall

accu

racy

"SIR

S in

tern

al Q

A"

Rur

al a

ccur

acy

"Val

idat

ion"

Rur

al a

ccur

acy

"SIR

S in

tern

al Q

A"

Urb

an a

ccur

acy

"Val

idat

ion"

Urb

an a

ccur

acy

"SIR

S in

tern

al Q

A"

Bruxelles 90,77% 88,40% 94,76% 93,10% 91,27% 89,40%

Kobenhavn 86,76% 95,70% 95,71% 98,10% 87,01% 95,90%

Valetta 85,96% 91,10% 94,78% 96,40% 85,96% 91,10%

Warszawa 91,29% 85,20% 95,21% 88,90% 92,39% 85,40%

Lisboa 90,92% 85,20% 96,45% 91,30% 92,76% 85,50%

Kaunas 92,13% 85,00% 96,06% 90,60% 93,18% 85,00%

Tempere 91,32% 87,50% 95,21% 93,80% 92,62% 88,50%

Lyon 93,97% 91,70% 96,99% 94,60% 94,79% 93,90%

Iraklion 90,24% 96,30% 96,75% 98,10% 90,36% 96,30%

Craiova 89,57% 85,30% 94,96% 89,30% 90,29% 85,30%

Berlin 91,04% 87,00% 87,41% 95,00% 87,41% 86,20%

Burgas 98,13% 87,50% 99,79% 89,10% 95,54% 88,50%

Gyor 98,88% 86,00% 98,43% 86,50% 99,43% 88,60%

London 94,20% 87,50% 93,40% 92,90% 95,00% 86,90%

Madrid 90,00% 85,40% 86,65% 91,70% 94,09% 86,00%

Roma 97,88% 85,30% 97,76% 91,90% 97,99% 86,60%

Tartu 96,75% 91,50% 98,20% 91,70% 94,35% 86,90%

Wien 97,10% 87,70% 96,65% 95,00% 97,67% 88,00%

Zilina 94,92% 90,20% 95,04% 94,00% 94,75% 92,10%

Arnhem 92,98% 90,50% 92,25% 95,80% 93,75% 90,50%

Linkoping 93,64% 91,10% 91,52% 95,20% 96,78% 93,40%


European Commission




08/01/2013

Page 56/96

Figure 23 : Overall Accuracy

Figure 24: Rural Accuracy


European Commission




08/01/2013

Page 57/96

Figure 25:Urban Accuracy

The compared results of both controls (QA-SIRS and “expert validation”) confirm the legitimacy of the internal QA/QC results made up by the IGN-FI controllers throughout the LUZ production.

5.4.2.3. Main conclusions and experts’ remarks

Although the results of the experts’ controls appear to be satisfactory and the work carried out by the producer has been qualified as "good", several remarks aimed at further improving the quality of the data produced have been expressed:

- In the case of scrubs and forestry regeneration: as mentioned in paragraph "3.2.3", the "mapping guide" first specified the classification of scrubs and forestry regeneration into the post “20000: Agricultural, Semi-natural and Wetland areas”. This choice, even if justifiable, was discussed by the European experts who decided to that the post “30000: Forest” was preferable as it was more relevant.

- Density of road network: The choice to represent the whole of the road network sometimes led to an over density which may make the reading of the map uncomfortable, particularly for the city centre areas. In rural areas, the problem lies in the representation of private lanes and agricultural lanes as roads when they are not.


European Commission




08/01/2013

Page 58/96

However, this exhaustive representation also ensures that an exaggerated bias is not introduced in terms of under-estimating the artificial areas where the networks belong (an important aspect within the Urban Atlas). On the other hand, a choice orientated towards an under-representation could also introduce a lack in terms of the presence of substructures according to the different types of landscapes existing within the European Union. The decision taken at the beginning of the project was not discussed. However, as the production went on, photo interpreters tried as far as possible to remove lanes in the landscape that were not considered to be roads.

- The case of the "Urban Fabric" post: During their validation session, the experts did not have access to the data re-generated after the soil sealing corrections. They were therefore unable to consider the different levels of density of the residential urbanised areas. A remark has been made concerning the presence of distinct density within globalized polygons. This remark has been taken into consideration and it was decided to implement an extra phase of re-cutting the polygons encoded in "11" before processing using soil sealing in accordance with the contained densities.

- Green urban areas and forest: the distinction of forest (i.e. class 30000) from green urban areas proved to be difficult, e.g. when wood areas are rather large and located within the scattered urban fabric or the forest area obviously contains recreational elements, but is rather large and not fully surrounded by urban fabric. Unfortunately, this classification also depends on the subjectivity of the photo-interpretation. In order to ensure homogeneous processing, wooded areas in urbanised and sub-urbanised areas were systematically checked by the person in charge of the quality control before completion and data delivery.


European Commission




08/01/2013

Page 59/96

5.5. Land Cover areas statistics

Statistics on the surfaces of each type of land use (level “detail”, level 2 and 1) are collated in appendix 3. They show the major trends in the distribution of land cover classification for each class for each country and highlight the significant differences between different countries.

Figure 26: Types of land use/cover distribution by country on level 2 of the typology

0%

20%

40%

60%

80%

100%

BE MT NL PT UK HU FR IT RO DK DE GR AT IE FI PL SL CZ LU CY SK LT ES BG SE EE LV

50

30

20

14

13

12

11


European Commission




08/01/2013

Page 60/96

Figure 27: Types of land use/cover distribution by country on level 1 of the typology

0%

20%

40%

60%

80%

100%

BE NL UK FR DE DK AT FI CZ PL BG LT SE EE

5

3

2

1

6. Data delivered

At the end of the project the following data were delivered:

6.1. For each LUZ

6.1.1. Vector data

Data were delivered in ESRI® Geodatabase v9.3 format.

The data were delivered in advance processed by the cell quality and GIS systems to ensure full consistency and structural topology.


European Commission




08/01/2013

Page 61/96

Data structure. Geodatabase naming is standardised.

In this example, the geodatabase is named “de002l_Hamburg.gdb”. • Both characters indicate the country (e.g. “DE”); • The four characters below show the serial number of the LUZ in the country (e.g.

“002l”); • A variable number of characters preceded by an underscore to the name of the city

(e.g. “Koln”). In addition, specific characters are avoided (e.g. “Koln” and not “Köln”).

> Within the geodatabase are:


European Commission




08/01/2013

Page 62/96

The file "check geometry", testing inconsistencies in the geometry (an empty file by definition). A set of geodatabase feature classes with the English name of LUZ. The set of feature classes.

Feature classes are located in the following set of classes of entities:

> The « contour » class:

This class contains the outline of the LUZ.


European Commission




08/01/2013

Page 63/96

The attributes of this class are:

· CITIES: Attribute type text, 254 characters. LUZ Name (e.g. Hamburg);

· LUZ_OR_CIT: Attribute type text, 254 characters of the ID LUZ (e.g. DE002L).

The XXYYYY_city class (uses the same naming rules as those established for the naming of the geodatabase).



· LUZ_OR_CIT: Attribute type text, 254 characters of the ID LUZ (e.g. DE002L);

· Code: Type attribute text, 7 characters. Code identifying the land (e.g. 50000);

· ITEM: Type attribute text: 150 characters. Land in the clear (e.g. bodies of water);

· PROD_DATE: text attributes: 4 letters: Year the data was produced.


European Commission




08/01/2013

Page 64/96

> The “routes” class includes the polygons of roads.



· LUZ_OR_CIT: Attribute type text, 254 characters of the ID LUZ (e.g. DE002L);

· Code: Type attribute text, 7 characters. Code identifying the land (e.g. 12220);

· ITEM: Type attribute text: 150 characters. Land in the clear (i.e. other roads and

associated land)’

· PROD_DATE: text attributes: 4 letters: Year the data was produced.


European Commission




08/01/2013

Page 65/96

> Metadata Metadata were produced using the EEA’s "Metadata editor". The standard used is the EEA-MSGI/ISO19115.

> The reference system is WGS84 UTM zone (e.g. WGS_1984_UTM_Zone_32N)

6.1.2. Delivery report

Using the Geodatabase a LUZ Delivery Report was delivered. This document is an external metadata and contains any relevant information about the production of the LUZ (e.g. projection, images used, photointerpreters’ remarks, quality control results, etc.). An example of the metadata file delivered for Hamburg (DE) is presented in the appendix 4 (page 88).


European Commission




08/01/2013

Page 66/96

6.1.3. Land Use/cover Maps

For each LUZ, an A3 map was produced. The files are in "PDF" format. An example of the LUZ of Manchester (UK) is presented in the appendix 5 (page 95).

6.2. LUZs Deliveries Report

The "Final LUZ Deliveries Report" is a compilation of reports on the production of each LUZ (Delivery Report). A document is delivered for each of the two contracts.


European Commission




08/01/2013

Page 67/96

Conclusion

Overall, the project had proceeded satisfactorily. The feedback received from the users seems to show real satisfaction from the point of view of the usefulness of the data and from the point of view of its quality. Early on in the project and following the test production of 5 LUZs, it was considered as necessary to adjust some of the methodological points. These are already described in Chapter 3 of this report. In summary, these adjustments concerned: - adapting the "Mapping Guide" (MMU, "Urban Fabric" typology post, classification of

Scrubs and Forestry regenerations); - the need to complete the FTSP2006 which presented gaps and made it impossible

for a correct classification of the "Urban Fabric" items; - the need to correct the contours of some LUZs; - the requirement to cut routes because of the complexity of the polygons for the

largest LUZs. Once the necessary adjustments were made, the production did not meet with any other major technical problems except for the supply of images (see below). Based on the experience gained in producing Urban Atlas 2006, the following thematic points are suggested from the producer point of view to be considered in a view of a further effort aimed at updating the 2006 database: 1. Better compatibility between Urban Atlas and Corine Land Cover Considering the current definition of the classification schemes, UA and CLC are suitably compatible in the urban zones. However, such compatibility is no longer assured in the rural areas of the LUZs. In fact, in UA the 1ha classes are represented as Agriculture (2000, including wetlands and semi-natural areas), Forest (3000) and Water (5000) whereas CLC classifies wetlands in a specific separate class (4000) while the semi-natural areas are included in the class Forest (3000) instead of Agriculture (2000) as is the current case for UA. In order to assure a more encompassing compatibility between the two datasets, it is therefore suggested to slightly modify the current Urban Atlas classification scheme in the rural areas (1ha classes) according to the CLC classification scheme.That would imply at the start of the Urban Atlas 2012 campaign the requirement to consider some additional processing time necessary to update UA 2006 according to the new classes. 2. Soil Sealing Layer integration in Urban Atlas (Urban Fabric classes)


European Commission




08/01/2013

Page 68/96

The Soil Sealing Layer 2012, to be produced by the recently announced GIO HRL effort, and used for updating the Urban Fabric classes in Urban Atlas 2012, will require (as it was the case for Urban Atlas 2006), some additional processing time so that to make it compatible to UA due to some inconsistencies between the two datasets. As explained earlier in this report, these inconsistencies are due to the fact that those datasets are based on a different spatial structure (spatial resolution) but also on a different conceptual framework with the Soil Sealing Layer (or Imperviousness Layer) being a classical land cover product while Urban Atlas derived as result of mixture of land cover and land use analysis. 3. Allow flexibility in determining the suitable methodological approach (categorisation). According to the practical experience obtained in the production of Urban Atlas 2006, the quality of the different satellite imagery (frames) used for the interpretation/classification of the different LUZs (and, sometimes, within the same LUZ) is rather variable, highly depending on various and specific situation (such as clouds and other meteorological conditions affecting the image, or the time of image capture and issues related to sun declinations and shadows, or sensor inclinations differing to nadir acquisitions, etc) . As a consequence of all these factors affecting the data captured, it is consider important to guarantee certain flexibility in the approach to be adopted for feature determination. In particular, it is strongly advised not to impose, a priori monolithical specific methodology (ie: automatic classification vs visual photo-interpretation) but to allow flexible adaptation of the methodology regarded by the producer as more appropriated according to the specific situation. 4. Product validation It is suggested to adopt a system based on good practices in order to define a precise methodology for evaluating the accuracy in features’ classification. Concerning the imagery, the main real difficulty encountered during the project implementation concerned the supply of images due to them being either insufficient in terms of area coverage or to the presence of clouds. In order to envisage possible solutions that can be used to minimize these kind of problems for future efforts, we suggest that:

- each image Provider adapts their image QC based to the Urban Atlas specifications;

- images outside the specification in terms of cloud cover to be additionally delivered since they could supplement the images already selected that are regarded insufficient to suitably cover the entire AOI;


European Commission




08/01/2013

Page 69/96

- widen the choice of images (same sensors) by agreeing to use older images (3 years) or more recent ones (only one year);

- widen the choice of sensors and produce composite images on the cloudy areas, or use images from lower resolution sensors like Rapid Eye, IRS, etc.

In the future, especially with regard to the future update of the Urban Atlas, the difficulties encountered and the risks they represent should be considered from the start of the project. Overall, the project had proceeded satisfactorily. The feedback received from users seems to show real satisfaction from the point of view of the usefulness of the data and from the point of view of its quality.


European Commission




08/01/2013

Page 70/96

Appendix 1 : List of LUZs produced N

° O

N O

RD

ER

LU

Z

CO

UN

TR

Y

SU

RF

AC

E O

F L

UZ

(K

M²)

CAPI / DIGITAL CLASSIFICATION DELIVERY

ST

AR

T D

AT

E

EN

D D

AT

E

Inte

rnal

Qua

lity

Con

trol

Qua

lity

Ass

essm

ent

Fin

al Q

ualit

y C

ontr

ol

DA

TE

OF

DE

LIV

ER

Y

CO

NT

RA

CT

NU

MB

ER

1 Bruxelles BE 1639,73 km² 25/02/2009 13/03/2009 OK OK OK 04/12/2009 1

2 Praha CZ 7008,19 km² 03/09/2009 02/10/2009 OK OK OK 04/12/2009 1

3 Kobenhavn DK 2837,98 km² 19/02/2009 09/03/2009 OK OK OK 04/12/2009 1

4 Tallinn EE 4350,10 km² 13/05/2009 29/06/2009 OK OK OK 04/12/2009 1

5 Roma IT 3616,84 km² 16/03/2009 14/04/2009 OK OK OK 04/12/2009 1

6 Lefkosia CY 2724,32 km² 07/05/2009 20/05/2009 OK OK OK 04/12/2009 1

7 Vilnius LT 4266,07 km² 02/06/2009 08/09/2009 OK OK OK 04/12/2009 1

8 Luxembourg LU 2612,43 km² 15/05/2009 12/06/2009 OK OK OK 04/12/2009 1

9 Budapest HU 2536,07 km² 11/05/2009 22/05/2009 OK OK OK 04/12/2009 1

10 Valetta MT 243,61 km² 06/05/2009 13/05/2009 OK OK OK 04/12/2009 1

11 Amsterdam NL 1183,76 km² 31/03/2009 20/04/2009 OK OK OK 04/12/2009 1

12 Wien AT 4638,40 km² 30/09/2009 25/11/2009 OK OK OK 04/12/2009 1

13 Warszawa PL 5229,75 km² 11/05/2009 02/06/2009 OK OK OK 04/12/2009 1

14 Lisboa PT 1452,68 km² 13/04/2009 27/04/2009 OK OK OK 04/12/2009 1

15 Ljubljana SL 2573,05 km² 06/04/2009 23/06/2009 OK OK OK 04/12/2009 1

16 Bratislava SK 2055,36 km² 07/05/2009 27/05/2009 OK OK OK 04/12/2009 1

17 Bucarest RO 1086,51 km² 11/05/2009 25/05/2009 OK OK OK 04/12/2009 1

18 Antwerpen BE 952,00 km² 25/02/2009 17/03/2009 OK OK OK 04/12/2009 1

19 Brno CZ 3322,10 km² 14/06/2009 08/07/2009 OK OK OK 04/12/2009 1

20 Thessaloniki GR 1436,94 km² 25/05/2009 23/06/2009 OK OK OK 04/12/2009 1

21 Barcelona ES 1809,58 km² 16/04/2009 28/04/2009 OK OK OK 04/12/2009 1

22 Marseille FR 609,28 km² 13/05/2009 29/05/2009 OK OK OK 04/12/2009 1

23 Milano IT 1361,43 km² 07/04/2009 14/04/2009 OK OK OK 04/12/2009 1

24 Liepaja LV 3666,39 km² 17/04/2009 30/04/2009 OK OK OK 04/12/2009 1

25 Kaunas LT 1635,10 km² 27/02/2009 19/03/2009 OK OK OK 04/12/2009 1

26 Miskolc HU 1015,11 km² 25/02/2009 12/03/2009 OK OK OK 04/12/2009 1

27 Rotterdam NL 713,66 km² 24/02/2009 12/03/2009 OK OK OK 04/12/2009 1

28 Graz AT 1243,17 km² 04/03/2009 16/03/2009 OK OK OK 04/12/2009 1

29 Lodz PL 2875,44 km² 16/06/2009 03/08/2009 OK OK OK 04/12/2009 1

30 Kosice SK 1783,67 km² 15/04/2009 27/04/2009 OK OK OK 04/12/2009 1


European Commission




08/01/2013

Page 71/96

31 Tempere FI 2392,63 km² 28/05/2009 03/07/2009 OK OK OK 04/12/2009 1

32 Liege BE 1067,06 km² 20/04/2009 04/05/2009 OK OK OK 04/12/2009 1

33 Munchen DE 5238,55 km² 28/08/2009 08/10/2009 OK OK OK 04/12/2009 1

34 Koln DE 1634,70 km² 07/04/2009 28/04/2009 OK OK OK 04/12/2009 1

35 Patrai GR 510,58 km² 25/02/2009 13/03/2009 OK OK OK 04/12/2009 1

36 Valencia ES 1454,65 km² 03/06/2009 07/07/2009 OK OK OK 04/12/2009 1

37 Sevilla ES 3093,06 km² 04/05/2009 15/05/2009 OK OK OK 04/12/2009 1

38 Lyon FR 3339,29 km² 20/04/2009 03/07/2009 OK OK OK 04/12/2009 1

39 Lille FR 620,09 km² 16/06/2009 09/07/2009 OK OK OK 04/12/2009 1

40 Napoli IT 571,40 km² 02/06/2009 22/06/2009 OK OK OK 04/12/2009 1

41 Panevezys LT 2243,19 km² 28/08/2009 19/10/2009 OK OK OK 04/12/2009 1

42 Debrecen HU 1687,67 km² 31/07/2009 06/08/2009 OK OK OK 04/12/2009 1

43 Utrecht NL 394,91 km² 11/06/2009 17/06/2009 OK OK OK 04/12/2009 1

44 S'GRAVENHAGE NL 421,28 km² 25/05/2009 02/06/2009 OK OK OK 04/12/2009 1

45 Linz AT 1762,08 km² 11/09/2009 05/10/2009 OK OK OK 04/12/2009 1

46 Braga PT 499,94 km² 16/06/2009 06/07/2009 OK OK OK 04/12/2009 1

47 Malmö SE 1852,48 km² 25/08/2009 09/10/2009 OK OK OK 04/12/2009 1

48 Gent BE 546,57 km² 03/05/2009 16/06/2009 OK OK OK 04/12/2009 1

49 Dresden DE 2635,29 km² 02/11/2009 25/11/2009 OK OK OK 04/12/2009 1

50 Trier DE 1217,41 km² 16/07/2009 22/07/2009 OK OK OK 04/12/2009 1

51 Iraklion GR 605,95 km² 20/04/2009 28/04/2009 OK OK OK 04/12/2009 1

52 Zaragoza ES 2304,94 km² 01/07/2009 21/07/2009 OK OK OK 04/12/2009 1

53 Nice FR 334,65 km² 07/08/2009 19/08/2009 OK OK OK 04/12/2009 1

54 Metz FR 1854,83 km² 02/07/2009 09/07/2009 OK OK OK 04/12/2009 1

55 Palermo IT 1190,49 km² 17/07/2009 04/08/2009 OK OK OK 04/12/2009 1

56 Firenze IT 1273,13 km² 01/07/2009 09/07/2009 OK OK OK 04/12/2009 1

57 Pecs HU 578,28 km² 29/07/2009 04/08/2009 OK OK OK 04/12/2009 1

58 Eindhoven NL 332,64 km² 22/05/2009 29/05/2009 OK OK OK 04/12/2009 1

59 Poznan PL 3738,26 km² 28/08/2009 21/10/2009 OK OK OK 04/12/2009 1

60 Coimbra PT 1264,74 km² 13/10/2009 02/11/2009 OK OK OK 04/12/2009 1

61 Cluj Napoca RO 599,60 km² 05/08/2009 14/08/2009 OK OK OK 04/12/2009 1

62 Timisoara RO 241,25 km² 06/05/2009 20/05/2009 OK OK OK 04/12/2009 1

63 Craiova RO 347,57 km² 17/07/2009 23/07/2009 OK OK OK 04/12/2009 1

64 Braila RO 442,25 km² 06/08/2009 13/08/2009 OK OK OK 04/12/2009 1

65 Plovdiv BG 1235,84 km² 03/07/2009 21/07/2009 OK OK OK 04/12/2009 1

66 Charleroi BE 628,14 km² 24/06/2009 01/07/2009 OK OK OK 04/12/2009 1

67 Brugge BE 412,00 km² 08/06/2009 16/06/2009 OK OK OK 04/12/2009 1

68 Namur BE 404,02 km² 06/05/2009 15/05/2009 OK OK OK 04/12/2009 1

69 Usti nad Labem CZ 880,00 km² 19/05/2009 02/06/2009 OK OK OK 04/12/2009 1

70 Olomouc CZ 1629,92 km² 28/07/2009 04/08/2009 OK OK OK 04/12/2009 1

71 Liberec CZ 1339,86 km² 12/10/2009 04/11/2009 OK OK OK 04/12/2009 1


European Commission




08/01/2013

Page 72/96

72 Hannover DE 2993,81 km² 28/08/2009 08/10/2009 OK OK OK 04/12/2009 1

73 Wuppertal DE 167,81 km² 06/05/2009 13/05/2009 OK OK OK 04/12/2009 1

74 Karlsruhe DE 1268,72 km² 07/07/2009 23/07/2009 OK OK OK 04/12/2009 1

75 Mönchengladbach DE 171,21 km² 22/04/2009 29/04/2009 OK OK OK 04/12/2009 1

76 Larisa GR 1567,87 km² 110/07/2009 23/07/2009 OK OK OK 04/12/2009 1

77 Volos GR 308,82 km² 17/06/2009 26/06/2009 OK OK OK 04/12/2009 1

78 loannina GR 1339,41 km² 07/07/2009 23/07/2009 OK OK OK 04/12/2009 1

79 Las Palmas ES 875,33 km² 02/07/2009 28/07/2009 OK OK OK 04/12/2009 1

80 Bilbao ES 991,32 km² 26/05/2009 17/06/2009 OK OK OK 04/12/2009 1

81 Palma di Mallorca ES 2174,65 km² 05/08/2009 08/09/2009 OK OK OK 04/12/2009 1

82 Valladolid ES 3055,61 km² 27/08/2009 18/09/2009 OK OK OK 04/12/2009 1

83 Cordoba ES 1265,45 km² 30/06/2009 07/07/2009 OK OK OK 04/12/2009 1

84 Vigo ES 1435,67 km² 26/08/2009 22/10/2009 OK OK OK 04/12/2009 1

85 Gijon ES 524,70 km² 14/08/2009 25/08/2009 OK OK OK 04/12/2009 1

86 Montpellier FR 599,07 km² 06/08/2009 24/08/2009 OK OK OK 04/12/2009 1

87 Toulon FR 340,72 km² 06/08/2009 13/08/2009 OK OK OK 04/12/2009 1

88 Le Havre FR 641,72 km² 08/10/2009 22/10/2009 OK OK OK 04/12/2009 1

89 Galway IE 51,95 km² 02/11/2009 18/11/2009 OK OK OK 04/12/2009 1

90 Trieste IT 215,22 km² 11/08/2009 18/08/2009 OK OK OK 04/12/2009 1

91 Brescia IT 548,75 km² 12/08/2009 26/08/2009 OK OK OK 04/12/2009 1

92 Szeged HU 760,28 km² 22/10/2009 02/11/2009 OK OK OK 04/12/2009 1

93 Tilburg NL 394,39 km² 02/11/2009 23/11/2009 OK OK OK 04/12/2009 1

94 Groningen NL 949,09 km² 06/11/2009 23/11/2009 OK OK OK 04/12/2009 1

95 Nijmegen NL 322,00 km² 02/11/2009 25/11/2009 OK OK OK 04/12/2009 1

96 Breda NL 503,76 km² 06/11/2009 20/11/2009 OK OK OK 04/12/2009 1

97 Banska Bystrica SK 815,47 km² 02/11/2009 18/11/2009 OK OK OK 04/12/2009 1

98 Ostrava CZ 3904,63 km² 16/06/2009 13/10/2009 OK OK OK 20/01/2010 1

99 Torino IT 1893,59 km² 19/11/2009 15/12/2009 OK OK OK 20/01/2010 1

100 Krakow PL 3024,73 km² 14/10/2009 04/11/2009 OK OK OK 20/01/2010 1

101 Bremen DE 5920,41 km² 25/08/2009 24/11/2009 OK OK OK 20/01/2010 1

102 Bielefeld DE 2943,17 km² 02/11/2009 18/12/2009 OK OK OK 20/01/2010 1

103 Reggio di Calabria IT 492,83 km² 01/12/2009 21/12/2009 OK OK OK 20/01/2010 1

104 Gyor HU 1451,53 km² 15/12/2009 08/01/2010 OK OK OK 20/01/2010 1

105 Apeldoorn NL 630,94 km² 17/11/2009 30/11/2009 OK OK OK 20/01/2010 1

106 Lublin PL 2902,02 km² 30/11/2009 17/12/2009 OK OK OK 20/01/2010 1

107 Torun PL 1360,06 km² 01/12/2009 21/12/2009 OK OK OK 20/01/2010 1

108 Olzstyn PL 2949,00 km² 08/12/2009 18/12/2009 OK OK OK 20/01/2010 1

109 Presov SK 944,63 km² 27/11/2009 08/12/2009 OK OK OK 20/01/2010 1

110 Jonkoping SE 3491,95 km² 02/11/2009 07/12/2009 OK OK OK 20/01/2010 1

111 Darmstadt DE 790,62 km² 15/12/2009 22/12/2009 OK OK OK 20/01/2010 1

112 Frankfurt (Oder) DE 151,48 km² 28/12/2009 07/01/2010 OK OK OK 20/01/2010 1


European Commission




08/01/2013

Page 73/96

113 Kavala GR 355,57 km² 28/12/2009 07/01/2010 OK OK OK 20/01/2010 1

114 Kalamata GR 447,36 km² 28/12/2009 21/01/2010 OK OK OK 20/01/2010 1

115 Lens - Lievin FR 243,99 km² 08/01/2010 19/01/2010 OK OK OK 20/01/2010 1

116 Pescara IT 683,45 km² 15/12/2009 10/01/2010 OK OK OK 20/01/2010 1

117 Ancona IT 409,83 km² 15/12/2009 05/01/2009 OK OK OK 20/01/2010 1

118 Nitra SK 878,07 km² 12/01/2010 19/01/2010 OK OK OK 26/02/2010 1

119 Alicante ES 679,03 km² 12/01/2010 25/01/2010 OK OK OK 26/02/2010 1

120 Clermont Ferrand FR 1833,60 km² 08/01/2010 02/02/2010 OK OK OK 26/02/2010 1

121 Nancy FR 1856,04 km² 05/01/2010 25/01/2010 OK OK OK 26/02/2010 1

122 Catania IT 591,58 km² 25/01/2010 12/02/2010 OK OK OK 26/02/2010 1

123 Taranto IT 1427,07 km² 08/12/2009 22/01/2010 OK OK OK 26/02/2010 1

124 Modena IT 647,57 km² 25/01/2010 16/02/2010 OK OK OK 26/02/2010 1

125 Cagliari IT 1695,82 km² 08/12/2009 08/01/2010 OK OK OK 26/02/2010 1

126 Gdansk PL 3351,19 km² 08/12/2009 22/01/2010 OK OK OK 26/02/2010 1

127 Kielce PL 2373,78 km² 26/11/2009 07/12/2009 OK OK OK 26/02/2010 1

128 Radom PL 1644,72 km² 15/12/2009 08/01/2010 OK OK OK 26/02/2010 1

129 Setubal PT 177,02 km² 07/01/2009 19/01/2010 OK OK OK 26/02/2010 1

130 Aveiro PT 275,07 km² 11/01/2010 20/01/2010 OK OK OK 26/02/2010 1

131 Sheffield UK 1878,19 km² 27/01/2010 19/02/2010 OK OK OK 26/02/2010 1

132 Leicester UK 1407,80 km² 12/01/2010 26/01/2010 OK OK OK 26/02/2010 1

133 Oradea RO 205,95 km² 19/01/2010 26/01/2010 OK OK OK 26/02/2010 1

134 Bacau RO 225,45 km² 12/01/2010 26/01/2010 OK OK OK 26/02/2010 1

135 Arad RO 527,64 km² 20/01/2010 22/01/2010 OK OK OK 26/02/2010 1

136 Sibiu RO 597,74 km² 19/01/2010 22/01/2010 OK OK OK 26/02/2010 1

137 Varna BG 887,78 km² 26/01/2010 12/02/2010 OK OK OK 26/02/2010 1

138 Burgas BG 1402,52 km² 12/01/2010 02/02/2010 OK OK OK 26/02/2010 1

139 Hradec Kralove CZ 879,82 km² 25/01/2010 05/02/2010 OK OK OK 26/02/2010 1

140 Pardubice CZ 901,41 km² 21/01/2010 09/02/2010 OK OK OK 26/02/2010 1

141 Zlin CZ 1040,37 km² 01/02/2010 10/02/2010 OK OK OK 26/02/2010 1

142 Jihlava CZ 1191,10 km² 22/01/2010 05/02/2010 OK OK OK 26/02/2010 1

143 Dusseldorf DE 1207,09 km² 01/02/2010 15/02/2010 OK OK OK 26/02/2010 1

144 Santander ES 597,97 km² 19/01/2010 01/02/2010 OK OK OK 26/02/2010 1

145 Santa Cruz de Tenerife ES 608,49 km² 26/01/2010 04/02/2010 OK OK OK 26/02/2010 1

146 Foggia IT 1059,32 km² 25/01/2010 18/02/2010 OK OK OK 26/02/2010 1

147 Trento IT 787,41 km² 20/01/2010 12/02/2010 OK OK OK 26/02/2010 1

148 Catanzaro IT 771,39 km² 21/01/2010 04/02/2010 OK OK OK 26/02/2010 1

149 Caserta IT 678,96 km² 21/01/2010 29/01/2010 OK OK OK 26/02/2010 1

150 Cremona IT 669,47 km² 26/01/2010 10/02/2010 OK OK OK 26/02/2010 1

151 Faro PT 488,32 km² 25/01/2010 27/01/2010 OK OK OK 26/02/2010 1

152 Piatra Neamt RO 150,31 km² 19/02/2010 22/02/2010 OK OK OK 26/02/2010 1

153 Calarasi RO 249,38 km² 16/02/2010 19/02/2010 OK OK OK 26/02/2010 1


European Commission




08/01/2013

Page 74/96

154 Giurgiu RO 113,32 km² 16/02/2010 19/02/2010 OK OK OK 26/02/2010 1

155 Alba Iulia RO 264,02 km² 17/02/2010 22/02/2010 OK OK OK 26/02/2010 1

156 Ceske Budějovice CZ 1638,78 km² 29/01/2010 12/03/2010 OK OK OK 22/03/2010 1

157 Nantes FR 2322,91 km² 26/01/2010 14/03/2010 OK OK OK 22/03/2010 1

158 Strasbourg FR 1382,84 km² 02/03/2010 11/03/2010 OK OK OK 22/03/2010 1

159 Saint-Etienne FR 575,48 km² 25/01/2010 02/03/2010 OK OK OK 22/03/2010 1

160 Orleans FR 2085,79 km² 18/03/2010 19/03/2010 OK OK OK 22/03/2010 1

161 Venezia IT 1215,81 km² 11/01/2010 12/03/2010 OK OK OK 22/03/2010 1

162 Belfast UK 968,22 km² 24/02/2010 05/03/2010 OK OK OK 22/03/2010 1

163 Saarbrucken DE 1552,03 km² 02/03/2010 15/03/2010 OK OK OK 22/03/2010 1

164 Málaga ES 952,67 km² 28/01/2010 05/03/2010 OK OK OK 22/03/2010 1

165 Badajoz ES 1487,68 km² 26/02/2010 09/03/2010 OK OK OK 22/03/2010 1

166 Amiens FR 1783,98 km² 01/03/2010 18/03/2010 OK OK OK 22/03/2010 1

167 Reims FR 1810,41 km² 02/03/2010 16/03/2010 OK OK OK 22/03/2010 1

168 Salerno IT 955,86 km² 25/01/2010 17/03/2010 OK OK OK 22/03/2010 1

169 Stoke-on-trent UK 890,10 km² 02/03/2010 18/03/2010 OK OK OK 22/03/2010 1

170 Targu Mures RO 145,67 km² 26/02/2010 02/03/2010 OK OK OK 22/03/2010 1

171 Ruse BG 898,95 km² 26/02/2010 12/03/2010 OK OK OK 22/03/2010 1

172 Murcia ES 1334,60 km² 31/03/2010 16/04/2010 OK OK OK 29/04/2010 1

173 Grenoble FR 1618,78 km² 13/01/2010 20/04/2010 OK OK OK 29/04/2010 1

174 Tours FR 1825,20 km² 12/03/2010 30/03/2010 OK OK OK 29/04/2010 1

175 Bari IT 901,27 km² 10/03/2010 17/03/2010 OK OK OK 29/04/2010 1

176 Verona IT 1220,18 km² 02/02/2010 15/03/2010 OK OK OK 29/04/2010 1

177 Padova IT 988,28 km² 20/01/2010 19/04/2010 OK OK OK 29/04/2010 1

178 Bydgoszcz PL 3416,42 km² 19/03/2010 19/04/2010 OK OK OK 29/04/2010 1

179 Katowice PL 2658,55 km² 13/01/2010 11/03/2010 OK OK OK 29/04/2010 1

180 Santiago de Compostela ES 1366,16 km² 07/04/2010 20/04/2010 OK OK OK 29/04/2010 1

181 Logrono ES 1448,93 km² 06/04/2010 22/04/2010 OK OK OK 29/04/2010 1

182 Perugia IT 814,53 km² 14/04/2010 21/04/2010 OK OK OK 29/04/2010 1

183 Székesfehérvár HU 1154,24 km² 06/04/2010 23/04/2010 OK OK OK 29/04/2010 1

184 Enschede NL 357,92 km² 09/04/2010 16/04/2009 OK OK OK 29/04/2010 1

185 Arnhem NL 492,28 km² 02/04/2010 14/04/2010 OK OK OK 29/04/2010 1

186 Leeuwarden NL 455,99 km² 13/04/2010 21/04/2010 OK OK OK 29/04/2010 1

187 Plock PL 1899,85 km² 31/03/2010 23/04/2010 OK OK OK 29/04/2010 1

188 Koszalin PL 1762,20 km² 01/04/2010 16/04/2010 OK OK OK 29/04/2010 1

189 Trnava SK 748,64 km² 06/04/2010 14/04/2010 OK OK OK 29/04/2010 1

190 Trencin SK 681,01 km² 20/04/2010 25/04/2010 OK OK OK 29/04/2010 1

191 Cambridge UK 952,37 km² 15/03/2010 26/03/2010 OK OK OK 29/04/2010 1

192 Wolverhampton UK 484,08 km² 02/04/2010 14/04/2010 OK OK OK 29/04/2010 1

193 Nottingham UK 910,66 km² 09/03/2010 01/04/2010 OK OK OK 29/04/2010 1

194 Vidin BG 524,97 km² 18/03/2010 29/03/2010 OK OK OK 29/04/2010 1


European Commission




08/01/2013

Page 75/96

195 Rennes FR 2575,62 km² 31/03/2010 23/04/2010 OK OK OK 18/05/2010 1

196 Aix en Provence FR 1305,51 km² 27/04/2010 10/05/2010 OK OK OK 18/05/2010 1

197 Bologna IT 2065,98 km² 11/01/2010 28/04/2010 OK OK OK 18/05/2010 1

198 Czestochowa PL 2586,64 km² 19/04/2010 03/05/2010 OK OK OK 18/05/2010 1

199 Vitoria-Gasteiz ES 2334,43 km² 31/03/2010 29/04/2010 OK OK OK 18/05/2010 1

200 Heerlen NL 215,68 km² 12/04/2010 15/04/2010 OK OK OK 18/05/2010 1

201 Gorzów Wielkopolski PL 1317,55 km² 12/04/2010 03/05/2010 OK OK OK 18/05/2010 1

202 Zielona Góra PL 1641,18 km² 28/04/2010 04/05/2010 OK OK OK 18/05/2010 1

203 Konin PL 767,03 km² 20/04/2010 03/05/2010 OK OK OK 18/05/2010 1

204 Zilina SK 824,54 km² 31/03/2010 19/04/2010 OK OK OK 18/05/2010 1

205 Leipzig DE 2818,46 km² 29/05/2010 15/06/2010 OK OK OK 23/06/2010 1

206 Oulu FI 3773,58 km² 28/04/2010 20/05/2010 OK OK OK 23/06/2010 1

207 Linkoping SE 4252,50 km² 06/05/2010 17/06/2010 OK OK OK 23/06/2010 1

208 Magdeburg DE 4350,33 km² 03/05/2010 17/06/2010 OK OK OK 23/06/2010 1

209 Oviedo ES 2354,95 km² 06/05/2010 25/05/2010 OK OK OK 23/06/2010 1

210 Toledo ES 3638,13 km² 03/05/2010 20/05/2010 OK OK OK 23/06/2010 1

211 Dijon FR 2300,63 km² 28/04/2010 19/05/2010 OK OK OK 23/06/2010 1

212 Poitiers FR 1773,87 km² 27/04/2010 19/05/2010 OK OK OK 23/06/2010 1

213 Potenza IT 1513,30 km² 03/05/2010 26/05/2010 OK OK OK 23/06/2010 1

214 l'Aquila IT 1599,70 km² 07/05/2010 31/05/2010 OK OK OK 23/06/2010 1

215 Kalisz PL 3123,16 km² 29/04/2010 02/06/2010 OK OK OK 23/06/2010 1

216 Pleven BG 1807,06 km² 26/04/2010 27/05/2010 OK OK OK 23/06/2010 1

217 Athens GR 3031,26 km² 11/06/2010 12/07/2010 OK OK OK 23/07/2010 1

218 Tartu EE 3017,94 km² 22/06/2010 19/07/2010 OK OK OK 23/07/2010 1

219 Wroclaw PL 4608,38 km² 12/05/2010 22/06/2010 OK OK OK 23/07/2010 1

220 Frankfurt am Main DE 4309,05 km² 03/05/2010 21/06/2010 OK OK OK 23/07/2010 1

221 Bialystok PL 5140,38 km² 15/06/2010 15/07/2010 OK OK OK 23/07/2010 1

222 Orebro SE 3707,05 km² 14/06/2010 23/06/2010 OK OK OK 23/07/2010 1

223 Pamplona-Iruna ES 4405,30 km² 11/06/2010 21/07/2010 OK OK OK 23/07/2010 1

224 London UK 8991,68 km² 22/06/2010 23/07/2010 OK OK OK 30/08/2010 1

225 Toulouse FR 4064,45 km² 17/06/2010 21/07/2010 OK OK OK 30/08/2010 1

226 Umea SE 9792,33 km² 27/06/2010 19/07/2010 OK OK OK 30/08/2010 1

227 Odense DK 3487,07 km² 06/07/2010 17/08/2010 OK OK OK 30/08/2010 1

228 Schwerin DE 4915,06 km² 19/07/2010 20/082010 OK OK OK 30/08/2010 1

229 Koblenz DE 934,08 km² 22/04/2010 29/04/2010 OK OK OK 30/08/2010 1

230 Madrid ES 8054,23 km² 06/07/2010 12/10/2010 OK OK OK 13/10/2010 1

231 Plzen CZ 3124,93 km² 23/07/2010 06/10/2010 OK OK OK 13/10/2010 1

232 Bordeaux FR 3909,90 km² 23/06/2010 13/10/2010 OK OK OK 13/10/2010 1

233 Szczecin PL 6078,12 km² 06/08/2010 22/09/2010 OK OK OK 13/10/2010 1

234 Besançon FR 1682,77 km² 09/07/2010 21/09/2010 OK OK OK 13/10/2010 1

235 Campobasso IT 1320,38 km² 16/07/2010 22/09/2010 OK OK OK 13/10/2010 1


European Commission




08/01/2013

Page 76/96

236 Nyíregyháza HU 1448,92 km² 21/07/2010 30/08/2010 OK OK OK 13/10/2010 1

237 Kecskemét HU 1493,78 km² 21/07/2010 27/08/2010 OK OK OK 13/10/2010 1

238 Jelenia Góra PL 592,93 km² 29/04/2010 25/08/2010 OK OK OK 13/10/2010 1

239 Berlin DE 17504,83 km² 23/09/2010 21/10/2010 OK OK OK 21/10/2010 1

240 Sofia BG 3435,78 km² 14/06/2010 19/11/2010 OK OK OK 26/11/2010 2

241 Aarhus DK 4554,46 km² 06/07/2010 26/11/2010 OK OK OK 26/11/2010 2

242 Gozo MT 68,03 km² 28/04/2010 28/04/2010 OK OK OK 26/11/2010 2

243 Liverpool UK 650,57 km² 10/08/2010 18/11/2010 OK OK OK 26/11/2010 2

244 Genova IT 933,65 km² 29/04/2010 18/11/2010 OK OK OK 26/11/2010 2

245 Edinburgh UK 1739,52 km² 16/07/2010 04/11/2010 OK OK OK 26/11/2010 2

246 Waterford IE 43,19 km² 17/03/2010 18/03/2010 OK OK OK 26/11/2010 2

247 Weimar DE 890,88 km² 26/10/2010 25/11/2010 OK OK OK 26/11/2010 2

248 Caen FR 1630,38 km² 19/10/2010 23/11/2010 OK OK OK 26/11/2010 2

249 Ajaccio FR 1024,68 km² 17/06/2010 27/10/2010 OK OK OK 26/11/2010 2

250 Derry UK 390,46 km² 20/10/2010 09/11/2010 OK OK OK 26/11/2010 2

251 Stara Zagora BG 88,14 km² 15/10/2010 27/10/2010 OK OK OK 26/11/2010 2

252 Helsinki FI 3095,87 km² 06/07/2010 20/12/2010 OK OK OK 10/01/2011 2

253 Turku FI 1746,07 km² 07/12/2010 16/12/2010 OK OK OK 10/01/2011 2

254 Coventry UK 820,94 km² 17/12/2010 21/12/2010 OK OK OK 10/01/2011 2

255 Karlovy Vary CZ 1630,00 km² 20/10/2010 13/12/2010 OK OK OK 10/01/2011 2

256 Sassari IT 1230,36 km² 22/06/2010 01/12/2010 OK OK OK 10/01/2011 2

257 Opole PL 1708,14 km² 17/12/2010 23/12/2011 OK OK OK 10/01/2011 2

258 Wrexham UK 955,58 km² 02/11/2010 16/12/2010 OK OK OK 10/01/2011 2

259 Worcester UK 1289,73 km² 02/12/2010 16/12/2010 OK OK OK 10/01/2011 2

260 Cork IE 2112,87 km² 07/12/2010 11/02/2011 OK OK OK 23/02/2011 2

261 Oporto PT 567,08 km² 04/01/2011 04/02/2011 OK OK OK 23/02/2011 2

262 Göteborg SE 4220,49 km² 02/12/2010 24/01/2011 OK OK OK 23/02/2011 2

263 Birmingham UK 1596,74 km² 08/02/2011 18/02/2011 OK OK OK 23/02/2011 2

264 Uppsala SE 6849,51 km² 22/12/2010 01/02/2011 OK OK OK 23/02/2011 2

265 Paris FR 12106,34 km² 28/01/2011 18/03/2011 OK OK OK 31/03/2011 2

266 Riga LV 5413,94 km² 28/02/2011 29/03/2011 OK OK OK 31/03/2011 2

267 Kiel DE 3394,09 km² 09/02/2011 31/03/2011 OK OK OK 31/03/2011 2

268 Nowy Sacz PL 454,65 km² 14/02/2011 29/03/2011 OK OK OK 31/03/2011 2

269 Ponto Delgada PT 536,91 km² 15/02/2011 28/03/2011 OK OK OK 31/03/2011 2

270 Essen DE 4462,57 km² 08/02/2011 10/05/2011 OK OK OK 13/05/2011 2

271 Bonn DE 1300,33 km² 09/05/2011 12/05/2011 OK OK OK 13/05/2011 2

272 Funchal PT 260,00 km² 10/05/2011 13/05/2011 OK OK OK 13/05/2011 2

273 Portsmouth UK 202,69 km² 05/05/2011 06/05/2011 OK OK OK 13/05/2011 2

274 Erfurt DE 2879,46 km² 15/02/2011 26/04/2011 OK OK OK 13/05/2011 2

275 Mainz DE 708,18 km² 29/04/2011 06/05/2011 OK OK OK 13/05/2011 2

276 Limoges FR 1855,93 km² 24/02/2011 22/04/2011 OK OK OK 13/05/2011 2


European Commission




08/01/2013

Page 77/96

277 Rzeszów PL 1287,50 km² 06/04/2011 22/04/2011 OK OK OK 13/05/2011 2

278 Dublin IE 7013,88 km² 16/05/2011 23/06/2011 OK OK OK 01/07/2011 2

279 Stockholm SE 7005,77 km² 11/05/2011 23/06/2011 OK OK OK 01/07/2011 2

280 Maribor SL 2185,66 km² 09/05/2011 26/05/2011 OK OK OK 01/07/2011 2

281 Salzburg AT 1757,81 km² 20/06/2011 28/06/2011 OK OK OK 01/07/2011 2

282 Cardiff UK 1185,35 km² 21/04/2011 24/05/2011 OK OK OK 01/07/2011 2

283 Wiesbaden DE 1027,10 km² 28/02/2011 27/06/2011 OK OK OK 01/07/2011 2

284 Freiburg im Breisgau DE 2226,95 km² 16/05/2011 14/06/2011 OK OK OK 01/07/2011 2

285 Limerick IE 3541,09 km² 15/06/2011 28/07/2011 OK OK OK 29/07/2011 2

286 Manchester UK 1287,70 km² 15/06/2011 08/07/2011 OK OK OK 29/07/2011 2

287 Aalborg DK 6169,39 km² 07/07/2011 28/07/2011 OK OK OK 29/07/2011 2

288 Halle an der Saale DE 1583,08 km² 14/06/2011 20/07/2011 OK OK OK 29/07/2011 2

289 Göttingen DE 2405,87 km² 13/07/2011 28/07/2011 OK OK OK 29/07/2011 2

290 Augsburg DE 2001,06 km² 20/06/2011 20/07/2011 OK OK OK 29/07/2011 2

291 Exeter UK 2466,69 km² 12/07/2011 28/07/2011 OK OK OK 29/07/2011 2

292 Kingston-upon-Hull UK 2500,06 km² 15/06/2011 21/07/2011 OK OK OK 29/07/2011 2

293 Hamburg DE 7232,45 km² 03/08/2011 26/08/2011 OK OK OK 02/09/2011 2

294 Nürnberg DE 2725,27 km² 12/08/2011 29/08/2011 OK OK OK 02/09/2011 2

295 Rouen FR 1600,20 km² 19/08/2011 31/08/2011 OK OK OK 02/09/2011 2

296 Innsbruck AT 2104,27 km² 12/08/2011 31/08/2011 OK OK OK 02/09/2011 2

297 Newcastle upon Tyne UK 3401,18 km² 15/06/2011 01/09/2011 OK OK OK 02/09/2011 2

298 Suwalki PL 625,47 km² 22/08/2011 26/08/2011 OK OK OK 02/09/2011 2

299 Lincoln UK 731,55 km² 25/07/2011 01/09/2011 OK OK OK 02/09/2011 2

300 Leeds UK 5131,92 km² 07/09/2011 26/09/2011 OK OK OK 30/09/2011 2

301 Glasgow UK 3390,53 km² 01/09/2011 21/09/2011 OK OK OK 30/09/2011 2

302 Stuttgart DE 3679,26 km² 12/09/2011 27/09/2011 OK OK OK 30/09/2011 2

303 Bristol UK 1343,86 km² 22/08/2011 19/09/2011 OK OK OK 30/09/2011 2

304 Aberdeen UK 6526,44 km² 23/08/2011 23/09/2011 OK OK OK 30/09/2011 2

305 Regensburg DE 2392,36 km² 13/09/2011 30/09/2011 OK OK OK 30/09/2011 2


European Commission




08/01/2013

Page 78/96

Appendix 2 : Results of confusion matrices by LUZ

Luz Nb of points Urban points Rural points Urban Accuracy Rural Accuracy Overall Accuracy

Aalborg 687 239 448 88,50% 91,12% 85,74%

Aarhus 589 287 302 92,02% 94,91% 90,66%

Aberdeen 871 380 491 90,59% 93,23% 88,29%

Aix-en-Provence 210 108 102 92,38% 92,86% 89,52%

Ajaccio 162 66 96 93,83% 90,12% 87,04%

Alba_Iulia 87 20 67 95,40% 98,85% 95,40%

Alicante_Alacant 117 66 51 88,89% 91,45% 88,03%

Amiens 415 212 203 88,43% 93,73% 87,71%

Amsterdam 278 167 111 96,40% 97,48% 96,40%

Ancona 66 38 28 90,91% 98,48% 90,91%

Antwerpen 725 466 259 95,17% 97,38% 94,90%

Apeldoorn 81 40 41 86,42% 95,06% 85,19%

Arad 84 42 42 96,43% 96,43% 92,86%

Arnhem 95 56 39 90,53% 95,79% 90,53%

Athina 474 280 194 92,41% 93,67% 89,24%

Augsburg 285 150 135 94,04% 96,49% 92,98%

Aveiro 51 28 23 94,12% 94,12% 88,24%

Bacau 80 22 58 90,00% 92,50% 87,50%

Badajoz 241 104 137 93,36% 92,53% 90,04%

Banska_Bystrica 165 74 91 87,27% 90,30% 85,45%

Barcelona 440 262 178 93,41% 93,18% 89,55%

Bari 137 69 68 90,51% 93,43% 89,05%

Belfast 181 121 60 91,16% 96,13% 89,50%

Berlin 2710 1469 1241 86,13% 93,17% 85,17%

Besançon 362 193 169 90,33% 93,09% 87,57%

Bialystok 628 290 338 91,08% 95,70% 89,97%

Bielefeld 562 298 264 89,50% 93,77% 88,43%

Bilbao 137 70 67 95,62% 96,35% 94,16%

Birmingham 522 396 126 87,55% 95,79% 86,78%

Bologna 273 141 132 89,74% 95,24% 88,28%

Bonn 245 150 95 87,76% 94,29% 87,76%

Bordeaux 677 331 346 90,99% 91,29% 86,56%

Braga 67 28 39 95,52% 95,52% 94,03%

Braila 54 28 26 87,04% 96,30% 87,04%

Bratislava 367 183 184 89,92% 95,64% 88,83%

Breda 94 64 30 88,30% 96,81% 88,30%

Bremen 759 385 374 89,72% 94,20% 88,80%

Brescia 114 56 58 87,72% 95,61% 86,84%

Bristol 245 161 84 88,16% 94,29% 87,35%

Brno 443 163 280 88,26% 92,33% 86,46%

Brugge 84 41 43 98,81% 97,62% 97,62%

Bruxelles 1203 661 542 89,44% 93,10% 88,45%

Bucuresti 284 185 99 96,83% 98,59% 96,83%

Budapest 677 482 195 85,52% 93,65% 85,23%

Burgas 192 79 113 88,54% 89,06% 87,50%

Bydgoszcz 645 311 334 89,46% 94,73% 88,06%


European Commission




08/01/2013

Page 79/96

Caen 218 95 123 88,99% 95,41% 88,07%

Cagliari 224 94 130 94,64% 95,09% 92,41%

Calarasi 84 17 67 92,86% 91,67% 89,29%

Cambridge 153 68 85 93,46% 97,39% 92,81%

Campobasso 116 32 84 93,10% 91,38% 87,93%

Cardiff 250 116 134 91,20% 91,20% 87,60%

Caserta 119 73 46 93,28% 96,64% 91,60%

Catania 115 62 53 92,17% 96,52% 91,30%

Catanzaro 107 27 80 94,39% 92,52% 88,79%

Ceske_Budejovice 330 145 185 89,70% 92,42% 88,79%

Charleroi 180 86 94 95,56% 97,22% 95,56%

Clermont-Ferrand 250 120 130 91,20% 92,40% 89,60%

Cluj-Napoca 123 43 80 86,99% 93,50% 85,37%

Coimbra 158 66 92 91,14% 90,51% 86,08%

Cordoba 280 143 137 91,07% 92,14% 86,79%

Cork 500 282 218 87,00% 94,40% 85,60%

Coventry 183 100 83 87,43% 92,35% 86,89%

Craiova 75 38 37 85,33% 89,33% 85,33%

Cremona 152 67 85 93,42% 97,37% 93,42%

Czestochowa 555 278 277 88,65% 92,25% 87,39%

Darmstadt 164 84 80 94,51% 96,34% 93,29%

Debrecen 211 92 119 90,05% 88,15% 85,31%

Derry 78 23 55 93,59% 100,00% 93,59%

Dijon 524 260 264 87,98% 94,27% 87,40%

Dresden 447 221 226 86,13% 93,29% 85,01%

Dublin 862 445 417 87,24% 93,27% 85,85%

Dusseldorf 358 226 132 90,22% 98,60% 90,22%

Edinburgh 357 205 152 88,52% 94,12% 87,96%

Eindhoven 87 58 29 94,25% 96,55% 93,10%

Enschede 68 46 22 91,18% 98,53% 91,18%

Erfurt 393 173 220 90,08% 94,66% 89,57%

Essen 1227 878 349 86,15% 94,21% 85,66%

Exeter 271 95 176 90,41% 88,19% 85,61%

Faro 82 39 43 91,46% 90,24% 86,59%

Firenze 171 79 92 88,30% 92,98% 85,38%

Foggia 172 66 106 93,02% 97,09% 93,02%

Frankfurt_(Oder) 70 21 49 90,00% 94,29% 90,00%

Frankfurt_am_Main 915 515 400 91,37% 96,39% 90,60%

Freiburg im Breisgau 239 110 129 88,70% 90,79% 85,77%

Funchal 103 22 81 95,15% 87,38% 85,44%

Galway 67 47 20 91,04% 92,54% 91,04%

Gdansk 399 197 202 87,72% 92,23% 87,47%

Genova 119 43 76 90,76% 89,92% 87,39%

Gent 206 120 86 94,66% 98,06% 94,66%

Gijon 63 29 34 82,54% 85,71% 74,60%

Giurgiu 103 15 88 98,06% 99,03% 98,06%

Glasgow 532 308 224 89,47% 92,67% 86,65%

Gorzow_Wielkopolski 203 96 107 85,71% 92,61% 85,22%


European Commission




08/01/2013

Page 80/96

Gozo 69 8 61 95,65% 95,65% 94,20%

Graz 488 209 279 92,42% 95,29% 92,01%

Grenoble 211 100 111 91,00% 96,21% 89,10%

Groningen 146 84 62 87,67% 93,84% 87,67%

Gyor 193 100 93 88,60% 86,53% 86,01%

Göteborg 651 335 316 89,09% 92,63% 87,25%

Göttingen 587 300 287 92,50% 96,08% 91,99%

Halle an der Saale 368 169 199 87,23% 91,58% 85,60%

Hamburg 1411 896 515 89,37% 95,68% 88,87%

Hannover 592 313 279 85,47% 93,92% 85,14%

Heerlen 67 42 25 86,57% 92,54% 86,57%

Helsinki 626 338 288 87,22% 90,73% 85,78%

Hradec_Kralove 105 41 64 90,48% 97,14% 89,52%

Innsbruck 382 145 237 93,46% 94,50% 90,31%

Ioannina 226 59 167 90,71% 88,50% 85,40%

Iraklion 108 61 47 96,30% 98,15% 96,30%

Jelenia_Gora 138 74 64 88,41% 92,75% 86,96%

Jihlava 214 76 138 89,72% 93,46% 88,79%

Jonkoping 518 239 279 89,38% 88,61% 85,91%

Kalamata 73 22 51 91,78% 91,78% 86,30%

Kalisz 525 203 322 92,57% 96,57% 91,81%

Karlovy Vary 280 104 176 90,00% 90,36% 88,21%

Karlsruhe 193 96 97 87,56% 96,37% 84,97%

Katowice 754 459 295 87,27% 91,91% 86,07%

Kaunas 287 155 132 85,02% 90,59% 85,02%

Kavala 88 23 65 94,32% 90,91% 88,64%

Kecskemet 379 239 140 91,56% 94,46% 89,97%

Kielce 483 232 251 89,23% 92,55% 87,99%

Kiel 399 188 211 90,73% 95,49% 89,97%

Kingston-upon-Hull 305 135 170 86,89% 93,77% 86,56%

Kobenhavn 1335 826 509 95,88% 98,13% 95,66%

Koblenz 151 78 73 92,05% 96,69% 92,05%

Koln 513 319 194 94,15% 98,25% 94,15%

Konin 111 31 80 92,79% 92,79% 90,09%

Kosice 249 100 149 90,36% 93,98% 88,35%

Koszalin 239 102 137 92,47% 93,72% 90,79%

Krakow 354 186 168 87,29% 92,66% 85,88%

Larisa 236 89 147 87,71% 92,37% 86,86%

Las_Palmas 77 25 52 89,61% 96,10% 89,61%

Leeds 995 551 444 85,73% 92,16% 85,13%

Leeuwarden 77 36 41 89,61% 96,10% 88,31%

Lefkosia 452 170 282 91,59% 92,04% 88,05%

Leicester 235 121 114 92,34% 97,02% 92,34%

Leipzig 496 209 287 88,31% 90,73% 87,50%

Lens_Lievin 83 51 32 87,95% 92,77% 86,75%

Le_Havre 149 89 60 87,25% 92,62% 85,91%

Liberec 342 190 152 86,84% 92,69% 85,67%

Liepaja 355 132 223 90,42% 92,96% 88,17%


European Commission




08/01/2013

Page 81/96

Lille 206 138 68 85,92% 95,15% 85,44%

Limerick 499 210 289 91,78% 93,39% 89,18%

Limoges 438 229 209 87,90% 93,15% 86,30%

Lincoln 121 50 71 88,43% 89,26% 85,95%

Linköping 559 240 319 93,38% 95,17% 91,06%

Linz 279 126 153 89,96% 96,06% 88,53%

Lisboa 358 223 135 85,47% 91,34% 85,20%

Liverpool 272 206 66 85,66% 92,28% 84,93%

Liège 327 151 176 96,33% 98,78% 96,02%

Ljubljana 518 213 305 89,38% 94,02% 87,84%

Lodz 404 200 204 87,13% 94,31% 85,40%

Logrono 146 56 90 96,58% 97,95% 94,52%

London 2382 1596 786 87,53% 92,91% 86,82%

Lublin 481 287 194 87,94% 93,35% 87,53%

Luxembourg 344 126 218 96,80% 98,55% 96,22%

Lyon 555 279 276 93,87% 94,59% 91,71%

l_Aquila 126 28 98 98,41% 95,24% 94,44%

Madrid 1406 773 633 85,99% 91,75% 85,42%

Magdeburg 563 232 331 88,10% 94,14% 87,57%

Mainz 143 65 78 86,71% 92,31% 85,31%

Malaga 142 82 60 92,25% 95,07% 91,55%

Malmö 350 213 137 86,00% 95,43% 85,71%

Manchester 500 375 125 87,80% 93,80% 86,80%

Maribor 372 145 227 89,25% 95,97% 88,98%

Marseille 164 120 44 90,24% 92,68% 87,20%

Metz 290 123 167 95,86% 97,59% 95,52%

Milano 933 633 300 88,42% 96,25% 88,21%

Miskolc 323 149 174 90,40% 93,81% 89,78%

Modena 92 50 42 94,57% 96,74% 92,39%

Monchengladbach 116 73 43 95,69% 99,14% 95,69%

Montpellier 121 73 48 89,26% 90,08% 85,12%

Murcia 324 169 155 86,11% 91,98% 85,49%

München 908 501 407 87,11% 95,26% 86,23%

Namur 103 38 65 98,06% 99,03% 98,06%

Nancy 233 109 124 89,27% 90,56% 86,27%

Nantes 379 200 179 90,77% 94,46% 88,92%

Napoli 173 123 50 88,44% 95,95% 86,13%

Newcastle upon Tyne 524 251 273 87,98% 93,51% 85,88%

Nice 70 45 25 92,86% 90,00% 88,57%

Nijmegen 59 36 23 89,83% 98,31% 89,83%

Nitra 151 66 85 90,07% 92,05% 88,74%

Nottingham 235 137 98 91,49% 95,74% 90,64%

Nowy Sacz 75 30 45 90,67% 89,33% 86,67%

Nyiregyhaza 198 99 99 90,91% 95,45% 90,91%

Nürnberg 497 293 204 90,95% 95,98% 90,14%

Odense 412 182 230 91,99% 96,60% 91,50%

Olomouc 402 172 230 89,05% 92,54% 87,56%

Olsztyn 351 140 211 91,45% 94,59% 90,31%


European Commission




08/01/2013

Page 82/96

Opole 467 191 276 91,43% 94,65% 90,79%

Oporto 198 123 75 88,89% 93,43% 86,87%

Oradea 48 26 22 89,58% 95,83% 89,58%

Orebro 589 277 312 90,15% 92,36% 87,44%

Orleans 277 126 151 90,97% 96,75% 90,97%

Ostrava 653 356 297 87,60% 92,65% 86,68%

Oulu 969 405 564 90,20% 91,12% 86,48%

Oviedo 251 96 155 94,02% 94,02% 89,64%

Padova 209 123 86 92,82% 98,09% 92,82%

Palermo 194 99 95 87,63% 93,81% 86,60%

Palma_di_Mallorca 258 124 134 90,31% 93,02% 87,60%

Pamplona_Iruna 431 116 315 96,52% 93,74% 91,65%

Panevezys 357 155 202 90,20% 94,96% 89,36%

Pardubice 105 44 61 93,33% 95,24% 91,43%

Paris 2279 1333 946 87,71% 94,30% 86,97%

Patra 207 113 94 95,17% 96,62% 93,72%

Perugia 98 39 59 90,82% 92,86% 87,76%

Pescara 80 36 44 90,00% 93,75% 87,50%

Piatra_Neamt 90 12 78 94,44% 93,33% 92,22%

Pleven 216 74 142 95,83% 96,76% 95,37%

Plock 263 116 147 92,02% 95,82% 91,63%

Plovdiv 194 95 99 87,63% 91,75% 87,11%

Plzen 634 288 346 90,85% 95,43% 89,91%

Poitiers 198 79 119 95,96% 95,96% 93,94%

Ponto Delgada 101 7 94 100,00% 98,02% 98,02%

Portsmouth 89 73 16 86,52% 92,13% 86,52%

Potenza 123 25 98 97,56% 95,12% 94,31%

Poznan 473 244 229 88,16% 91,97% 86,05%

Praha 1015 502 513 87,29% 92,81% 86,11%

Presov 110 53 57 91,82% 94,55% 90,00%

Pécs 81 44 37 91,36% 91,36% 86,42%

Radom 378 217 161 91,27% 92,86% 90,48%

Regensburg 281 110 171 91,10% 93,59% 89,68%

Reggio_di_Calabria 44 20 24 88,64% 95,45% 88,64%

Reims 400 181 219 93,00% 95,75% 92,00%

Rennes 327 149 178 92,05% 96,33% 90,83%

Riga 1479 889 590 87,49% 90,67% 84,99%

Roma 1162 618 544 86,57% 91,91% 85,28%

Rotterdam 526 363 163 85,93% 95,63% 85,55%

Rouen 270 148 122 86,67% 92,59% 85,19%

Ruse 135 50 85 91,11% 92,59% 88,89%

Rzeszów 162 75 87 88,89% 93,83% 87,65%

s'Gravenhage 131 92 39 87,79% 91,60% 87,79%

Saarbrucken 297 149 148 88,89% 92,26% 87,54%

Saint-Etienne 93 52 41 93,55% 97,85% 93,55%

Salerno 95 32 63 89,47% 91,58% 86,32%

Salzburg 201 70 131 91,54% 94,03% 88,06%

Santander 75 29 46 90,67% 92,00% 85,33%


European Commission




08/01/2013

Page 83/96

Santa_Cruz_de_Tenerife 78 35 43 94,87% 96,15% 92,31%

Santiago_de_Compostela 125 40 85 96,00% 88,00% 87,20%

Sassari 266 127 139 89,47% 91,73% 87,22%

Schwerin 849 359 490 87,16% 92,82% 86,34%

Setubal 81 32 49 93,83% 95,06% 92,59%

Sevilla 351 153 198 89,46% 94,02% 88,60%

Sheffield 378 203 175 88,36% 93,39% 87,30%

Sibiu 87 24 63 93,10% 91,95% 88,51%

Sofia 552 248 304 92,75% 95,11% 91,85%

Stara Zagora 88 30 58 89,77% 92,05% 88,64%

Stockholm 1265 721 544 88,14% 92,17% 86,64%

Stoke-on-trent 193 122 71 91,19% 95,85% 90,67%

Strasbourg 298 146 152 94,97% 96,64% 94,30%

Stuttgart 613 303 310 90,21% 93,31% 88,42%

Suwalki 124 51 73 87,90% 92,74% 86,29%

Szczecin 976 390 586 88,11% 91,60% 86,27%

Szeged 121 63 58 89,26% 93,39% 88,43%

Szekesfeharvar 145 68 77 90,34% 98,62% 90,34%

Tallinn 1127 576 551 88,82% 90,06% 85,00%

Tampere 321 148 173 88,47% 93,77% 87,54%

Taranto 184 83 101 90,76% 91,30% 88,04%

Targu_Mures 72 25 47 94,44% 97,22% 94,44%

Tartu 351 145 206 91,45% 91,74% 86,89%

Thessaloniki 150 73 77 88,67% 92,00% 85,33%

Tilburg 72 45 27 90,28% 98,61% 90,28%

Timisoara 61 39 22 100,00% 100,00% 100,00%

Toledo 399 98 301 94,74% 96,49% 93,48%

Torino 292 173 119 88,36% 92,12% 85,62%

Torun 305 186 119 86,56% 91,15% 85,90%

Toulon 89 62 27 88,76% 93,26% 85,39%

Toulouse 626 346 280 90,89% 95,37% 89,30%

Tours 268 126 142 92,16% 97,01% 91,04%

Trencin 94 35 59 94,68% 97,87% 94,68%

Trento 88 25 63 98,86% 95,45% 95,45%

Trier 126 41 85 87,30% 93,65% 85,71%

Trieste 34 19 15 91,18% 91,18% 85,29%

Trnava 96 43 53 94,79% 96,88% 94,79%

Turku 316 159 157 87,03% 93,04% 86,39%

Umea 767 196 571 96,48% 87,35% 85,40%

Uppsala 812 293 519 92,98% 91,13% 87,56%

Usti_nad_Labem 171 82 89 87,72% 89,47% 86,55%

Utrecht 84 60 24 97,62% 98,81% 97,62%

Valencia 269 165 104 87,36% 93,68% 85,87%

Valladolid 425 211 214 89,18% 95,06% 87,76%

Valletta 56 33 23 91,07% 96,43% 91,07%

Varna 175 66 109 93,14% 91,43% 89,71%

Venezia 206 102 104 93,69% 97,57% 93,69%

Verona 192 94 98 92,71% 95,31% 92,71%


European Commission




08/01/2013

Page 84/96

Vidin 75 22 53 93,33% 96,00% 93,33%

Vigo 143 77 66 88,81% 91,61% 86,71%

Vilnius 1042 534 508 87,14% 92,51% 85,80%

Vitoria_Gasteiz 238 115 123 94,96% 94,54% 91,18%

Volos 41 21 20 85,37% 92,68% 85,37%

Warszawa 945 631 314 85,40% 88,89% 85,19%

Waterford 68 37 31 91,18% 94,12% 91,18%

Weimar 116 48 68 88,79% 91,38% 87,93%

Wien 924 497 427 87,99% 95,02% 87,66%

Wiesbaden 165 91 74 88,48% 93,94% 86,67%

Wolverhampton 108 52 56 94,44% 97,22% 93,52%

Worcester 148 55 93 93,92% 94,59% 93,92%

Wrexham 167 71 96 89,22% 91,62% 87,43%

Wroclaw 1071 615 456 89,36% 94,96% 88,24%

Wuppertal 69 49 20 98,55% 100,00% 98,55%

Zaragoza 565 309 256 89,03% 95,93% 88,50%

Zielona_Gora 257 123 134 91,44% 92,22% 87,94%

Zilina 215 92 123 92,09% 93,95% 90,23%

Zlin 119 44 75 93,28% 91,60% 89,08%


European Commission




08/01/2013

Page 85/96

Appendix 3 : Land/Use Cover statistics

Area percentage by Country level 3

Countries Area (km²) 11100 11210 11220 11230 11240 11300 12100 12210 12220 12230 12300 12400 13100 13300 13400 14100 14200 20000 30000 50000

Austria AT 11426,13 0,39 1,96 2,40 1,32 0,11 0,59 2,09 0,25 2,05 0,30 0,09 0,21 0,28 0,13 0,11 0,72 0,56 52,25 32,80 1,39

Belgium BE 5590,52 1,51 4,18 6,07 5,72 1,63 0,54 5,62 0,80 3,66 0,68 1,45 0,39 0,38 0,19 0,30 1,71 0,92 48,73 13,57 1,96

Bulgaria BG 10209,32 0,98 2,21 0,98 0,27 0,01 0,14 2,48 0,06 1,61 0,20 0,07 0,16 0,31 0,07 0,07 0,37 0,32 60,80 26,96 1,93

Cyprus CY 2710,23 1,33 2,06 0,47 0,28 0,18 0,41 1,51 0,07 1,19 0,00 0,00 0,29 0,46 0,05 0,24 0,14 0,15 90,71 0,31 0,15

Czech Republic CZ 28295,68 1,03 3,04 1,13 0,20 0,02 0,32 2,28 0,09 1,61 0,27 0,00 0,11 0,33 0,15 0,06 0,62 0,51 53,28 33,93 1,01

Germany DE 101031,46 1,21 4,28 1,63 0,34 0,02 0,40 3,67 0,26 2,31 0,33 0,09 0,13 0,45 0,11 0,10 0,80 1,14 53,38 27,61 1,72

Denmark DK 17188,46 0,32 2,87 2,12 1,11 0,09 1,72 2,52 0,16 1,99 0,13 0,13 0,16 0,20 0,12 0,07 0,64 1,07 69,10 13,63 1,86

Estonia EE 7336,77 0,18 0,83 1,04 0,36 0,01 0,95 1,05 0,00 0,99 0,10 0,09 0,14 0,22 0,13 0,04 0,30 0,13 40,57 51,69 1,16

Spain ES 47965,42 0,90 0,86 0,84 1,07 0,52 0,44 2,17 0,26 2,57 0,12 0,06 0,21 0,43 0,63 0,13 0,47 0,42 69,78 17,37 0,74

Finland FI 11008,48 0,13 0,39 0,80 1,82 1,71 1,25 1,69 0,19 1,78 0,13 0,10 0,21 0,41 0,14 0,04 0,91 0,53 24,74 56,30 6,75

France FR 61064,88 1,05 3,95 2,32 1,02 0,14 0,63 3,31 0,25 2,65 0,28 0,10 0,23 0,26 0,18 0,13 0,83 0,77 57,36 23,16 1,38

Greece GR 9563,02 1,83 2,37 1,55 1,04 0,16 0,47 2,80 0,15 2,75 0,07 0,17 0,31 0,43 0,16 0,13 0,39 0,30 72,48 11,70 0,74

Hungary HU 12028,59 2,34 4,96 1,29 0,29 0,01 0,41 2,91 0,14 1,61 0,27 0,01 0,22 0,27 0,26 0,17 0,55 0,45 60,97 21,49 1,39

Ireland IE 12746,69 0,13 1,92 1,83 0,81 0,04 1,75 1,92 0,11 1,56 0,08 0,06 0,12 0,47 0,17 0,11 0,52 1,29 75,70 9,05 2,36

Italia IT 35072,33 1,15 2,26 1,82 1,71 0,69 1,17 4,07 0,23 3,10 0,25 0,16 0,18 0,35 0,28 0,18 0,52 0,55 61,73 17,57 2,04

Lithuania LT 8095,35 0,84 1,99 0,82 0,08 0,01 0,90 1,54 0,07 1,16 0,10 0,00 0,07 0,12 0,21 0,10 0,85 0,17 50,86 38,04 2,06

Luxembourg LU 2596,59 0,31 2,63 1,56 0,10 0,00 0,14 2,43 0,25 2,44 0,22 0,02 0,15 0,23 0,14 0,07 0,23 0,49 51,60 36,51 0,48

Latvia LV 9044,75 0,25 0,80 0,84 0,51 0,08 0,78 1,26 0,00 1,14 0,12 0,10 0,12 0,20 0,03 0,09 0,41 0,35 38,57 50,86 3,47

Malta MT 315,58 3,41 4,37 2,72 2,17 1,53 0,70 6,40 0,00 6,71 0,00 0,59 1,09 1,33 0,14 0,34 0,45 0,77 66,64 0,05 0,58

Netherlands NL 7288,39 3,64 5,17 2,24 0,83 0,10 1,02 6,93 0,84 4,27 0,38 0,97 0,45 0,16 0,61 0,41 2,57 2,06 47,73 11,48 8,15

Poland PL 68644,09 1,54 3,06 0,59 0,04 0,00 0,79 1,95 0,04 1,49 0,23 0,02 0,09 0,22 0,13 0,07 0,40 0,51 57,22 29,92 1,70

Portugal PT 5469,78 3,33 4,29 2,61 1,62 0,33 0,73 3,97 0,34 3,72 0,16 0,18 0,27 0,46 0,46 0,43 0,94 0,61 49,65 22,48 3,42

Romania RO 5111,96 5,20 3,04 0,29 0,04 0,00 0,20 4,78 0,02 1,66 0,40 0,06 0,37 0,20 0,26 0,18 0,53 0,26 63,19 16,88 2,42

Suede SE 41306,67 0,04 0,10 0,38 1,24 1,57 0,82 0,99 0,11 1,29 0,09 0,04 0,08 0,16 0,07 0,02 0,51 0,54 22,10 62,13 7,70

Slovenia SL 4722,54 0,08 1,05 1,74 1,38 0,16 1,59 1,37 0,15 2,29 0,15 0,00 0,02 0,13 0,14 0,05 0,15 0,16 35,88 52,82 0,70

Slovakia SK 8662,56 0,88 2,81 0,69 0,16 0,03 0,12 2,30 0,14 1,35 0,22 0,02 0,14 0,21 0,14 0,06 0,28 0,45 47,45 41,71 0,83

United-Kingdom UK 51843,40 0,29 4,72 3,83 1,21 0,12 0,83 4,07 0,18 2,65 0,23 0,13 0,17 0,42 0,14 0,11 1,64 2,30 66,81 9,28 0,87

Average 1,27 2,67 1,65 0,99 0,34 0,74 2,89 0,19 2,28 0,20 0,17 0,23 0,34 0,19 0,14 0,68 0,66 55,16 27,01 2,18


European Commission




08/01/2013

Page 86/96


Countries Area (km²) 11 12 13 14 20 30 50

Austria AT 11426,13 6,77 4,99 0,52 1,28 52,25 32,80 1,39

Belgium BE 5590,52 19,65 12,60 0,86 2,63 48,73 13,57 1,96

Bulgaria BG 10209,32 4,60 4,58 0,45 0,69 60,80 26,96 1,93

Cyprus CY 2710,23 4,72 3,06 0,75 0,29 90,71 0,31 0,15

Czech Republic CZ 28295,68 5,75 4,36 0,55 1,13 53,28 33,93 1,01

Germany DE 101031,46 7,89 6,81 0,66 1,94 53,38 27,61 1,72

Denmark DK 17188,46 8,22 5,08 0,39 1,72 69,10 13,63 1,86

Estonia EE 7336,77 3,37 2,37 0,39 0,44 40,57 51,69 1,16

Spain ES 47965,42 4,63 5,39 1,19 0,89 69,78 17,37 0,74

Finland FI 11008,48 6,10 4,09 0,59 1,44 24,74 56,30 6,75

France FR 61064,88 9,11 6,82 0,57 1,60 57,36 23,16 1,38

Greece GR 9563,02 7,42 6,25 0,72 0,69 72,48 11,70 0,74

Hungary HU 12028,59 9,29 5,16 0,70 1,00 60,97 21,49 1,39

Ireland IE 12746,69 6,48 3,85 0,75 1,80 75,70 9,05 2,36

Italia IT 35072,33 8,80 7,99 0,81 1,07 61,73 17,57 2,04

Lithuania LT 8095,35 4,64 2,95 0,43 1,02 50,86 38,04 2,06

Luxembourg LU 2596,59 4,76 5,50 0,44 0,72 51,60 36,51 0,48

Latvia LV 9044,75 3,28 2,73 0,33 0,76 38,57 50,86 3,47

Malta MT 315,58 14,91 14,79 1,81 1,22 66,64 0,05 0,58

Netherlands NL 7288,39 13,01 13,84 1,18 4,62 47,73 11,48 8,15

Poland PL 68644,09 6,03 3,80 0,42 0,91 57,22 29,92 1,70

Portugal PT 5469,78 12,90 8,65 1,35 1,55 49,65 22,48 3,42

Romania RO 5111,96 8,79 7,28 0,65 0,79 63,19 16,88 2,42

Suede SE 41306,67 4,15 2,61 0,26 1,05 22,10 62,13 7,70

Slovenia SL 4722,54 6,00 3,98 0,32 0,30 35,88 52,82 0,70

Slovakia SK 8662,56 4,69 4,17 0,41 0,74 47,45 41,71 0,83

United-Kingdom UK 51843,40 10,98 7,43 0,68 3,94 66,81 9,28 0,87

Average 7,66 5,97 0,67 1,34 55,16 27,01 2,18


European Commission




08/01/2013

Page 87/96


Countries Area (km²) 1 2 3 5

Austria AT 11426,13 13,56 52,25 32,80 1,39

Belgium BE 5590,52 35,74 48,73 13,57 1,96

Bulgaria BG 10209,32 10,32 60,80 26,96 1,93

Cyprus CY 2710,23 8,83 90,71 0,31 0,15

Czech Republic CZ 28295,68 11,78 53,28 33,93 1,01

Germany DE 101031,46 17,29 53,38 27,61 1,72

Denmark DK 17188,46 15,42 69,10 13,63 1,86

Estonia EE 7336,77 6,57 40,57 51,69 1,16

Spain ES 47965,42 12,10 69,78 17,37 0,74

Finland FI 11008,48 12,21 24,74 56,30 6,75

France FR 61064,88 18,09 57,36 23,16 1,38

Greece GR 9563,02 15,08 72,48 11,70 0,74

Hungary HU 12028,59 16,15 60,97 21,49 1,39

Ireland IE 12746,69 12,89 75,70 9,05 2,36

Italia IT 35072,33 18,66 61,73 17,57 2,04

Lithuania LT 8095,35 9,04 50,86 38,04 2,06

Luxembourg LU 2596,59 11,42 51,60 36,51 0,48

Latvia LV 9044,75 7,11 38,57 50,86 3,47

Malta MT 315,58 32,73 66,64 0,05 0,58

Netherlands NL 7288,39 32,65 47,73 11,48 8,15

Poland PL 68644,09 11,16 57,22 29,92 1,70

Portugal PT 5469,78 24,45 49,65 22,48 3,42

Romania RO 5111,96 17,51 63,19 16,88 2,42

Suede SE 41306,67 8,07 22,10 62,13 7,70

Slovenia SL 4722,54 10,60 35,88 52,82 0,70

Slovakia SK 8662,56 10,01 47,45 41,71 0,83

United-Kingdom UK 51843,40 23,04 66,81 9,28 0,87

Average 15,65 55,16 27,01 2,18


European Commission




08/01/2013

Page 88/96

Appendix 4 : LUZ Delivery Report - Hamburg example

LUZ HAMBURG_DE

GENERAL INFORMATION Area of product 7206,95 km²

Format (geodatabase esri) Ok

SOIL SEALING cover and projection Ok

COTS validation Ok

Coordinate system reference Projection: UTM 32 Central Meridian: 9° Latitude of Origin: 0° False Easting: 500000,00 False Northing: 0,00 Ellipsoid: WGS84

IMAGE data used Satellite &

Sensor Image Name Date (y/m/d) Remark (e.g.

clouds)

Spot 5 2,50 m 50492401006171100402B8 (PS) 2010/06/17 Pan-sharpening Images created by

SPOT IMAGE 50502400908201045112B0 (PS) 2009/08/20

50502410908201045192B0 (PS) 2009/08/20

50502420908201045272B0 (PS) 2009/08/20

50532410909011014392B0 (PS) 2009/09/01

50532420909011014472B0 (PS) 2009/09/01

RapidEye 5 m 2009-04-10T110514_RE2_3A-NAC_6212971_110437 (XS) 2009/04/10

2009-04-18T111046_RE5_3A-NAC_6212756_110437 (XS) 2009/04/18

2009-04-18T111047_RE5_3A-NAC_6212766_110437 (XS) 2009/04/18

2009-04-18T111049_RE5_3A-NAC_6212773_110437 (XS) 2009/04/18

2009-04-18T111050_RE5_3A-NAC_6212700_110437 (XS) 2009/04/18

2009-04-18T111050_RE5_3A-NAC_6212763_110437 (XS) 2009/04/18

2009-04-18T111052_RE5_3A-NAC_6212771_110437 (XS) 2009/04/18

2009-04-18T111053_RE5_3A-NAC_6212699_110437 (XS) 2009/04/18

2009-04-19T111354_RE1_3A-NAC_6213203_110437 (XS) 2009/04/19

2009-04-19T111358_RE1_3A-NAC_6213202_110437 (XS) 2009/04/19

2009-05-13T111747_RE1_3A-NAC_6212970_110437 (XS) 2009/05/29

2009-05-29T111207_RE3_3A-NAC_6212767_110437 (XS) 2009/05/29

2009-05-29T111211_RE3_3A-NAC_6212758_110437 (XS) 2009/05/29

2009-05-29T111213_RE3_3A-NAC_6212760_110437 (XS) 2009/05/29

2009-05-29T111214_RE3_3A-NAC_6212768_110437 (XS) 2009/05/29

2009-05-29T111217_RE3_3A-NAC_6212757_110437 (XS) 2009/05/29

2009-05-29T111221_RE3_3A-NAC_6212974_110437 (XS) 2009/05/29

2009-05-31T111327_RE5_3A-NAC_6212764_110437 (XS) 2009/05/31

2009-05-31T111330_RE5_3A-NAC_6212765_110437 (XS) 2009/05/31

2009-05-31T111330_RE5_3A-NAC_6212973_110437 (XS) 2009/05/31

2009-07-03T110649_RE5_3A-NAC_6212772_110437 (XS) 2009/07/03

2009-07-03T110652_RE5_3A-NAC_6212769_110437 (XS) 2009/07/03

2009-07-03T110653_RE5_3A-NAC_6212770_110437 (XS) 2009/07/03

2009-07-03T110656_RE5_3A-NAC_6212755_110437 (XS) 2009/07/03

2009-07-05T112955_RE1_3A-NAC_6212762_110437 (XS) 2009/07/05

2009-07-27T111044_RE5_3A-NAC_6212972_110437 (XS) 2009/07/27


European Commission




08/01/2013

Page 89/96

Ancillary data used (thematic data, satellite images, aerial photos, city maps)

Id. Data

source/type

Title

(if relevant)

Production date

(y/m/d)

Scale

(spatial detail) Remarks

Google Earth

City map Falk Hamburg 1 : 39000e

CAPI/ DIGITAL CLASSIFICATION

Photointerpreter(s)

Name/surname interpretation

start (y/m/d)

end (y/m/d)

Remarks

Bonnemains diane Sladkowski audrey Bitschene antoine Lecocq aurelien Prevost aurelien Stievenard christophe Auger emeric Fretin david Cornuet jeanne

11/07/21 11/08/19 - Rapideye source image used under the clouds on the other datas

- Airbus factory and runway test, not an airport but 121 53°32’18’’N / 09°50’02’’E

- old burying ground, now weather station area 121

53°30’36’’N / 10°01’47’’E

Internal quality control

Internal quality check

Date of control (y/m/d)

Result Remarks (errors, corrections, etc.)

Courmont Yoann

11/08/22 OK Helgoland island (80km from the border of the rest of the LUZ) with no images code 999

54°10’55’’N / 07°53’11’’E

Quality assesment

Internal quality check

Date of control (y/m/d)

Result Remarks (errors, corrections, etc.)

IGNFI 11/08/23 ok

CONFUSION MATRIX Urban strata accuracy calculation The sample population is that of the entire LUZ and non urban classes are aggregated in class 600. The main advantage of this method is to provide both omission and commission errors on the urban strata (omission - commission class 1xx / class 600).


European Commission




08/01/2013

Page 90/96

IFI

110 113 121 122 123 124 131 133 134 141 142 600 TotalProducer

accuracy

Commission

error

SIRS 110 437 1 11 4 2 5 2 8 470 93.0% 7.0%

113 1 0 1 2 0.0% 100.0%

121 17 143 9 2 1 3 4 2 181 79.0% 21.0%

122 24 24 100.0% 0.0%

123 16 18 34 52.9% 47.1%

124 7 7 100.0% 0.0%

131 11 2 13 84.6% 15.4%

133 3 3 100.0% 0.0%

134 1 4 3 2 10 30.0% 70.0%

141 2 3 48 2 3 58 82.8% 17.2%

142 3 2 1 70 76 92.1% 7.9%

600 3 4 2 2 1 2 22 497 533 93.2% 6.8%

Total 464 1 183 40 20 9 11 7 5 78 78 515 1411

Producer

accuracy 94.2% 0.0% 78.1% 60.0% 90.0% 77.8% 100.0% 42.9% 60.0% 61.5% 89.7% 96.5%

Omission

error 5.8% 100.0% 21.9% 40.0% 10.0% 22.2% 0.0% 57.1% 40.0% 38.5% 10.3% 3.5%

Overall accuracy = 89.4%

CONFUSION MATRIX (URBAN)


European Commission




08/01/2013

Page 91/96

IFI

100 200 300 500 TotalProducer

accuracy

Commission

error

SIRS 100 860 12 5 1 878 97.9% 2.1%

200 13 373 1 387 96.4% 3.6%

300 19 6 87 112 77.7% 22.3%

500 4 30 34 88.2% 11.8%

Total 896 391 93 31 1411

Producer

accuracy 96.0% 95.4% 93.5% 96.8%

Omission

error 4.0% 4.6% 6.5% 3.2%


CONFUSION MATRIX (RURAL)


European Commission




08/01/2013

Page 92/96

IFI

110 113 121 122 123 124 131 133 134 141 142 200 300 500 TotalProducer

accuracy

Commission

error

SIRS 110 437 1 11 4 2 5 2 5 3 470 93.0% 7.0%

113 1 0 1 2 0.0% 100.0%

121 17 143 9 2 1 3 4 1 1 181 79.0% 21.0%

122 24 24 100.0% 0.0%

123 16 18 34 52.9% 47.1%

124 7 7 100.0% 0.0%

131 11 2 13 84.6% 15.4%

133 3 3 100.0% 0.0%

134 1 4 3 2 10 30.0% 70.0%

141 2 3 48 2 2 1 58 82.8% 17.2%

142 3 2 1 70 76 92.1% 7.9%

200 2 2 2 1 2 4 373 1 387 96.4% 3.6%

300 1 18 6 87 112 77.7% 22.3%

500 2 2 30 34 88.2% 11.8%

Total 464 1 183 40 20 9 11 7 5 78 78 391 93 31 1411

Producer

accuracy 94.2% 0.0% 78.1% 60.0% 90.0% 77.8% 100.0% 42.9% 60.0% 61.5% 89.7% 95.4% 93.5% 96.8%

Omission

error 5.8% 100.0% 21.9% 40.0% 10.0% 22.2% 0.0% 57.1% 40.0% 38.5% 10.3% 4.6% 6.5% 3.2%


CONFUSION MATRIX (OVERALL)


European Commission




08/01/2013

Page 93/96

TOPOLOGICAL QUALITY CHECK The topological quality check of all of the LUZs was based on the same rules (see diagramme).

Thematic QC on the CAPI phase;

Creation of Arcinfo’s Topology;

Geometric check and validation of Geodatabase’s Topology.

However, in some cases, especially in large LUZs, the geometric check and arcgis’s topology failed. The main cause was the size and shape of a very large and complex polygon composed of roads (code 12220). So, to avoid this problem, SIRS created a 10 km size grid on these LUZs which split this large polygon into smaller ones. Only the 12220 code roads are split using this method. After this operation, the geometric and topology checks passed successfully.

Internal quality

check Date of control

(y/m/d) Result Remark (errors, corrections, etc.)

SIRS 11/08/31

ok


European Commission




08/01/2013

Page 94/96

Appendix 5 : LUZ Map - Manchester (UK) example

Date post:	25-Jan-2017
Category:	Documents
Upload:	dokhue
View:	227 times
Download:	0 times