1. Provision of interoperable datasets to open GI to EU communities Magistrato alle Acque di Venezia Project funded by eContentplus Programme Considerations for Harmonising Cross-Border Geospatial Datasets Gobe Hobona 1, Carmelo Attardo 2, Robert Laurini 3, Mike Jackson 1, Maria Pla 4, Stefania de Zorzi 5, Anette Breu 6, Catherine Roussey 7, Alina Kmiecik 8 1 - Centre for Geospatial Science at the University of Nottingham, UK 2 - Intergraph Corporation, Italy 3 - Institut National des Sciences Applique'es de Lyon, France 4 - Institut Cartogràfic de Catalunya, Catalonia, Spain 5 - Corila, Venice, Italy 6 - Intergraph Corporation, Germany 7 - CNRS, LIRIS UMR5205, Université de Lyon, France 8 - Intergraph Corporation, Poland

2. Provision of interoperable datasets to open GI to EU communities Project Objectives To facilitate the sharing of base cartography with the intention of making it more accessible and interoperable for purposes of aggregation To provide reference data for specific themes defined in the INSPIRE directive To create a GIS4EU server that will enable access to data from multiple providers through standard web services To enable spatial databases provided by partners to be shared through web services To analyse lack of information at dataset level To validate INSPIRE Implementation Rules and make suggestions for addressing critical issues To examine, the usability of data and metadata implementation rules during the design stages, allowing for obstacles and potential problems to be identified. To provide guidelines and training

3. Provision of interoperable datasets to open GI to EU communities

4. Merging Uniform Scale Cross-border Datasets Illustration of datasets at the same scale Slovakia (Blue) Hungary (Green)

5. Provision of interoperable datasets to open GI to EU communities Merging Uniform Scale Cross-border Datasets (1) RULE 1: Each dataset produced through aggregation should be accompanied by metadata that describes the lineage of the aggregated dataset. RULE 2: The quality of the input datasets shall be evaluated before any aggregation is undertaken by examining the metadata of the input datasets according to the five elements of data quality listed in ISO 19114… RULE 3: If an input dataset is not in ETRS89, then the requirements listed in the INSPIRE specification for Coordinate Reference Systems should be applied. RULE 4: If Hausdorff Distance between BA and BB < epsilon, then BA and BB are both replaced by Midline(BA,BB); if and only if, the aggregated datasets are not to be used in a legal context… RULE 5: If two adjacent polygonal features have been force fit and they are of EXACTLY the same type, then they can be replaced by their union.

6. Provision of interoperable datasets to open GI to EU communities Merging Uniform Scale Cross-border Datasets (2) RULE 6: If topological connectivity between two zones is explicitly known and the quality of the input datasets is different, then the topologically equal edges may be replaced with a shared edge taking the boundary data with the higher level of quality. RULE 7: If a linear feature is represented in two adjacent datasets, then coincidence of the end nodes should be checked… RULE 8: If the end nodes are not coincident, then the nodes should be translated to their midpoint. RULE 9: If the generated line is too jagged(i.e. zig zag formation), then other vertices should be moved to make the line smooth. RULE 10: If adjacent raster datasets are of different geometries, then transform both into triangles and fill the intermediary zones by additional triangles. A regular grid may then be interpolated from the triangular irregular network (TIN); if and only if, the aggregated datasets are not to be used in a legal context…

7. Provision of interoperable datasets to open GI to EU communities Merging Uniform Scale Cross-border Datasets (3) RULE 11: If the meaning of a feature type or attribute is known to be similar to one in the INSPIRE data model, then the feature type or attribute name will be modified to adopt the name of its INSPIRE equivalent. Similarly the data types (floating point, character strings, integers) will adopt those of the INSPIRE data model. RULE 12: If the complete codelist of an input attribute is known to be a subset of the multilingual INSPIRE/GIS4EU glossary AND if the attribute does NOT hold placenames, then the values shall be translated to English. RULE 13: if a placename has several names, then use the local name. RULE 14: If the attributes of features in input datasets show quantities or magnitudes AND if the quantities or magnitudes are not in the same units as the output data, then a conversion factor shall be applied to the attributes on the input datasets to match the unit used by the output dataset. RULE 15: If any features contain exactly the same attributes AND the attributes have the same names and type, then the duplicated features should be deleted If and Only If the semantics of the dataset are not affected.

8. Provision of interoperable datasets to open GI to EU communities Merging Uniform Scale Cross-border Datasets (4) RULE 16: If two raster datasets being merged are known to have different classifications (or codelists), then the pixel values of one of the datasets has to be reclassified into the classification of the output raster; if and only if the semantics of the classification labels are known. RULE 17: If character encoding of an input dataset is not in Unicode UTF-8, then the dataset will be converted to UTF-8. RULE 18: if the objects in input datasets do not have identifiers, then they will be given unique identifiers in the merged dataset RULE 19: if the objects have been given identifiers prior to merging, then new unique identifiers will be generated for each spatial object. RULE 20: if a single feature is artificially cut into two object parts, each belonging to a different dataset, then the two objects should be reconstructed into a single new object through a geometric union with attributes copied from one dataset to the other. The newly reconstructed object should have a unique ID.

9. Provision of interoperable datasets to open GI to EU communities Merging Multiple Scale Cross-border Datasets Illustration of datasets at different scales : Slovakia province(Left) Hungary (Right)

10. Provision of interoperable datasets to open GI to EU communities Merging Multiple Scale Cross-border Datasets (1) Rule 1: Requirements of the generalized data should be formalized to allow automation of the evaluation of the results and the resolution of conflicts. Rule 2: Systems for evaluating the results of the automated generalization process should be provided, and used as input for any further optimization. RULE 3: If the datasets are of significantly different scales then the merged dataset may appear clattered. To improve clarity, some of the least relevant feature types should be selected and removed. Relevance can be judged based on theme, for example in a transport network dataset, rivers are more relevant than buildings (e.g. in Venice, Italy), whereas rivers are less relevant than roads (e.g. in Turin, Italy). Rule 4: Elements that are too short or too small at the target scale may be deleted, even if they are relevant to the theme of the output dataset…

11. Provision of interoperable datasets to open GI to EU communities Merging Multiple Scale Cross-border Datasets (2) Rule 5: If two adjacent features have been force fit and they are of exactly the same type, then they can be replaced by their union. Rule 6: If line simplification is applied on line shared by different objects, the process must guarantee the same result for all objects including the processed line. Rule 7: Where the input datasets also contain symbols, the Legend of the aggregated features should be consistent with the Legend of the output map in scale and symbol. Rule 8: Often when merging datasets of significantly different scales, it is necessary to amalgamate some of the features, for example a big area covered by a set of small lakes may be represented by some aggregations of some of the smaller lakes… Ordnance Survey Crown Copyright

12. Provision of interoperable datasets to open GI to EU communities Merging Multiple Scale Cross-border Datasets (3) Rule 10:For area or linear features, the geometries can be collapsed. For example, a river could be represented as a line along the centreline rather than as the polygon defined by the river bank, or a road service area could be represented as a point rather than as an area. Rule 11: When Collapsing polygonal features into linear features, if the end nodes are not coincident, then the nodes should be translated to their midpoint. Rule 12: When the difference between the scales of datasets is small, simplification of geometries should be considered. This involves reducing the vertices or coordinates within a geometry resulting in a ‘smoother’ feature. However, care should be taken that the simplified feature maintains topological consistency in relation to nearby features (Laurini, 1998). For example, a simplified road should not introduce overlaps with an adjacent river.

13. Provision of interoperable datasets to open GI to EU communities Discussion Each modification of geometry or alphanumeric attributes can lead to degradation in data quality. Therefore there is need for harmonisation processes to include a validation step. Due to the variety of rules, there is a need for more transparency from software vendors about which rules they adopt in their products. Perhaps leading to a meta(data) standard for harmonisation software? Inadequate metadata of existing datasets potentially reduces the quality of harmonisation

14. Provision of interoperable datasets to open GI to EU communities Conclusion Majority of harmonisation rules are applicable to existing datasets (taking GIS4EU datasets as a sample) There is motivation for a Harmonisation ‘Best Practice’ specification from INSPIRE Objective approaches for evaluating the quality of the harmonisation results are needed