European Commission logo
INSPIRE Community Forum

Interpretation/Implementation of GRIDs for raster data (especially elevation)

Dear INSPIRE community!

I'm a GIS technician working for the Federal Office of Metrology and Surveying (BEV) of Austria. With the attentive help of the Italy-based company GeoSolutions we are implementing the provision of our INSPIRE relevant raster data.

The inspire directive describes in "2.2. Grids" a geographical grid based on the ETRS89-LAEA CRS. Raster data has to align with it.
In section "2.2.2(.1) Other Grids" it is stated that other grids may be specified in this Annex for specific data themes. There is no other grid specified in this Annex for the data theme elevation. There is another grid specified in Annex D of the TG for elevation data, which I interpret as recommendation, not an obligation, since it is not mentioned in the Annex of the original directive.

The software used by us is GeoServer, which is extended with necessary functionality to serve INSPIRE conformant data. This is done mainly by contracting the company GeoSolutions which is adding these components to the open-source-code of GeoServer, ready for everyone (not just us) to be used.

(following, I use the terms "foreward" and "backward" to phrase my explanation, I apologize, if these have other meanings in your eyes)
To my understanding, transformation performed by GeoServer (and many other GIS software), is done in a "foreward" way: The user is defining an area-of-interest, which is extracted from original data and then reprojected into the desired CRS. A slight misalignment with the original bounding-box is happening, because of reprojection errors.

In contrast, when using a "backwards" transformation (as was already described by other posts here, e.g. https://inspire.ec.europa.eu/forum/discussion/view/4609/transformation-of-gridded-reporting-data-from-national-grids-to-the-inspire-equal-area-grid#elgg-object-4920) , the target grid is set before the reprojection occurs. By sampling the original data, the target grid is filled with values of the original grid. There is no visible (!) reprojection error.
I already could re-enact a manual backwards transformation by just using basic QGIS Processing tools.

The question that did arise was, how exactly do other INSPIRE members implement the geographical grid as described in 2.2 of the directive? Is this grid really interpreted that strictly? Do other member states see the need for a reprojection without visible error?

I would like to be able to reproject INSPIRE data on-the-fly without the need to pre-transform the original data and would be glad about comments on this issue!

Best Regards,
Markus Mayr

  • Markus Mayr

    Dear All,

    I now have implemented the prototype of a software that is transforming any raster data source to a data set that is stored in EPSG:3035 and is georeferenced in a way that it alignes with the affordance of the INSPIRE statistical GRIDs.

    Since the minimum resolution for raster data to be provided in the form of statistical grids is 1 meter, this dataset can be stored additionally to the original data without consuming too much space. Other resolution levels can also be computed and need even less space for storage.

    By using this data as a unique layer (or ImagePyramid, in case of multiple resolutions) in GeoServer, different map sheets can be requested by either using WCS or WPS.

    The provision of metadata is a different story, but this should be solveable, too: All our available metadata is already stored in a self-developed system based on a PostGIS database. So, whatever form or layout is necessary, it should be possible to implement this transformation. (I'm thinking of e.g. a database view for each INSPIRE grid-system, conflating and transforming metadata of intersecting regions)

    Best Regards,
    Markus Mayr