European Commission logo
INSPIRE Community Forum

How to implement the concept of coverage aggregation

Introduction

This discussion topic was originated in the following thread dedicated to publish Experiences on encoding of Elevation and Orthoimagery coverages.

After the identification of the issue some dicussion took place in the topic How to implement tiling / model mosaic elements, coverages and coverage aggregations in GMLCOV files.

However the topic was partially mixed with other ones, which lead to some misunderstandings.

During the Workshop "Implementation and potential of INSPIRE coverage data and WCS" (30th September 2016 - INSPIRE Conference 2016, Barcelona) there was the opportunity to meet with the community of Thematic Cluster #3, discuss face-to-face and clearly delimitate the most important issues on encoding of coverages - being this topic one of them.

Summary

The benefit of modelling “coverage aggregation” in INSPIRE raster data models shall be examined. In principle it was included by TWG-OI in the INSPIRE TG on OI as a result of a request from ESA as a user requirement (also included in the INSPIRE TG on EL).

An “aggregated coverage” is a collection of contributions from different individual coverages. Each of them has a defined “contributing footprint” in relation to the aggregated coverage (mosaic), that determines which of its pixels are present in the mosaic. This makes sense from a satellite operator’s perspective where individual scenes taken should be visible to the end-user. End users need uncomplicated answers to their domain-oriented questions, and the extra complexity that is added by an “aggregated coverage” is not paired by an extra user benefit for the majority of users – a coherent orthoimage mosaic is what users in most cases expect.

The coverages to be aggregated are determined by the tiling schema, which is again selected by the data provider (tiling schema to be used for delivering data through predefined datasets).

  • A tiling schema is needed in the user / client side when providing / delivering coverage data through Predefined Datasets (see this Dedicated page). The tiling schema selected is just suited for a particular purpose.
  • The tiling schema is NOT needed in the user / client side when providing / delivering coverage data through WCS (see this Dedicated page),  where tiling is usually managed by the server in order to store the information and operate in an efficient way.

Therefore, modelling of coverage aggregation is dependent on the tiling schema used in the user / client side (if any is used).

Any change in the tiling schema would require a change of the corresponding coverages (tiles). In other words: the rationale used by the implementer to set what it is considered a "coverage" dataset, could suddenly change, without any apparent justification. 

Additionally, the INSPIRE model proposes a recursive modelling of coverage aggregation (coverages can contain coverages, which in turn can contain further coverages themselves, and so on) which introduces further complexity.

All these facts make this information very costly and tedious to maintain for INSPIRE (e.g. aggregation relationships established, contributing footprints associated to each of them). Additionally, implementing sophisticated features, such as overlapping tiles (allowed now in the INSPIRE TGs), would maximize the issues.

Some data providers have also come up with doubts regarding how to model all this information in the implementation standards (currently GMLCOV).

On the other hand, WCS is tiling schema independent. It does neither require any particular tiling at the client nor the server side (although most WCS servers use an internal tiling for data storage, which is transparent to the user / implementer). Data can be stored in the server as a whole (even though it is not recommendable) or tiled, leaving to the user the freedom to select the tiling schema (between other parameters, such the CRS) in which they want to obtain the data. In other words, it makes neither sense to model the data based on any tiling schema, nor maintain relationship information (e.g. coverage aggregation) based on it.

Finally, the modelling of “coverage aggregation” constitutes an INSPIRE Extension – i.e. it is not implemented by the underlying OGC standards. As a result, it will be ignored by the standard WCS interfaces.

Pending discussion

The proposal to be discussed would be to remove coverage aggregation from the INSPIRE raster data models, leaving to data providers the freedom to select what they consider a coverage without asking to fill-in additional tiling schema dependent information. This would simplify a lot the INSPIRE data models and schemas, making them more independent from the delivery options, while facilitate their maintenance a lot - In line with the INSPIRE REFIT.

It would be necessary to contact ESA to check if all their needs could then be satisfied by using WCS to exploit their raster data. Modelling of “coverage aggregation” seems not necessary if data is exploited through WCS. One way forward might be to have a simple, straightforward “core” function­al­ity which covers most day-to-day use cases plus an “extension” where the high-end, complex funct­ion­ality is gathered.

An online webinar will be scheduled in early 2017 to discuss this specific implementation aspect.

Let's discuss on it!

 

 

  • Jordi ESCRIU

    Dear All,

    coverage service for various (raster data) INSPIRE themes has been setup as a result of the related activities performed in the scope of the INSPIRE Community Forum (Thematic Clusters) during the last years, particularly based on the results from the Workshop “Feel the power of INSPIRE WCS / WCPS in your hands” organised in the INSPIRE Helsinki 2019 Technical Conference:

    http://inspire.rasdaman.org/

    The demo service allows the user to play with different service requests to get INSPIRE coverage data and metadata, analyse and combine different data layers (cross-theme fusion), download and present the final results. 

    It is specially designed to demonstrate use of coverages and WCS/WCPS for several INSPIRE themes: Elevation, Orthoimagery and Land cover - The service is powered by the rasdaman WCS reference implementation and offers sample data from a regional data provider (Institut Cartogràfic i Geològic de Catalunya, ICGC).

    The objectives are to:

    • Experience the potential of these technologies.
    • Fully-understand the main benefits of integrating them within their daily geospatial businesses and analytical tasks.

    ...but mainly:

    • Get feedback from the community on this functional good practice to align INSPIRE coverages to OGC standards, by amending in a simple, OGC compliant way the INSPIRE XSD coverage schemas (as INSPIRE coverage metadata).

    A related discussion thread has been open in the Software & Tools group of the INSPIRE Community Forum for this purpose:

    https://inspire.ec.europa.eu/forum/discussion/view/265095/inspire-coverages-demo-service

    Jordi

Elevation, Ortho & Grids

Elevation, Ortho & Grids

INSPIRE Thematic Cluster Elevation, Orthoimagery, Reference systems, Geographical grids - Join this group to share your knowkledge, learn and collaborate in solving issues related to the Elevation, Orthoimagery, Reference systems and Geographical grids themes