Layers in LPIS

From Wikicap - European Commission

Go up to the main MLL page

Layers in general

The modern GIS stores the geospatial and associated alphanumeric information in specific tabular form, structured and organized according to the rules of the relation databases management system (RDBMS) applied. However, the representation (in the sense of portrayal and display) of the geospatial information is done through the concept of “layers”. Layer can be considered the visual representation of a geographic dataset in any digital map environment. Conceptually, a layer is a stratum of the geographic reality in a particular area. Every single stratum holds geospatial features sharing common characteristics and properties associated to one or more given feature classes (or types). The layer is more or less equivalent to a legend item on a paper map. On a road map, for example, roads, national parks, political boundaries, and rivers might be considered different layers. Grouping geospatial features with common characteristics into layers allows the user to manipulate each set of features individually. The specific portrayal rules can be defined and applied for each layer making the different features classes well distinguishable. Layers are of key importance for any spatial analysis and data querying, as they provide the interface needed by the user to be able to understand the spatial relationship between the geospatial features and classes represented. Although the notion of the layer in GIS is quite close to the old analogue map production concept, based of overlaying of different “hardcopy layers” of thematic data, digital layers in GIS do not have the same physical nature, as they are not the „containers” where the spatial data is stored. Layers can be simply regarded as sets of queries and the rules for representation of particular spatial features that reside (and remain residing) in the geodatabase. Whereas ISO 19128:2005 defines layer as “basic unit of geographic information that can be requested as a map from a server”, different GIS software use the term for slightly different concepts. Still, but the term “layer” is mostly used in one of two GIS functionalities:

  • for controlling the display and portrayal of sets of spatial features. This use is quite similar to the concept of colour film in the cartographic printing process,
  • for organising selected spatial feature in sets in preparation of and storing the results of spatial queries operators such as masking, vicinity analysis, intersections, etc.

There are two technical implementations for organising data in layers; one is raster-based, the alternative is vector-based. Orthoimagery is the raster layer produced from raw sensor images by removing perspective and topographic distortion. It has is no practical vector equivalent.

Raster layer

Raster layer is usually a layer that references a raster as its data source and a raster renderer that defines how the raster data should be rendered, as well as any additional display properties. Any modifications made in the raster layer do not affect the raster data, as they only control the way how it is rendered at the time of viewing. Certainly, any modification of the original raster dataset, would be reflected by the raster layer. The most important viewing properties having direct impact on the photointerpretation for the LPIS purposes are as follows:

  • band combination: the image bands selected for visualization and the way they are mapped into the RGB space,
  • image data stretch: the function used to position (stretch) the DN values of the image histogram in the given display range,
  • resampling: the methods used to resample the image data for display purposes.

Orthoimage.jpg
Figure 8: An orthoimage rendered with different mapping of the image bands to the RGB (Red, Green, Blue) space

In order to obtain correct positioning of the image data in the selected coordinate reference system (CRS) of the particular GIS instance (project), it is essential to specify correctly the CRS in which the (ortho) image data is delivered. Although, it seems straightforward, experience shows the different GIS environments uses slightly different descriptions and libraries for the same CRS. They can also differ from the description of the CRS in which the orthoimage was created. This require a careful check of the CRS parameters, such as datum, geographic projections, position of pixel coordinate, any specific shifts, in order to guarantee correct transformation and display of the rendered image data. Another important aspect is the presentation of the original spatial resolution of the acquired images. Not only the resulted orthoimage should have pixel size correspondent to the ground sampling distance (GSD), but the resampling approach for display of screen in the GIS environment, should be selected in a way to preserve the information content. Same is valid for the representation of the DN values of the image pixel on screen, as the correct colour palettes compatible with the radiometric resolution of the image rendered, has to be used.

Vector layer

Vector layer is a layer that references a set of feature data. Such set may include points, lines, and polygons. Similarly to the raster layer, vector layer controls how and which of the feature data is displayed and annotated, without an impact on the actual data content stored in the RDBMS. The most common displaying properties (portrayal rules) are:

  • symbology: the map symbols used and rules to rendering your data (colours, line types, fill options),
  • labels: the label expressions, label classes, and labelling options for label placement and symbology,
  • display order: the way feature data is displayed, while moving in the view.

As a vehicle to hold the spatial attributes of database features, layers have to comply with the following requirements:

  • the layer (data sets behind it) must be defined in a well-documented coordinate reference system, appropriate for the purpose of the layer. In particular the Regulation imposes the use of a nationally recognised CRS. If several regional CRS are used within a MS, features of different regions must be transformed (re-projected) into the national CRS. Note: since all CRS involve some kind of distortion an area attribute of the spatial feature calculated from the original geometry, will no longer perfectly match the area of the converted geometry. This is not necessarily an issue.
  • positional accuracy of each spatial co-ordinate must be of roughly the same level. Or specified from a cartographic viewpoint, the scale range must be applicable for all features in the layer.
  • a different lineage (historic cartographic material, different aerial flights) may cause discontinuities (small coordinate shifts) between different features on either side of the line between lineage. These have to be addressed once and for all for the set of effected features

Topological and hierarchical rules must be respected. In particular,

  • layers that are used to represent unique areas within a given theme (e.g. eligible hectares, EFA elements) cannot support overlaps (no “double slivers”),
    • adjacent features must have a common boundary in the layer (no “empty slivers”),
    • any hierarchical rule on the geometry of the features must be reflected in the layer. Merely allowing the population process sequence to determine the final result (e.g. “last one on top” is not a good practice),
    • Tessellation (complete coverage of the layer) is not required; controlled gaps are allowed.

It goes without saying that many of the above rules and conditions should have been considered when the spatial database was made and kept up. In that case, the layer creation process becomes a pretty straightforward extraction.

Classesofcore1.jpg Classesofcore2.jpg Classesofcore3.jpg
Figure 9. Classes of “core” LPIS layers – RP, EFA, LF

Single feature layers in LPIS

The Reference parcel layer is the oldest and most familiar layer in the LPIS. But in fact it is a single feature layer only in those block systems that combine identification of the land (RPid) and quantification of eligible hectares (reference area). Any eligibility or ineligibility mask represents a layer of consistent land cover observations. For eligibility masks, it groups the various land cover classes of the eligibility profile, the ineligibility mask represents the contrary. The geospatial application produces the feature that populates the agricultural parcel layer. It is an annual layer and although many farmers co-author that layer. Series of automatic crosschecks ensures it meets the above requirements.

Interacting between different layers

Just like certain conditions apply when a single spatial layer is created from a set of spatial features, a number of requirements need to be respected when two layers are combined for spatial analysis or data processing for the generation of new set of features. To produce useful information, all layers:

  • must be defined in a common, appropriate CRS,
  • hold features with a compatible positional accuracy or with an compatible range of cartographic scale,
  • represent the earth’s surface within a compatible time frame.

An illustration of how critical these conditions are is given by the ortho-rectification process. Indeed the GNSS-surveyed GCPs, the sensor position (from flight navigation system) and the digital elevation model represent three layers from different sources that together produce an orthoimage layer (in the national CRS). The DG JRC has plenty of examples of cases where a small defect in any of the sources causes defective imagery (page 15. ftp://mars.jrc.ec.europa.eu/LPIS/Training_2011/PDF/tallinn2011_ETS_12_final.pdf). Very often, the interaction process directly propagates the spatial errors of each contributing layer in the result and the final quality can at best equal the quality of the poorest layer. Often the result can be even worse. It is therefore essential that the quality of each layer is documented and controlled and the appropriateness of the layer combination is assessed and appropriate action is taken.

Multiple feature layers in LPIS

Non-Block reference parcel designs use a combination of a cadastral parcel or topographic block with an (in)eligibility mask. This can be a viable option, but a quality problem in either of the combining layer, most likely present in the third party source layer, can make the result substandard. In particular, cadastral parcels can run the risk of being obsolete or have insufficient positional accuracy to represent the actual physical state of the land concerned. In that case, these data should not be considered. On the other hand, if a cadastral index layer does accurately match the current and true situation, it can freely be combined to act as reference parcel even to constitute a hybrid block system because the resulting information has not been subject to degradation. EFA layer is explicitly mentioned in the Regulation but, pending on the implementation choices made by the Member States, it can hold the geometries of a broad range of features, such as landscape features, strips and practices. Compiling a technical EFA-layer is potentially not a simple or straightforward process, because of the heterogeneity of the spatial features that could constitute an EFA-element.

Area calculation

Area values stored in the different layers of LPIS refer to the ortho-projected area; i.e. the area of the two-dimensional graphic representation of a physical object on Earth in which all the projection lines were orthogonal to the projection plane, defined by the CRS. For objects with well-distinct third dimension (trees, buildings) the considered area is the ortho-projected area of their physical footprint. In the case of tree, it will be the trunk with the associated area; in the case of building, it will be the footprint of its base, and associated area.

Go up to the main MLL page