Geospatial features in LPIS

From Wikicap - European Commission

Go up to the main MLL page

To transfer the observed physical world into a digitized and computerized system as LPIS, we use geospatial features classes as a digital representation in space and time - a class groups features with a common definition, like “trees”, or “buildings”. Each LPIS entry (“instance”) is a one-to-one representation of the real object on the ground without making any assumptions in shape and size for the business needs. To associate information to this physical features geospatial feature classes have their properties in form of a geometric representation and alphanumerical information about these objects.

Feature properties

Geometry (spatial attributes)

Geometries deal with co-ordinates to store location and shape information. In LPIS, a feature type can have one or more basic geometry representation type in a form of Point, Curve or Surface. Each geometric object is associated with a Spatial Reference System, which describes the coordinate space in which the geometric object is defined. Since the objects are being represented in a two dimensional perspective (x, y), a collection of feature classes are corresponding to Points, Lines and Polygons.

Basic geometry types in LPIS.jpg
Figure 1: Basic geometry types in LPIS

Geometries are selected and acquired according to the common geometric appearance of an object and the need for information about those. E.g:

  • the reference parcel is represented as polygon for display on the computer screen and in the pre-established form/map,
  • an agricultural parcel can be represented by a polygon where its boundary is an individual spatial feature, but it can “inherit” the geometry of a reference parcel when occupies that RP completely.

In GIS modelling, a set of specific topology rules between features types and their geometries can be established to model more complex conditions: Two adjacent polygons share a boundary. The “sharing” of a single line segment prevents the occurrence of gaps and overlaps between the two adjacent real world features. Understanding the rules and restriction that apply to the geometric representation of a spatial feature is essential for the correct application of spatial operations such as the spatial crosscheck.

Alphanumeric attributes

Attributes store information of properties of the feature. A single feature type can set out many attributes to record the id number, area, length, farm id, name, non-coordinate location, etc... In a GIS environment, some of the attribute values associated to a feature can come from a pre-defined code list, an enumeration or can be user defined free text. Values can be entered into the system or derived from the geometry residing inside. Some attribute values (e.g. area values) can be essential for any alphanumeric crosscheck.

Pre-defined code1.jpg Predefined code2.jpg
Figure 2: Pre-defined code lists for EFA type values and LF type value

Exampleofattribute.jpg
Figure 3: Example of attribute information associated to a reference parcel

Spatial relationships – geometry rules

Topology

To assure the concept of “unique area” definition in LPIS, topological relationships must be respected during the digitizing process and data processing. The common topological rules applied for vector data - line and polygon feature types - are: Line errors - dangles, switchbacks and knots, are not acceptable in case one spatial object or its boundary or border is represented by polyline. The resulting geometric length of the line cannot be used to calculate length, perimeter or area of that feature. Polygon errors – slivers and gaps, create either less respectively more (double counted) areas in the system than there is present in the real world. Within a single data set of “unique area” in LPIS, topology must be free of these common artefacts. Artefact means that the set of coordinates do not represent (a part of) the feature in the real world.

Topological.jpg
Figure 4: Topological errors

Examplesofgap2.jpg
Figure 5: Example of gaps and slivers within the ”RP layer” in the system

To some extent, spatial overlapping between geospatial features from different data sets will be possible and it largely depends on the parcel and landscape feature definition from each individual system. The figure below illustrates a controlled overlap between data sets. The green polygon (1st data set) holds an area, and on the top of it is another yellow polygon (2nd data set) holding its area. The final accounted area is not the sum of the two polygon areas, but the area from the polygon with the biggest extent.

Controlled overlap.jpg
Figure 6: Controlled overlap of two features

Controlled overlap is possible for the eligible features, whose area doesn’t have to be excluded from the maximum eligible area of the reference parcel, but the presence of the feature should be recorded and graphically represented in LPIS. Note that the “unique area” concept implies a 2-dimensional space: iIn IACS, it is relevant only for features’ spatial properties on the ground level. Furthermore:

  • ground level (from the DTM) defines the spatial canvas for orthoimagery production,
  • eligible hectares such as arable land, permanent grassland are defined by the ground level vegetation.

The unique area concept and subsequent topologic rules are irrelevant between spatial attributes that relate to different levels of representation of one and the same natural phenomenon: For instance, a tree canopy has its largest expansion several meters above ground level, so

  • it can easily cover and overlap with arable land below (and possibly obstruct the latter’s visibility),
  • its position on orthoimagery will become over-susceptible to perspective displacement.

First.jpg Second.jpg Third.jpg
Figure 7: WV2 0.5 m resolution image taken at two different moments with a different off nadir angle.

Figure 7 shows the low positional accuracy of the digitized polygons of the tree canopies visible when viewing them against a later image. The positional accuracy of the digitized lines of the tree trunks is suitable for both images. To deal with the 3 dimensional reality of features and respecting the unique area requirement, some feature types will require multiple geometries, including one for the representation at ground level.

Inheritance principles

Based on the MS’s choice, some features may have common geometry and some of the features might share a common boundaries, Depending on these choices, ensure that at the level of the coordinates that build up these geometries, lines would have to coincide with the polygon edges, points would have to fall along the line feature, or would have to be within the polygon areas, etc. If a particular feature is classified into two spatial feature types (e.g. a hedge is BPS eligible area and EFA element or a crop acts as agricultural parcel and reference parcel), then the corresponding geometry must be common for all feature types. I.e. a common polygon is shared by all instances of that feature.

Consistency

If a spatial or alphanumeric rule applies to one feature, it should apply to all instances of that feature types. Rules cannot be applied on a feature by feature basis but always on the class.

Go up to the main MLL page