Scale and Temporality

Geographic scale

The intended geographic unit of this classification system is the property parcel.
However, it may be appropriate to map sub-parcel geographic entities for particular classes - particularly if the boundary of natural features (forests, waterways) is pertinent, if the parcel is very large, and where source-data scale permits such definition.

Whether to map sub-parcel areas is therefore left to operator discretion, but the intended and minimum level of attribution is the property parcel, and therefore property-parcel identifiers and geographic boundaries must be present in output land-use data.

Source scale

source_scale

To indicate that one of the endpoints is to be excluded from the set, use a parenthesis and not a square bracket (the latter indicates inclusion).

For example:

• If source data is precise between 0.02 m and 0.75 m, use the notation (0,1), indicating precision at the sub-metre scale, between 0–1 m (but not including the endpoints).
• If vector data is no more precise than 60 m, use [60,) if there is no suitable upper bound; or else determine an appropriate nominal upper bound (e.g. [60,x], where x > 60) since it is unlikely that in reality there is no upper bound.

Rules of thumb for converting nominal scales of input data to this notation:

• Raster data pixel size:
  Given the smaller of the pixel height or width (a) and the larger (b): [a,b].
  If a = b, this notation remains valid ([a,a] = the singleton set {a}; (a,a) would be empty).
  Example: a raster with 30 m pixels → [30,30].

• 1:50 000 scale map data:
  Precision = (map-scale denominator) / 1 000.
  Example: a 1:50 000 scale map has positional precision ≈ 50 m → [50,) (with optional practical or QA-derived upper limit).

• Derived data from mixed sources:
  A map produced from raster data with 10 m pixels, manually digitised at 1:25 000 scale →
  Precision = 10 + (25 000 / 1 000) = 35 m → [35,) (with optional upper limit).

When multiple input datasets are used together to inform a classification decision,
the ranges should be merged - compute the smallest interval that includes all given ranges
(e.g. PostgreSQL range_merge function).
Example: [1,2) and [3,4) → [1,4).

Temporality

Interannual.
A crop planted for a whole year is a commodity, but not necessarily a land use -
for instance, if it forms part of a livestock-farm system.
The primary economic purpose over an interannual period is the determining factor
for assigning a land-use class.