Land cover/ land use changes
Land cover / land use changes product represents confidence of changes as detected from Landsat satellite imagery between consecutive years or selected time period. This layer expresses the probability with which a change occurred in the monitored period. The probability is presented in percentage and takes values ranging from 0 to 100. Land cover / land use change in the form of polygons can be obtained by setting an appropriate threshold of probability. The confidence layer can also be used as a quality layer indicating the degree of uncertainty of the changes.
Three classes of land cover/land use have been defined depending on the information content of the input imagery (especially spatial resolution) and the capabilities of the automatic detection: artificial surfaces, non-artificial surfaces and other surfaces. They do not distinguish various types of the built-up area or vegetated area. Thus single land cover flow was defined to provide information about urban expansion. It includes all changes from vegetation classes to urban classes. An overview of the classes based on aggregated PUMA nomenclature brings the following Table 1.
|Class Name||Class CODE||Included classes by PUMA||Class Description|
111 Continuous Urban Fabric (S.L. > 80%)
113 Discontinuous Urban Fabric (S.L. < 80%)
121 Industrial units
|Includes all types of urban fabric, industrial, commercial and transport units.|
141 Green urban areas
200 Agricultural Areas, semi-natural areas and wetlands
|Includes all types of vegetation (Natural and Semi-natural) and water bodies and river bed.|
133 Construction sites
233 Bare Land
Table 1: UD nomenclature
The UD processor provides two land cover / land use change products with different level of post-classification adjustment. The first product includes all changes as detected from Landsat satellite imagery. No post-classification editing is applied. These layers are suitable for statistical evaluation of urban expansion or for manual post-editing. The second product utilizes the auto-context principles. These layers present more generalized change objects and therefore they are appropriate for visual interpretation but in some cases also for manual post-editing (depending on its purpose).
Land Accounting System (LAS)
Land accounting system represents a standard way for presentation of the change detection. The change segments representing urban expansions are generalized to the 1.5 km regular grid. This method allows detection of urban sprawl intensity, comparison of products with different spatial resolution or generalized assessment of changes in a larger region.
The metadata products are enclosed to each delivered area of interest in form of raster file. Two types of metadata products are produced. First metadata file provides information about number of available Landsat satellite images and valid values in each pixel and for each year in the 2000-2014. Second metadata file brings information about which pixel has been used to change detection, i.e. Landsat acquisition date. The file contains values from 1 to 53, which represents week of the year.
The metadata files are used to identify pixels/areas with the absence of information necessary for correct change detection. These areas have lower likelihood of UD processor result and should be used together with metadata layers. They include a small number of valid values in the monitored time period or Landsat imagery is acquired during the no growing season.
The UD processor result is unusable when the number of acquired scenes is zero (no information about the land cover status in this year). Utilization of acquisition date is dependent on locality or rather occurrence of growing season. The layer serves to identify the temporary bare soil in farmland.