Multivariate classification analysis of cultural landscapes: An example from the Czech Republic

A method of landscape typology and a determination of diagnostic variables for delimiting landscape types was developed for the Czech Republic, thus fulfilling the first step of the European Landscape Convention. Eight datasets describing elevation, aspect, slope, soils, reconstructed natural vegetation, mean annual temperature, mean annual precipitation and land cover were used. The method uses a modified TWINSAPN classification, measuring cluster heterogeneity using JUICE software, and defines a statistically significant diagnostic variable for each landscape type by calculating a measure of fidelity. This hierarchical approach combines statistical methods with GIS techniques, enabling instant visualisation and a further analysis of the results obtained. It is universal when zooming in on the landscape, though higher spatial resolution datasets are needed.

Research highlights▶ Modified TWINSPAN classification yields meaningful hierarchical landscape typology. ▶ Diagnostic variables of landscape types are well defined using fidelity measure. ▶ Interconnection between TWINSPAN and GIS enables instant visualisation of the results obtained.