The role of geomorphometric predictors in LUCC modelling: A case study of Slovakia from 1990 to 2018

Abstract

An adequate choice of independent variables – geographical factors – represents a fundamental step in creating a successful model of Land Use/Cover Change (LUCC). Spatial predictors derived from the digital elevation model (DEM) encompass a significant group of variables that are relatively easy to obtain. However, a detailed review of more than 100 scientific articles has revealed that predominantly only some basic morphometric characteristics are used, specifically elevation, slope, aspect, and solar radiation characteristics. In this study, geomorphometric variables and their importance in land cover models are analysed. The modelled response consists of five distinct changes in land cover in the territory of Slovakia from 1990 to 2018: urbanisation, agricultural intensification, agricultural extensification, afforestation, and deforestation; derived from the 3rd level of CORINE Land Cover. Simple logistic regression models were used to determine the predictive power of factors on land cover changes. The hierarchical partitioning method was used to identify independent contributions in multiple regression models. For ruggedness characteristics, the calculation scale was crucial: The quantities computed in the window with radius of 100 m performed best, regardless of their type. Topographic Wetness Index confirmed its significant role in other studies. The duration of solar radiation achieved significantly better results than global radiation. Casorati curvature achieved the best result among curvatures, pointed out that distinguishing between positive and negative curvature is undesirable, and recorded a relatively significant share of independent contributions compared to other factors.