Original paper
Automated landform classification and mapping using a combined textural-morphometric approach: the Congo basin and surroundings
Viennois, Gaëlle; Bétard, François; Freycon, Vincent; Barbier, Nicolas; Couteron, Pierre
Journal of Geomorphology Volume 1 Issue 1 (2022), p. 79 - 102
published: Dec 15, 2022
published online: Aug 19, 2022
manuscript accepted: Jun 2, 2022
final revised version received: May 25, 2022
manuscript revision requested: Apr 14, 2022
manuscript received: Dec 23, 2021
DOI: 10.1127/jgeomorphology/2022/0752
Open Access (paper may be downloaded free of charge)
Abstract
An automatic method of landform mapping applicable to large continental areas is presented, based on 30-meter SRTM (Shuttle Radar Topography Mission) data and combining texture analysis using Fourier 2D periodograms (FOTO method) with a set of morphometric variables. This integrated strategy was applied to the whole Congo Basin and adjacent regions in Central Africa, where landscapes and landforms mapping remains heterogeneous and partial with existing expert maps differing in aims and scales. Through the FOTO method, a principal component analysis (PCA) on obtained Fourier r-spectra yielded six textural features, which were further combined with seven morphometric criteria into a global PCA. A k-means classification from these output results provided an automatic mapping of 12 landform classes (at a final resolution of 900 m) which were successfully interpreted in terms of geomorphological meaning together with some hydrological and soil attributes. Finally, comparison of our landform map with existing, independent geomorphological sheets revealed a good spatial congruence. Overall, our method proved effective to depict landform assemblages at regional or continental scales based on complementary textural information and morphometric parameters. As such, it could serve as a sound basis for further predicting and mapping soil units at the landscape scale, given the close soil-landform imbrications and interactions at the catena level. It could serve as well as a predictive variable for biodiversity measures and biomass estimates, especially in the humid tropics where environmental data are lacking whilst ecological modelling is urgently needed to support land planning and forest management.
Keywords
SRTM • geomorphological mapping • texture • morphometry • Congo basin