Premium
Application of the Hilbert–Huang transform for recognition of active gully erosion sites in the Loess Plateau of China
Author(s) -
Li Jilong,
Na Jiaming,
Yang Xue,
Cao Jianjun,
Dai Wen,
Tang Guoan
Publication year - 2019
Publication title -
transactions in gis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.721
H-Index - 63
eISSN - 1467-9671
pISSN - 1361-1682
DOI - 10.1111/tgis.12512
Subject(s) - loess , landform , loess plateau , geology , erosion , digital elevation model , geomorphology , soil science , remote sensing
The loess shoulder‐lines are known to be the most critical topographic structural lines in the loess landform area of the Loess Plateau, China. Previous studies have shown that gully erosion activity can be reflected in the dynamic changes of the shoulder‐lines. In this study, active gully erosion sites along the shoulder‐lines are investigated. Three typical test areas with different loess landforms are selected, and the loess shoulder‐lines are derived from 1 m resolution digital elevation models. First, height profiles of the loess shoulder‐lines are extracted. Then, from the perspective of frequency domains, active gully erosion sites are identified using the Hilbert–Huang transform (HHT). The experimental results show that the HHT method can effectively mine the intrinsic characteristics of the shoulder‐lines. The detailed information components (non‐components) and trend components of the height profiles can be obtained via ensemble empirical mode decomposition and a significance test. In addition, the HHT method was employed to identify the instantaneous frequency curve catastrophe points of the loess shoulder‐line height profiles, and the physical meaning is clear: a higher frequency denotes greater energy and more severe erosive processes, and the catastrophe point locations represent the active gully erosion sites. Evidence from a recent field survey and a comparison with another method supports the results, indicating that the method proposed in this study could potentially be effective for the exploration and analysis of geomorphologically sensitive locations in loess landform studies.