Landslides are a frequently occurring threat to human settlements. Along with global climate change, the occurrence of landslides is the forecast to be even more frequent than before. Among numerous factors, topography has been identified as a correlated subject and from which hillslope landslide-prone areas could be analyzed. Geometric signatures, including statistical descriptors, topographic grains, etc., provide an analytical way to quantify terrain. Various published literature, fast Fourier transform, fractals, wavelets, and other mathematical tools were applied for this parameterization. This study adopts the Hilbert-Huang transform (HHT) method to identify the geomorphological features of a landslide from topographic profiles. The sites of the study are four “large-scale potential landslide areas” registered in the government database located in Meinong, Shanlin, and Jiasian in southern Taiwan. The topographic mapping was conducted with an airborne light detection and ranging instrument. The resolution of the digital elevation model is 1 m. Each topographic profile was decomposed into a number of intrinsic mode function (IMF) components. Terrain characterization was then performed with the spectrum resulting from IMF decomposition. This research found that the features of landslides, including main scarp-head, minor scarp, gully, and flank, have strong correspondence to the features in the IMF spectrum, mainly from the first and the second IMF components. The geometric signatures derived with HHT could contribute to the delineation of the landslide area in addition to other signatures in the terrain analysis process.