Since topographic surfaces exhibit multi-scale features, the Hilbert-Huang Transform (HHT) is proposed for exploring the geomorphic features of landslides in this study. HHT is a powerful approach to analyze a non-liner and non-stationary signal of time series. In brief, HHT consists of two parts: The empirical mode decomposition (EMD) and the Hilbert Transform (HT). The original version of EMD was recently refined to the ensemble empirical mode decomposition (EEMD). This paper aims to adopt the method, by using space series (distance) instead of time series, to explore the landslide features and delineate the location and range of landslide. Each data set of topographic parameters, such as elevation, slope and curvature could be decomposed into a number of pseudo-intrinsic mode functions (IMF components). Furthermore, frequency-like domain signals of the data set could be obtained; it is then possible to catch the local characteristics of the system (such as geomorphic features of landslide) by HT. A potential landslide in southern Taiwan was taken as an illustrated example in this study. A 5×5m digital elevation model (DEM) data is available; topographic parameters including altitude, slope, and various curvatures could be obtained through a topographic analysis. The results indicate that (1) landslide features and locations along the analytical profile may be extracted from the Hilbert amplitude and marginal spectrum, and (2) this method provides an alternative way to efficiently search for landslide features.