Terracettes are repeating step-like microtopographic features roughly following the contours of hillslopes that are often associated with livestock tracks. These common features in many semi-arid rangelands have been shown to alter soil moisture, slope stability, sediment transport, infiltration rates and coincident vegetation patterns. The spatial extent and distribution of terracettes is currently unknown and therefore their landscape-scale hydrological influence is absent in modelling and land management decision making. When viewed in uncalibrated very-high-resolution satellite imagery, terracettes appear as repetitious parallel lines within a specific frequency range. Here we used the two-dimensional discrete Fourier transform to identify terracettes at three test sites in the Inland Pacific Northwest, USA. We created an automated rule-based classification of terracetted sites based on spatial frequency, orientation, slope angle and land-use class. Results show a detection accuracy of 77% based on an optimized spatial frequencies search window between 0.3 and 0.7m À1. Terracette orientation did not contribute significantly to detection accuracy because orientations varied ±50°from digital elevation model-derived aspects. We found terracettes occurred predominantly on north-facing slopes at our test sites, although this estimate may be exaggerated by the timing of image capture. We feel that the method developed in this paper provides a way forward to map terracettes at large scales and enable new insights into the functions of terracettes in the landscape.