【摘要】 垂直节理发育、直立性强的黄土常形成陡崖地貌。黄土陡崖是侵蚀沟、滑坡等诸多地貌的侧面景观，是系统映射黄土地貌发育演化的重要窗口。受以往数字高程模型（DEM）数据精度的限制，黄土陡崖地貌常呈平面隐性特征，限制了水土流失和地质灾害广泛性的认知精度。本研究选择黄土高原典型研究区，基于山体阴影法和无人机倾斜摄影测量获得的高分辨率DEM数据，构建面向黄土陡崖提取的地貌元素法，将黄土陡崖地貌划分为顶坡、陡崖坡和崖麓缓坡三元素，并与坡度阈值法的提取结果对比评估黄土陡崖地貌的平面隐性特征，研究结果发现：样区内发现陡崖5,093个，陡崖密度为853个/km~2，其中能够被坡度阈值法发现的陡崖仅为2,145个，占42.12%，称为显性陡崖，反之为隐性陡崖；坡度阈值法提取结果与陡崖坡元素重叠面积达95.2%，主要针对陡崖坡面，而地貌元素法包含陡崖地貌三元素，当陡崖坡隐藏时，仍能通过顶坡和崖麓缓坡推测陡崖地貌的存在；显性陡崖与隐性陡崖存在显著差异性，显性陡崖面积和高度平方的相关性优于隐性陡崖，并且显性陡崖普遍面积较大、高度较高，而隐性陡崖大多面积较小、高度较低，面积10 m~2以下、高度15 m以下的陡崖容易隐藏；人工作用陡崖面积和高度小于自然形成的陡崖，人工作用的隐性陡崖占比较多。黄土陡崖地貌元素的引入为黄土地区水土流失和地质灾害隐患点的发现提供更精确的科学依据。更多还原

【Abstract】 The development of vertical joints and strong upright characteristics in loess often leads to the formation of steep cliff landforms. Loess cliffs serve not only as lateral landscapes for various landform features such as gullies and landslides but also as critical windows for systematically mapping the evolutionary processes of loess landforms. However, due to the limitations in the resolution of traditional digital elevation model(DEM) data, loess cliff landforms have often exhibited concealed planar characteristics, thereby restricting the precision in understanding the extent and distribution of geological hazards. This study focuses on a typical region of the Loess Plateau, utilizing hillshade method and high-resolution DEM data derived from the UAV photogrammetry. A landform element based method for loess cliff extraction was developed, dividing the landforms into three elements: the cliff crest, cliff slope, and cliff foot. The results were compared with those extracted using the slope threshold method to evaluate the planar latent characteristics of loess cliff landforms. The findings reveal that: The planar morphology of loess cliffs exhibits significant diversity, with cliffs in different geomorphic settings displaying distinct shapes. Within the study area, 5, 093 cliffs were identified, with a cliff density of 853 per square kilometer. Of these, only 2,145 cliffs(42.12%) were detected using the slope threshold method and are referred to as apparent cliffs, while the remainder are classified as latent cliffs. The slope threshold method primarily targets the cliff slope, with its extraction results overlapping 95.2% of the cliff slope areas. In contrast, the landform element based approach encompasses all three elements of loess cliffs, enabling the inference of cliff presence even when the cliff slope is concealed through analysis of the cliff crest and foot. Furthermore, the slope threshold method often yields fragmented pixel distributions within cliffs, whereas the landform element based method maintains the structural integrity necessary for statistical analysis of cliffs. Apparent and latent cliffs exhibit significant differences. The correlation between area and the square of height is stronger for apparent cliffs compared to latent cliffs. Apparent cliffs generally have larger areas and greater heights, whereas latent cliffs are predominantly smaller in area and lower in height, with cliffs under 10 m2 in area and 15 m in height being more likely to remain hidden. The area and height of artificially created cliffs are smaller than those of naturally formed cliffs. There is a relatively high proportion of hidden cliffs that are artificially created. The introduction of landform elements for loess cliffs provides a more precise scientific basis for identifying geological hazards and potential soil erosion risks in loess regions.更多还原