Abstract: Landslide dams exhibit typical spatial variation characteristics, with disordered composition, significant particle sorting, and complex spatial structures, making them prone to dam-break flood disasters under the scouring action of water flow. How to quantitatively investigate the spatial distribution characteristics of parameters such as void ratio and gradation of landslide dams is a fundamental problem that urgently needs to be addressed. In physical model tests, Structure-from-Motion (SfM), a three-dimensional reverse reconstruction technology, is introduced, and the discrete block sampling method is simultaneously integrated. A comprehensive methodology is proposed for landslide dams, encompassing orderly discrete block design, image acquisition, and quantitative analysis of the material spatial variation characteristic field. Subsequently, experimental validation and application are conducted based on this methodology. The test results reveal that both particle size distribution and void ratio exhibit significant variability in three-dimensional space for landslide dam materials. The proposed methodology obtains data on the distribution of void ratio and gradation compositions, endowing landslide materials with spatial distribution information properties. This lays a data foundation and provides a methodological approach for quantifying the accumulation form and spatial variability of wide-graded landslide dams.