<p>Conceptual models of the longitudinal evolution of landslide dam geometries during overtopping are essential components of simplified physically based models, which are most widely used for predicting outburst floods. Previous studies have proposed various conceptual models, which are classified as backward erosion models (BEMs), surface erosion models (SEMs), and surface erosion and deposition models (SEDMs) based on the temporal migration of a downstream dam toe location. However, the role of suspended sediments in overtopping erosion processes has not yet been quantified, limiting understanding of the factors and mechanisms controlling the evolution of longitudinal dam geometry. Herein, we report laboratory experiments on the overtopping erosion processes of a landslide dam under 37 different theoretical suspended sediment fraction conditions during overtopping. The theoretical suspended sediment fraction had strong negative correlations with the mean dam height and deposition volume when the main outburst flood occurred. The experimental relationships among the theoretical suspended sediment fraction, bed gradient, and deposition volume were consistent with the equilibrium gradient theory of debris flow. Based on the equilibrium gradient theory, we proposed a method to physically predict longitudinal dam geometry evolution and applied it to the Baige, Tangjiashan, and Ambon dam breaching cases. This method predicted the Baige dam breach as the BEM type and the Tangjiashan and Ambon dam breaches as the SEDM type. We conclude that longitudinal dam geometry evolution is significantly different for each landslide dam breach, emphasizing the importance of identifying an appropriate conceptual model of longitudinal dam geometry evolution for outburst flood prediction.</p>