Characterizing Gravel Bed Structures Using Significance Testing of Second-Order Structure Functions

Gravel riverbeds play an important role in shaping how water flows and how sediment is transported in rivers. These beds often contain complex grain structures that influence bed roughness and flow turbulence. However, identifying and measuring these structures typically relies on expert judgment, making it difficult to compare results across different studies or sites. In this research, we developed a new, objective method to detect and describe grain structures on gravel riverbeds using a statistical approach. To determine whether observed patterns are meaningful or simply random, we generated artificial “random” gravel beds with a machine learning model. These synthetic surfaces serve as a control, allowing us to statistically test the significance of structural features. We applied this method to both laboratory-created and natural riverbeds and found that it successfully identifies key structural features such as grain alignment and clustering—without requiring manual interpretation. This approach offers a new way to better understand how grain structures form and function in riverbeds. In the future, this method could also be used to evaluate other surface features of riverbeds in a similarly objective and statistical manner.