Characterizing Experimental Alluvial Fan Dynamics Using Dense Optical Flow and DEMs of Difference

Abstract. Alluvial fan surfaces reorganize through channel migration, avulsion, and lobe switching, yet existing methods for mapping this activity in laboratory experiments — topographic differencing via DEMs of Difference (DoDs) and dye-based wetted-area mapping — each capture only part of the signal. DoDs record net elevation change but miss activity that cancels within a survey window, while dye-thresholding detects wetness rather than motion and cannot distinguish actively transporting flow from quasi-static ponded water. In this study we introduce a framework that pairs dense optical-flow activity maps, derived from high-frequency overhead imagery, with DoDs computed over identical 30-minute windows. We apply the framework to a constant-forcing alluvial fan experiment that transitions from free progradation to downstream-limited conditions when the fan toe reaches a sediment-export boundary. Optical-flow sensitivity to DoD-detected change remained high throughout (median 0.92), confirming that optical flow reliably captures areas of net morphologic change. Precision, however, declined from a median of 0.61 during progradation to 0.38 during downstream limitation, indicating that a growing share of visible surface motion (approximately 62 % of the optical-flow footprint) no longer produced preserved net topographic change. Occupancy metrics derived from the optical-flow maps show that the fan shifted from spatially distributed, low-persistence activity during progradation (median active-area fraction 0.60; median corridor reuse 0.30) to spatially concentrated, high-persistence activity during downstream limitation (median active-area fraction 0.36; median corridor reuse 0.43). This reorganization occurred without any change in external forcing and therefore reflects an autogenic response to the downstream boundary condition. The paired framework demonstrates that near-zero net elevation change in the downstream-limited phase does not indicate inactivity but rather continued internal reworking in which erosion and deposition increasingly cancel within each analysis window. By separating where activity occurred from where it left a preserved topographic residual, the framework reveals aspects of fan dynamics that neither optical flow nor topographic differencing can resolve alone.