AUTOMATING THE MAPPING AND MEASUREMENT OF GEOMORPHIC RESPONSE TO REGULATED RIVER FLOWS: A CASE STUDY IN GRAND CANYON, ARIZONA

Quantifying the effects of flow alteration on river channel and valley geomorphology is often difficult given the highly connected nature of fluvial and non-fluvial geomorphic processes. For geomorphologists, geomorphic change detection using high-resolution digital elevation models (DEMs) provides a tool for quantifying the relative contributions of sediment transport processes to local or valley-scale geomorphic change. Professional judgment is often used to discern the mechanism(s) of change in DEMs-of-Difference (DoDs), though this method is inherently subjective and may lead to significant differences in interpretation between researchers. Here we present an automated, two-part method for attributing geomorphic processes to apparent surface changes in DoDs that is both reproducible and computationally simple. Part 1 employs a landscape-based approach (LBA), wherein we use topographic characteristics from repeat surveys to predict primary and secondary geomorphic transport mechanisms for an area of interest. Part 2 uses spatial dimensions, orientation, and location of geomorphic change in a DoD-based approach (DBA) to predict a single geomorphic mechanism. The results of both methods are then evaluated for commonality to assess our confidence in each prediction. We evaluated this approach at 114 discrete sample points within seven field sites undergoing a combination of fluvial, aeolian, and hillslope geomorphic processes along the Colorado River within Grand Canyon, Arizona. We found that where the results of the LBA and DBA agreed, the common geomorphic mechanism was correctly predicted in > 95 percent of sample points. Overall performance for geomorphic attribution was strong, though additional refinement of the process will likely improve accuracy for both the LBA and DBA independently as well as increase agreement between these methods. We anticipate that this approach will provide a technique for geomorphic process attribution from repeat topographic data that is both rapid and reproducible at multiple spatial scales (e.g., 10¹ – 10⁵ m²) within the project area and is adaptable to other geomorphic contexts.