Delineating individual alluvial fans and morphological analysis based on digital elevation models

Abstract

Alluvial fans are common landforms deposited by streams and/or debris flows. Multiple fans can overlap and connect with each other, developing a bajada. Currently, studies about the evolutionary process of alluvial landscapes and environmental changes in arid areas urgently need clear, detailed and object-based evaluation within bajadas. However, existing methods still struggle to divide these alluvial slopes into individual fans. Accordingly, we proposed a novel method for delineating individual alluvial fans based on free digital elevation models. This method incorporated multi-directional hillshades to accurately extract alluvial slopes, and two flow direction algorithms to support the delineation of individual fans. The method was tested in Death Valley, where a total of 89 individual fans were delineated for morphological analysis. Our result demonstrated high accuracy validated by a Landset-9 L2 satellite image, with an intersection over union (IoU) of 95.51 % for alluvial slopes and a mean IoU of 86.19 % for individual fans, and it also aligned well with the geological map. Additionally, based on the delineated individual fans, we computed morphological features, distinguished the surface differences of both sides and analyzed the feature correlations. Overall, this study accomplished the delineation of individual fans and provided a viable solution for the automated and large-scale mapping of arid alluvial landscapes. The delineated results can act as the foundation for research about the evolutionary process and genetic mechanisms of alluvial fans, and provide opportunities for understanding the tectonic activities, sediment deposition and erosion process and environmental changes in arid regions.