Visualizing preferential flow paths using dye tracer and species diversity theory methods to explore their correlation to soil properties with random forest algorithm

The preferential flow path development is potentially the result of spatial variations in soil properties with soil depth. However, visualizing the evolution of the preferential flow path with soil depth remains a challenge. This paper presents dye tracer and species diversity theory methods for characterizing preferential flow paths. Field dye tracer experiments were performed at three sites (tree, bush, and grass) in the Yellow River Delta wetland and dye distribution diversity indices (Simpson index (D_s), Shannon-Wiener index (H), Margalef index (D_m), and Pielou index (E)) were applied to verify their availability for preferential flow assessment. The results showed that the uniformity of the shallow-infiltrated dye at the tree site, non-uniformity of the shallow-infiltrated dye at the bush site, and deep dye infiltration at the grass site were the three typical infiltration types. The average proportion of dye-stained areas (PDA) gradually decreased with increasing soil depth. The quantitative effects of soil properties on PDA changes were profound, indicating that soil clay content at 0–10 cm depth, soil sand content at 10–20 cm depth, soil drainage capacity at 20–30 cm depth, and soil bulk density at 30–40 cm depth were the most predictive factors controlling PDA changes. Our results also showed that dye-stained patches with extremely high and high dye concentrations were the most distributed; D_s, H, D_m, and E were the highest at the tree site and E was the diversity index with the greatest importance for PDA change. The findings reveal the soil properties controlling the formation of preferential flow paths, which will improve our understanding of water resource management in the vadose zones of coastal wetlands.