Quantitative evaluation of the effects of artificial grass and stem covers on overland flow hydrodynamics

Vegetation coverage on hillslopes substantially alters the hydrological and hydraulic erosion processes of overland flow; however, the mechanisms by which grass-shrub community coverage influences overland flow hydrodynamics remain insufficiently understood. To clarify the effects of varying grass-shrub coverage on flow dynamics, a systematic flume experiment with a non-erodible bed was conducted. The experiment design comprised 25 combinations of artificial grass (Cg) and stem cover (Cs) (five levels each), five unit discharges (q) ranging from 0.278 to 2.222 L·m−1·s−1, and four slope gradients between 2° and 12°. The results revealed that: 1) the observed flow was predominantly laminar and transitional, with the onset of transitional flow primarly governed by discharge, occurring at a critical threshold of 0.556 L·m−1·s−1. Higher vegetation coverage facilitated the transition of overland flow from supercritical to subcritical regimes, whereas steeper slopes increased the vegetation coverage threshold required for this transition. (2) Increasing vegetation coverage altered the relationship between Manning’s n and the discharge q from negative to positive, while steeper slopes reversed this trend—transforming positive correlations into negative ones and diminishing the proportional increase of Manning’s n with coverage. Vegetation also attenuated the rate at which mean velocity increased with q: as q rose from 0.278 to 2.222 L·s−1·m−1, velocity increased by 118 % at Cg = 0 but only 13 % at Cg = 65.97 %. This underscores vegetation’s critical role in modulating flow resistance and velocity. (3) A predictive model for mean flow velocity under artificial grass-shrub vegetation was developed and rigorously evaluated through error and sensitivity analyses. The model mechanistically incorporates both particle resistance (arising from substrate roughness) and form drag (induced by vegetation morphology). It demostrated high predictive accuracy when validated against the experimental dataset, achieving an adjusted R2 of 0.877 and a Nash–Sutcliffe efficiency of 0.875. For the study dataset, the mean relative error was − 0.029 (standard deviation = 0.164). These findings substantially advance the mechanistic understanding of overland flow hydrodynamics in the presence of grass and stem cover.