Abstract
Field measurements of resistance to flow are analyzed for 739 rivers vegetated with grass (281 channels), shrubs (150 channels), and trees (308 channels). The measured distribution of Manning roughness coefficients ranges from 0.015–0.250 for grass, 0.016–0.250 for shrubs, and 0.018–0.310 for trees. Significant trends are obtained between Darcy-Weisbach ($f$) or Manning ($n$) roughness coefficients and flow discharge ($Q$), friction slope ($S_f$), and relative submergence. The regression equations derived—such as $f_{veg}=3.305S_f^{0.508}$ and $n_{veg}=0.144S_f^{0.199}$—provide semi-empirical relationships useful for hydraulic engineering practice.
Introduction: The Challenge of Quantifying River Roughness
In hydraulic engineering, accurately calculating flow velocity requires a precise understanding of "roughness." Vegetation—including grasses, shrubs, and trees—creates significant drag, yet quantifying this effect remains a major hurdle. Traditionally, engineers have relied on subjective "look-up tables" or photographic comparisons, which often fail to account for the dynamic nature of roughness as water levels and flow conditions change.
Methods and Extensive Field Datasets
This study moves beyond subjective estimation by performing a meta-analysis of 739 field measurements from diverse geographical locations, including the USA, New Zealand, and Australia. The data were categorized into three dominant vegetation types:
- Grass (281 channels)
- Shrubs (150 channels)
- Trees (308 channels)
The researchers utilized consolidated datasets from landmark studies such as Hicks and Mason (1998), Coon (1998), and Soong et al. (2009). Statistical analysis was used to visualize distribution, followed by non-linear regression to establish power-law relationships between roughness coefficients and key hydraulic variables like flow discharge ($Q$) and friction slope ($S_f$).
Key Insights: Roughness as a Dynamic Variable
The analysis confirms that roughness is not a static property. Manning’s $n$ typically decreases as flow discharge ($Q$) increases, a critical consideration for flood modeling. While there is overlap between categories, a clear hierarchy exists: average roughness increases from Grass ($n_{avg} \approx 0.046$) to Shrubs ($n_{avg} \approx 0.058$). Trees, while having a similar mean, exhibit much higher maximum values (up to 0.310), reflecting the high drag of dense forest canopies during overbank flows.
Empirical Results and Engineering Formulas
The study successfully derived several robust regression equations for vegetated rivers. These formulas allow for more objective estimation of roughness based on measurable physical parameters:
- Based on Discharge ($Q$):
- $f_{veg} = 0.436Q^{-0.363}$
- $n_{veg} = 0.061Q^{-0.124}$
- Based on Friction Slope ($S_f$):
- $f_{veg} = 3.305S_f^{0.508}$
- $n_{veg} = 0.144S_f^{0.199}$
Furthermore, the study suggests that the traditional Manning-Strickler $1/6$ power law for submergence may need adjustment for vegetated flows, with a $1/8.3$ exponent fitting the field data more accurately.
Conclusion
By providing data-driven, semi-empirical formulas, this research offers hydraulic engineers a more reliable toolkit for predicting flow resistance in natural, vegetated channels. While limitations exist—such as the complexity of mixed vegetation types—these relationships represent a significant step toward objective and accurate river modeling.
Summary updated