› Stream Restoration & Bankfull Discharge Simulator Back
Stream Restoration

Stream Restoration & Bankfull Discharge Simulator

A planning tool for restoring engineered, straightened streams back to their natural form. Adjust drainage area, channel width, depth, slope, sediment size and Manning's n to watch the bankfull discharge, velocity, width-to-depth ratio, Rosgen stream type and particle mobility update in real time.

Parameters
Drainage area A
km²
Bankfull width W
m
Bankfull depth D
m
Bed slope S
m/m
Median particle d50
mm
Representative grain size of bed material
Manning's n
Typically 0.030-0.060 for natural streams
Rosgen stream type
Target form for the restoration
Results
—
Cross-section A_xs (m²)
—
Mean velocity V (m/s)
—
Bankfull discharge Q_bf (m³/s)
—
Regional estimate Q_reg (m³/s)
—
Width/Depth ratio W/D
—
Recommended Rosgen type
—
Stream cross-section & floodplain — Rosgen view

Central trough = channel; both sides = banks and floodplain. Blue intensity scales with velocity; yellow dots mark d50; the top label shows the recommended Rosgen type.

Regional curve — Q_bf vs drainage area
Typical W/D ratio by Rosgen type
Theory & Key Formulas

$$Q_{bf} = \frac{1}{n}\,A_{xs}\,R^{2/3}\,S^{1/2},\qquad \tau_b = \rho\,g\,R\,S$$

Manning's equation for bankfull discharge Q_bf and boundary shear stress τ_b. n = Manning's roughness, A_xs = flow area, R = hydraulic radius (= A_xs / P, P = W + 2D = wetted perimeter), S = bed slope, ρ = 1000 kg/m³, g = 9.81 m/s².

$$Q_{reg} = 2.5\,A^{0.85},\qquad \tau_c = 0.045\,(\rho_s-\rho)\,g\,d_{50}$$

Regional curve (Q_reg in m³/s, A in km²) and Shields critical shear stress τ_c. The particle d50 starts moving when τ_b > τ_c. The tool uses quartz submerged density ρs − ρ = 1650 kg/m³.

Stream Restoration & Bankfull Discharge Design — Rosgen Classification

🙋
What exactly does "stream restoration" restore? Is it just ripping out concrete bank linings?
🎓
That's part of it, but the idea is broader. Twentieth-century flood control left a lot of streams as straightened concrete trapezoids with no meanders and no floodplain. Stream restoration is the environmental-engineering effort to put back both the natural form (meanders, riffle-pool sequences, floodplains) and the natural functions (habitat, groundwater recharge, flood attenuation). The US drives this through NRCS and EPA programmes; in Europe it sits under the Water Framework Directive.
🙋
Got it. With the default settings I see "Q_bf = 35.9 m³/s" and "Q_reg = 69.5 m³/s". What is the difference?
🎓
Good catch. Q_bf is the discharge that Manning's equation gives for the cross-section you set: 12 m wide, 1.5 m deep, 0.5 % slope, n = 0.04. Q_reg is what regional regression curves say a typical 50 km² catchment in that region should carry at bankfull — about 69.5 m³/s. When the two diverge it's a red flag that the cross-section is unnaturally sized. Here Q_bf is about half Q_reg, so the channel is too narrow or too shallow for that drainage area.
🙋
The recommended Rosgen type shows "E" but I picked "C". Why the mismatch?
🎓
Rosgen sorts streams by width-to-depth ratio and slope. With W/D = 8 (12÷1.5) and slope 0.5 %, the rule "W/D < 12 and slope < 2 %" lands you in E (a deep, narrow stream). You picked C, which needs W/D ≥ 12 (a wide, meandering channel with broad floodplain). This is a classic restoration mistake — the geometry doesn't support the target form. To actually build a C-type channel you need to widen the bankfull width until W/D crosses 12.
🙋
How do we actually use the Shields particle-mobility check in restoration design?
🎓
The goal of natural-channel design is for the stream to "groom itself" — moving its bed material a little at bankfull and reshaping pools and bars. So we size d50 so the Shields number is just above the 0.045 threshold at bankfull discharge. Too coarse and the bed armours into a static pavement that kills habitat; too fine and a single flood sweeps it out and undermines the structures. The "is mobile" flag in this tool is doing exactly that check.
🙋
One last thing — why is bankfull tied to a 1.5-2 year return period? Wouldn't bigger floods do more to shape the channel?
🎓
It feels counterintuitive, but the dominant shaping work is done by moderate floods that happen often. A 100-year flood is powerful but rare; everyday low flows are frequent but weak. The product (work × frequency) peaks at around the 1.5-2 year return event — Wolman and Miller (1960) showed this and it's why we call bankfull the "dominant" or "formative" discharge. Banks, particle sizing and floodplain width are all keyed to it.

Frequently Asked Questions

Bankfull discharge is the flow that just fills the channel to the top of its banks, i.e. up to the floodplain. It corresponds to a return period of roughly 1.5-2 years and is treated as the 'dominant' or 'formative' discharge that shapes the channel. This tool computes it from Manning's equation Q = (1/n)·A·R^(2/3)·S^(1/2).
Rosgen's classification (Rosgen 1994/1996) splits streams into types A-G based on slope, width-to-depth ratio, sinuosity and bed material. Common targets are B (semi-confined, moderate slope), C (meandering with wide floodplain, W/D ≥ 12), D (braided, sand-gravel) and E (deep narrow streams with W/D < 12). Restoration designers compare the present (engineered) reach with a reference natural reach and pick a target type.
A regional curve is a statistical regression of bankfull discharge against drainage area for gauged streams: Q_bf = a·A^b (a regional, b ≈ 0.7-0.9). This tool uses Q_bf = 2.5·A^0.85 (A in km², Q in m³/s) as a generic value. Real coefficients vary by an order of magnitude between dry and humid regions, so always calibrate against locally gauged streams.
The Shields parameter (dimensionless shear stress) τ* = τ/((ρs−ρ)·g·d) compares bed shear stress to the critical value (~0.045 for sand-gravel). This tool compares τ = ρgRS with τc = 0.045·(ρs−ρ)·g·d50. Above τc the bed is 'mobile' (active gravel-bed); below it the bed is 'static' (armoured). It guides sizing of bed material in restoration.

Real-world applications

Urban stream daylighting and naturalisation: Projects such as Tokyo's Nogawa River, Los Angeles' Compton Creek and Seoul's Cheonggyecheon convert concrete-lined channels back to natural form. The workflow starts with a regional curve to estimate target bankfull discharge, picks a Rosgen type that fits the slope and valley confinement, and uses Manning's n in the 0.035-0.055 range (vegetation plus gravel bed) to check that the computed discharge matches the regional estimate.

NRCS / EPA Natural Channel Design (NCD): Since the 1990s the US Department of Agriculture's NRCS and the EPA have standardised Rosgen-based Natural Channel Design and applied it to hundreds of restoration projects in the eastern United States, especially North Carolina. The reference-reach approach measures W/D, meander wavelength and grain-size distribution on a healthy nearby stream, then transfers those metrics to the impaired reach.

Mountain stream and check-dam design: For steep streams the same calculations are used to pick bed material that won't lock up. If τ_c is more than twice τ_b the bed armours and biota suffer; if τ_c < τ_b a single flood washes out the rip-rap. Sizing d50 so τ_b ≈ τ_c is the field rule of thumb.

Fish-passage and habitat retrofit: Salmonid streams need bankfull-flow riffles where depth stays under ~0.3 m. A Rosgen C or E geometry produces the riffle-pool sequence those species require, and this tool is used to iterate W/D and slope until the simulated riffle velocity and depth fall in the target window.

Common pitfalls and pro tips

The biggest mistake is confusing bankfull discharge with the design flood. Bankfull is a 1.5-2 year "formative" flow, not the 50-100 year flood used in levee design. Rosgen typing and particle sizing should always use bankfull, while levee crest elevations and bridge openings use the design flood. Mixing them — for instance trying to pass a 100-year flood through a bankfull cross-section — leads to catastrophic overtopping.

Second, do not apply Rosgen mechanically. The classification was empirically built from North American temperate streams. In steep, flashy catchments (e.g. Japan's mountain rivers) the D vs F boundaries blur and judgement diverges. The recommendation in this tool is a guideline only — pair it with a reference-reach field survey and locally calibrated thresholds before committing to a design.

Third, do not trust a single regional curve everywhere. The Q_bf = 2.5·A^0.85 equation is a generic fit. Coefficient a varies by more than one order of magnitude — about 0.5 in arid US Southwest, around 8 in the Pacific Northwest. Always cross-check with locally gauged data from the same hydrologic region, and adjust the coefficient before relying on a regional-curve target.

How to Use

  1. Enter channel geometry: number of meander amplitude cycles (aNum 2-8) and their range (aRange 5-40 m), plus width at bankfull (wNum 4-20 m, wRange ±2 m)
  2. Input depth parameters: number of depth measurements (dNum 3-6) and their variability (dRange 0.3-1.5 m) to define cross-sectional profile
  3. Specify slope (sNum 0.002-0.05 m/m, sRange ±0.015 m/m) and click Calculate to solve Manning equation: Q_bf = (1/n)×A_xs×R_h^(2/3)×S^0.5, where n=0.04 for restored gravel-bed streams

Worked Example

For a straightened 8 m wide, 1.2 m deep alluvial channel in Colorado with slope 0.008 m/m: cross-sectional area A_xs = 9.6 m², hydraulic radius R_h = 1.85 m, Manning coefficient n = 0.04, yielding V = 0.68 m/s and bankfull discharge Q_bf = 6.5 m³/s. Regional estimate (USGS regression for headwater reaches) predicts Q_reg = 7.2 m³/s. Width/depth ratio W/D = 6.7 suggests Rosgen Type C3 (moderate entrenched, sinuous). Restoring meanders with aNum=4, aRange=12 m increases residence time without exceeding regional Q estimates.

Practical Notes

  1. Use dRange to represent seasonal variation and scour depths—overestimating depth by 0.3 m increases Q_bf by ~30%, risking underdesigned riprap protection in restored bends
  2. Set aRange conservatively: excessive meander amplitude (>30 m) in low-gradient streams (S<0.003) may trap sediment and cause localized aggradation
  3. Compare Q_bf to Q_reg: if Q_bf exceeds regional estimate by >15%, validate Manning coefficient or reduce slope assumption—confirms need for habitat roughness (logs, stone vanes)
  4. W/D ratios <4 indicate narrow, deep incised channels requiring grade-control structures; ratios >8 signal braided tendency, needing lateral confinement