A rating curve is a relationship used in hydrology to estimate the volume of water flowing past a particular location from the observed water level. In its most common form, it relates stage, the elevation of the water surface above a reference datum, to discharge, the volume passing a cross-section per unit time. Established from field measurements and hydraulic analysis, the relationship allows a stream gauge to convert a continuous record of water level into a continuous discharge record. A simple rating curve assumes that each stage corresponds to one discharge; where this assumption fails, additional observations or more complex relationships are required.
Physical basis
Water level is generally easier to record continuously than discharge. A stage sensor measures the elevation of the water surface, whereas determining discharge ordinarily requires measurements of channel area and water velocity.
Discharge is expressed as
where is discharge, is the wetted cross-sectional area, and is the component of velocity normal to the cross-section. A common practical approximation is , where is the cross-sectional mean velocity. Discharge is usually reported in cubic metres per second or cubic feet per second.
A stage–discharge relationship exists because increasing flow generally raises the water surface, enlarges the wetted area, and increases velocity. The precise relationship depends on channel shape, bed slope, hydraulic resistance, and downstream conditions. Stage is not necessarily water depth: its zero is a chosen gauge datum, which may lie above or below the channel bed.
The channel feature or reach governing the relationship is called the hydraulic control:
- Section control occurs where a relatively short feature, such as a rock ledge, riffle, or constructed weir, governs the upstream stage.
- Channel control occurs where the geometry, slope, and resistance of a longer reach govern stage.
- Compound control occurs when different features govern different flow ranges.
A site may have section control at low flows, channel control at intermediate flows, and a different relationship after water spreads onto the floodplain.
Establishing a rating curve
A rating is developed from paired observations of stage and independently measured discharge. Measurements should cover the range of conditions encountered at the station, including low flows, ordinary flows, and floods where practicable.
Common discharge-measurement methods include:
- Velocity–area measurements, using current meters at multiple locations across a cross-section.
- Acoustic Doppler measurements, which use sound to measure water velocity and can sample much of the cross-section from a moving instrument.
- Dilution gauging, which infers discharge from the mixing and downstream concentration of an introduced tracer.
- Hydraulic structures, such as calibrated weirs and flumes, whose geometry provides a defined relationship between upstream water level and flow.
Measurements are plotted and used for calibration of a suitable equation, piecewise relationship, or tabulated rating. The procedure combines statistical fitting with knowledge of channel hydraulics; a close numerical fit alone does not establish that a rating is physically appropriate.
The resulting relationship is applied to recorded stage to produce a discharge time series and a hydrograph. Measurements collected afterward test whether the rating remains valid. A rating therefore has a period of applicability rather than being a permanent property of a site.
Mathematical representation
A widely used form is the power law
where:
- is stage;
- is an effective zero-flow stage;
- is a coefficient determined by site geometry, hydraulic conditions, and measurement units;
- is an exponent describing how rapidly discharge increases with effective stage.
The parameter is not necessarily the gauge datum or the lowest surveyed bed elevation. It represents the stage offset appropriate to the controlling hydraulic relationship and may differ between rating segments.
For a selected , taking logarithms gives
This explains the traditional use of logarithmic rating plots, on which a power-law segment appears as a straight line. However, fitting in logarithmic space changes the treatment of errors; transforming predictions back to discharge requires attention to possible statistical bias.
Hydraulic interpretation of the exponent
Idealized hydraulic relations help explain the power-law form. For a wide rectangular channel under approximately uniform flow, Manning’s equation gives
in its standard SI form, where is the roughness coefficient, the hydraulic radius, and the friction slope. If width, roughness, and slope remain constant, then is proportional to depth and , giving .
This is an illustrative result, not a universal rating exponent. Idealized weirs yield other exponents, while natural channels change shape and resistance as stage rises. Many operational ratings consequently use several connected segments rather than one power law.
Changing and non-unique relationships
Shifting controls
A rating can change when the channel bed or hydraulic control changes. Causes include erosion, deposition, movement of bed material, vegetation growth, debris accumulation, and construction. Floods may alter the channel sufficiently to require a new rating.
Temporary or gradual changes are often represented by shifts applied to a base rating. These may be expressed as corrections to stage before discharge is calculated. Shift corrections are supported by discharge measurements and field observations; they do not remove the need to inspect and reassess the site.
Backwater
Backwater occurs when downstream conditions influence upstream water level. Reservoir levels, tides, tributary inflows, or obstructions can raise stage without a corresponding increase in local discharge. Under these conditions, stage alone may not uniquely determine flow.
A stage–fall–discharge relationship uses water levels at two locations to estimate the water-surface fall over a reach. This adds information about the hydraulic gradient. At tidally affected sites, flow reversal and rapidly changing conditions may require velocity measurements or a hydraulic model.
Unsteady flow and hysteresis
During a flood, the water-surface slope and velocity can differ between the rising and falling limbs. Consequently, the same stage can occur at different discharges, producing a loop rather than a single curve. This behavior is called hysteresis.
In many river-flood situations, discharge at a given stage is greater on the rising limb than on the falling limb, although the direction and size of the loop depend on local conditions. Methods for addressing this include corrections based on the rate of stage change, additional slope observations, or unsteady-flow models.
Ice effects
Ice cover and ice accumulations change hydraulic resistance and the effective flow area. An open-water rating may therefore be invalid during ice-affected periods. Discharge estimates for these periods require separate measurements or methods and commonly have greater uncertainty.
Extensions and alternatives
An index-velocity method is useful where stage is a poor predictor of velocity. A fixed instrument measures a representative velocity, which is related to cross-sectional mean velocity through a calibrated relationship. A separate stage–area relationship supplies the wetted area, allowing discharge to be calculated as their product. This approach can accommodate variable backwater more effectively than a stage-only rating.
Hydraulic models can also estimate or extend stage–discharge relationships using surveyed channel geometry, roughness, structures, and boundary conditions. Such models remain sensitive to input uncertainty and require evaluation against observations where available.
The term rating curve is sometimes used more broadly. A sediment rating curve, for example, relates sediment transport or sediment concentration to water discharge. It is distinct from the stage–discharge relationship and does not replace it.
Uncertainty and extrapolation
Rating-derived discharge is an estimate, not a direct continuous measurement of flow. Important sources of measurement uncertainty include errors in stage observations and discharge measurements, imperfect model form, changes in the control, and extrapolation beyond the measured range.
For the power-law rating, small stage errors have an approximate relative effect
when other parameters are held fixed. Thus, a small stage error can produce a large relative discharge error near effective zero flow.
High-flow estimates present a different difficulty. The largest floods are often difficult or unsafe to measure, leaving the upper rating poorly constrained. Floodplain inundation, bridge effects, and changes in roughness can make simple extension of the measured relationship unreliable. Surveyed geometry, hydraulic calculations, and indirect flood-discharge estimates can help constrain extrapolation, but do not eliminate uncertainty.
Systematic rating errors can affect an entire discharge record. Uncertainty analysis therefore needs to consider errors shared across many observations, not only random errors at individual times.
Uses and interpretation
Rating curves underpin river-flow records used in flood forecasting, drought assessment, irrigation planning, water allocation, and water-balance analysis. They also support estimates of sediment and pollutant loads, for which errors in discharge propagate into calculated transported quantities.
A rating curve describes the local relationship between stage and discharge. It differs from a flow-duration curve, which describes how often particular discharges are equalled or exceeded, and from a flood-frequency relationship, which describes the statistical occurrence of flood magnitudes. Changes to a station’s rating, gauge datum, or treatment of exceptional conditions must be accounted for when interpreting long-term records.