A flow-duration curve (FDC) is a graph showing the percentage of time that a specified discharge in a river or stream is equaled or exceeded during a defined period. Widely used in hydrology, it summarizes the distribution of high, intermediate, and low flows in a single curve. Unlike a hydrograph, which plots discharge in chronological order, a flow-duration curve ranks observations by magnitude and therefore does not show when particular flows occurred or how long individual events lasted.
Definition and interpretation
Discharge is the volume of water passing a cross-section per unit time, commonly expressed in cubic metres per second. A conventional flow-duration curve places discharge on the vertical axis and exceedance percentage on the horizontal axis, increasing from left to right. The curve consequently descends from high flows, exceeded infrequently, toward low flows, exceeded frequently.
A discharge denoted , for example, is equaled or exceeded approximately 95% of the time represented by the curve. It is therefore a low-flow measure, not a high-flow measure. Conversely, represents a relatively high discharge exceeded approximately 5% of the time. is the median discharge, subject to the conventions used to estimate quantiles.
This notation can be confused with ordinary statistical percentiles: an exceedance-based corresponds approximately to the fifth percentile of discharge when percentiles are defined by non-exceedance probability. Reports should state their convention explicitly.
For an equally spaced record containing observations, an empirical exceedance percentage at a threshold can be written as
where is an observed discharge and equals 1 when the stated condition is true and 0 otherwise. Thus, a flow-duration curve is closely related to the complementary cumulative probability distribution of discharge, with appropriate treatment of values equal to the threshold.
The percentage describes the sampled record directly. Treating it as a probability for future conditions requires additional assumptions about the record’s representativeness and the stability of the flow regime.
Construction
Data preparation
Flow-duration curves are commonly constructed from daily mean discharge, although hourly, monthly, or other averaging intervals may be used. Records usually originate from a stream gauge, where water levels are often converted into discharge using a rating curve.
The temporal resolution matters. A curve of daily means suppresses short-lived peaks and troughs, while a curve of monthly means smooths variability further. Curves based on different averaging intervals are therefore not directly interchangeable.
Before calculation, the record should be checked for missing values, changes in measurement practice, and changes caused by dams, diversions, or abstractions. A curve calculated from incomplete data describes the available observations; it does not necessarily represent the full period if the missing observations disproportionately concern floods, dry seasons, or other particular conditions.
Ranking and plotting positions
A common procedure is to:
- Select a period and a consistent discharge averaging interval.
- Sort the observations from largest to smallest.
- Assign an exceedance percentage to each ranked observation.
- Plot discharge against that percentage.
One widely used plotting-position formula is
where denotes the largest discharge and is the number of observations. This convention avoids assigning the largest and smallest observations probabilities of exactly 0% and 100%. It is an estimate of plotting position rather than the exact fraction of observations equaling or exceeding each value. Other conventions produce slightly different results, particularly near the extremes.
Tied observations, especially repeated zero flows, require consistent handling. Discharge is often plotted on a logarithmic scale to display high and low flows together, but zero values cannot be shown directly on such a scale.
Hydrological meaning of curve shape
The shape of a flow-duration curve reflects how a drainage basin transforms and stores incoming water.
A steep curve generally indicates a large contrast between high and low flows. Such variability may be associated with strongly seasonal precipitation, rapid runoff generation, limited storage, or intermittent streamflow. A relatively flat curve indicates less variation across the displayed exceedance range and may reflect sustained groundwater discharge, lake storage, or regulated releases.
Different portions of the curve emphasize different behavior:
- The high-flow end reflects relatively rare large discharges and is sensitive to storms, snowmelt, and flood regulation.
- The middle section describes more frequently occurring flows and the overall variability of the discharge regime.
- The low-flow end reflects dry-period conditions, storage depletion, abstractions, and contributions from baseflow.
These interpretations are not unique diagnoses. Similar curves can result from different combinations of geology, climate, land use, and water management. For example, storage in an aquifer and controlled releases from a reservoir may both sustain low flows, despite operating through different mechanisms.
Where a stream periodically dries up, its curve reaches zero discharge. The proportion of zero-flow observations is an important characteristic that can be obscured by plotting or interpolation choices.
Variants and comparison between catchments
A period-of-record curve pools all observations from a selected record. Long records provide a broad sample of conditions, but a single pooled curve can conceal seasonal differences and changes over time.
A seasonal curve uses observations from a particular season or group of months. It may distinguish, for example, wet-season availability from dry-season scarcity. Its percentages apply only to the selected season, not to the entire year.
An annual flow-duration curve is calculated separately for each year. Collections of annual curves reveal year-to-year variability and can be summarized at selected exceedance percentages. A median of annual curves is generally not identical to a curve obtained by pooling every daily observation.
For catchment comparisons, discharge may be divided by drainage area or by a reference discharge such as the long-term mean. Area-normalized discharge helps compare runoff production, while mean-normalized discharge emphasizes relative variability. Neither normalization eliminates differences in climate, record length, regulation, or data quality.
At ungauged locations, curves may be estimated using regional relationships, catchment characteristics, or hydrological models. Such estimates depend on how well the regional data or model represent the target catchment.
Applications
Water availability and allocation
Flow-duration curves describe how frequently specified discharge thresholds have been available. They support assessments involving water supply, irrigation, abstractions, and competing demands.
For example, a daily of indicates that daily mean discharge equaled or exceeded that value on approximately 90% of the represented days. It does not mean that the full discharge is available for withdrawal: existing uses, ecological requirements, and operational constraints remain separate considerations.
Hydropower
For run-of-river hydropower, a curve helps estimate how often discharge is sufficient to operate turbines and how often it exceeds turbine capacity. Under simplified assumptions, electrical power is
where is water density, is gravitational acceleration, is overall efficiency, is hydraulic head, and is turbine discharge. A flow-duration curve can support an estimate of average power when operating thresholds, required bypass flows, capacity limits, and head are specified.
It does not, by itself, represent reservoir scheduling or energy production when storage and operating decisions depend on the chronological sequence of inflows.
Ecology and water quality
Flow-duration curves help characterize water availability for aquatic habitat and compare altered and less-altered flow regimes. They can also support load-duration curves, which combine discharge with a concentration target to express allowable pollutant loads under different flow conditions.
Ecological and water-quality processes often depend on timing, temperature, event duration, and recovery intervals. A duration curve therefore provides only part of the relevant information.
Limitations and uncertainty
The central limitation is loss of chronology. Two records can have identical flow-duration curves even if one contains a single prolonged drought and the other contains many short low-flow episodes. The curve cannot independently determine consecutive days below a threshold, flood rise rates, or storage requirements during a dry spell.
Exceedance percentages must also be distinguished from flood return periods. A discharge exceeded on 1% of days is not a “100-year flood”: daily-flow exceedance and annual-maximum flood frequency describe different statistical quantities.
Estimates near the ends of the curve are particularly sensitive to record length, exceptional events, and measurement uncertainty. Serial dependence between successive flows also means that the number of daily observations is not equivalent to the same number of independent observations.
Finally, climate change, land-use change, groundwater depletion, reservoir operation, and altered abstraction can make a historical curve unrepresentative of current or future conditions. Interpretation therefore depends on the stated observation period, sampling interval, completeness of the record, and degree of human alteration.