A streamflow recession curve describes how river discharge falls after a period of high flow when new inputs become small enough for stored water to dominate the response. Its shape can reveal how a catchment releases water from soils, groundwater, wetlands and other stores. It cannot, by itself, prove that every litre reaching the river came from an aquifer.
This makes recession analysis both useful and easy to overinterpret. A falling hydrograph may help compare dry-weather behaviour, investigate low flows or test a hydrological model. However, rainfall, snowmelt, evaporation, abstraction and reservoir operation can all alter the decline. Reading the curve well means combining a simple mathematical description with knowledge of the catchment and evidence about what happened during the selected period.
Key takeaways
- A recession is a falling-flow period associated with the release of stored water.
- Slow decline suggests sustained drainage, but does not uniquely identify groundwater.
- Exponential recession loses a constant fraction, not a constant volume, over equal intervals.
- Rainfall, snowmelt, regulation and water use can interrupt or distort the curve.
- Independent groundwater and weather observations strengthen any interpretation.
Where does river water come from after rain stops?
Rainfall and river flow do not start and stop together. Some rain reaches channels rapidly, while other water enters soil, fills depressions or moves into permeable ground. Water stored during earlier weeks or seasons may continue reaching the river long after the latest storm has ended.
A hydrograph plots discharge against time. Discharge is a rate of flow, usually expressed in cubic metres per second, rather than the total volume held in a channel. Following a storm peak, the hydrograph often drops quickly and then more slowly. Several pathways can contribute to this changing shape.
Early decline may include draining surface depressions and shallow soil pathways. Later flow can be more strongly influenced by deeper storage, but the transition is not a universal boundary between surface water and groundwater. Bank storage, in which river water temporarily enters nearby sediments during high stages, can subsequently return to the channel.
The wider explanation of water moving through surface and underground stores is therefore important. A recession curve records the combined outlet response of a connected system, not a labelled inventory of each source.
How to read a streamflow recession curve
Begin by identifying the measurement interval. Hourly data may show short interruptions that disappear in daily averages. Next, examine precipitation, temperature and any operational records. A period without rain is not necessarily a period without new input: melting snow or delayed drainage from a lake may continue feeding the river.
On an ordinary graph, equal vertical distances represent equal changes in discharge. An exponential decline appears curved because the absolute loss becomes smaller as flow falls. On a logarithmic discharge axis, the same idealised exponential decline appears as a straight line.
The logarithmic transformation is a diagnostic convenience, not evidence that nature must behave exponentially. Real recession segments can bend, change slope or combine several drainage responses. Zero flow cannot be plotted directly on a logarithmic axis, and values close to the measurement limit need special care.
A simple worked example
Suppose a hypothetical stream carries 8 cubic metres per second, then 6, then 4.5 on three successive days. Each daily value is three-quarters of the previous one. The flow falls by 25 per cent per day, but the absolute changes are 2 and 1.5 cubic metres per second.
If that simplified pattern continued, the next value would be 3.375 cubic metres per second. This is a mathematical illustration, not a forecast. A shower, a pumping change or a different storage response could break the pattern before the next measurement.
Contrast a stream falling from 8 to 6 to 4. Its absolute daily decrease is constant, but the proportional decrease grows. These two sequences can look similar over a short interval yet imply different extrapolations. Reporting the model and fitting interval is therefore essential.
The linear-reservoir idea
One useful conceptual model assumes that discharge is proportional to the amount of drainable water remaining in a store. A fuller store releases more water; as storage declines, release slows. If no new water enters and there are no other losses, this relationship produces exponential recession.
The model is called a linear reservoir because of the assumed proportional relationship between storage and outflow. It need not represent an actual reservoir behind a dam. It can serve as a simplified mathematical description of a catchment store.
A recession timescale describes how quickly that idealised store empties. A longer timescale implies a slower proportional decline. Different publications may use a decay constant, a daily recession factor or the time taken for flow to fall by one logarithmic cycle. These quantities are related but have different units and numerical meanings.
Before comparing studies, check the definition. A larger daily retention factor means slower decline, whereas a larger decay-rate coefficient means faster decline. Treating both as a generic “recession constant” without reading the equation can reverse the interpretation.
Why a real catchment is more complicated
Aquifers are not uniform tanks. Hydraulic gradients, saturated thickness and connected drainage areas can change as water levels fall. Different parts of a basin may stop contributing at different times. Fractures and porous sediments can also release water on contrasting timescales.
Vegetation adds another pathway out of storage. During a warm growing season, evapotranspiration can steepen a recession even when the underground geology is unchanged. Comparing summer and winter curves without considering plant activity can therefore produce a misleading geological explanation.
What recession analysis can reveal
Repeated observations can show whether a river usually sustains flow between storms or loses it rapidly. This is useful when examining drought sensitivity, ecological low-flow conditions and the performance of catchment models. The comparison is strongest when measurement methods and seasonal conditions are similar.
Recession analysis can also help identify changes worth investigating. A stream that now declines faster than during comparable historical periods might have experienced altered abstraction, drainage, vegetation or connectivity. The curve identifies a pattern; additional evidence is needed to attribute its cause.
Understanding why river flow changes through the year helps separate an unusual recession from a normal seasonal transition. A dry-season decline should not automatically be interpreted as a permanent loss of catchment storage.
Some methods use recession characteristics to estimate groundwater recharge or discharge. These estimates depend on assumptions about flow pathways and water balance. A method designed for diffuse recharge and groundwater discharge to streams may be inappropriate where reservoirs, wetlands or snowmelt dominate the observed decline.
Why groundwater is not the only explanation
A slow, smooth recession can emerge from more than one store. A lake draining through an outlet, a wetland releasing water and a bank-storage zone returning river water can all prolong flow. Several sources combined may resemble one exponential curve over a limited period.
This is an example of non-uniqueness: different internal systems can produce similar observations at the outlet. The problem does not make recession analysis useless, but it changes the strength of the conclusion. “Consistent with sustained storage release” is often better supported than “proves a particular aquifer supplied the water”.
Groundwater levels in observation wells can test whether the surrounding aquifer is declining in a compatible way. Water temperature, electrical conductivity and chemical or isotopic tracers may provide additional clues. None is automatically decisive, but agreement among independent measurements narrows the plausible explanations.
Choosing and combining recession segments
A long record contains many falling-flow segments. Analysts may select periods with little rainfall, remove obvious regulation effects and separate seasons. Selection rules should be stated before comparing results, because choosing only the smoothest curves can hide important catchment behaviour.
Short segments offer many examples but can be strongly influenced by individual storms. Long segments reveal sustained drying but are more likely to include changes in weather, vegetation or management. There is no universally best duration: the appropriate choice depends on the question and the available observations.
Some analyses align multiple recessions to construct a representative or master curve. This can summarise common behaviour, but the resulting curve is a synthesis, not one event that actually occurred. Scatter around it is useful information about variability rather than merely an inconvenience to remove.
It is also important to distinguish interpolation from extrapolation. Describing flow within an observed range is less demanding than predicting conditions far below any measured discharge. A model that fits moderately low flows may fail as disconnected pools develop or a stream becomes dry.
Measurement uncertainty near low flow
Many continuous discharge records are calculated from water level using a stage–discharge relationship. At low flows, small errors in level or changes in the channel control can become important relative to the small discharge being estimated. Vegetation, sediment movement and debris can alter that relationship.
Background on how stream gauges support river-model calibration explains why a smooth-looking record still requires checking. Instrument resolution and the rating curve are part of the evidence, not separate technical details that can always be ignored.
Daily averaging can further hide short water-use cycles or intermittent releases. Before fitting a recession slope, inspect the original time resolution where possible and note any estimated or flagged values. A precise fitted coefficient cannot compensate for uncertain input observations.
Compare like with like
Catchment size can obscure a useful comparison. A large basin may have greater discharge throughout a recession simply because it drains more land. Comparing proportional decline or discharge per unit area can address different aspects of behaviour, but neither removes differences in geology, climate or regulation. State whether the question concerns total water delivery, drying rate or catchment response before choosing the comparison.
Conclusion: a diagnostic curve, not a source label
Recession curves turn the quiet period after a flood into a source of information. Their shapes help describe storage release, compare catchments and test hypotheses about low-flow behaviour. Their greatest value comes from making those hypotheses explicit and checking them against other observations.
The wider lesson is that a successful mathematical fit does not uniquely reveal the physical system behind it. Good hydrology combines the curve with weather, geology, water use and measurement quality, keeping the explanation no more specific than the evidence allows.
Frequently asked questions
Does a recession begin exactly when rain stops?
Not necessarily. Water may still be arriving through delayed pathways, and the hydrograph can continue rising after rainfall ends. Analysts define the selected falling-flow period explicitly.
Is baseflow always groundwater?
Groundwater is often important, but slowly varying flow may include other delayed sources. A hydrograph separation is a model-based division, not direct identification of every water molecule.
Why use a logarithmic axis?
It makes a constant proportional decline appear straight, helping test an exponential approximation. It does not establish that the approximation is physically unique.
Can recession curves estimate recharge?
Some methods use them for that purpose, but only with assumptions about storage, losses and stream–groundwater connections. Independent checks are needed, especially in regulated or snow-dominated basins.
Does a slower recession always mean more water is stored?
No. It describes the rate of decline relative to the chosen model. Storage size, hydraulic connectivity, ongoing inputs and other losses all influence the observed response.
