The Gold Standard of Hydrology | Lag Time, Time of Concentration and the Difference Between Catchment Clocks

Full chapter index

Before calculating a catchment response time, state the question. Are you estimating how long runoff takes to travel along a path? Are you choosing a duration for a peak-flow design method? Are you measuring delay between effective rainfall and a recorded hydrograph peak? These tasks use related language but different reference points.

Travel time is path-specific. A parcel or response moves through a segment of known or estimated length at a representative velocity. Under a simple constant-velocity model, time equals length divided by velocity. Real velocity varies, so the calculation is an approximation tied to a method and flow condition.

Time of concentration, Tc, is commonly defined in design hydrology as the time required for runoff to travel from the hydraulically most distant point in the drainage area to the outlet or point of interest. Hydraulically most distant means longest travel time, not necessarily longest straight-line or channel distance.

Lag has several uses. In hydrograph analysis, an important definition is the interval between the centre of mass of excess rainfall and the peak runoff rate. Another informal report may measure rainfall start to peak. Those two values can differ by more than an hour for the same storm. Name the reference points.

Time to peak is also used in hydrograph methods and may be related to rainfall duration and lag under a particular formulation. It should not be silently substituted for Tc. Formulas carry definitions from the method that created them.

Write units on every clock. Minutes and hours are easy to mix, especially when velocities are in metres per second and rainfall intervals are in minutes. A dimensional check catches many errors before any hydrologic interpretation begins.

Contents · Next chapter

The longest mapped flow line is not always the slowest. One path may be long but steep and channelised. Another may be shorter but cross a rough, shallow surface where movement is slow. Time of concentration follows the path with the greatest relevant travel time under the adopted representation.

Divide plausible routes into flow segments. A headwater route might begin as sheet flow, become shallow concentrated flow and then enter a defined channel. A different route might reach a drain quickly and travel faster despite a longer map distance. Calculate comparable times for each complete route before selecting the controlling one.

Do not add the slowest sheet segment from one route to the slowest channel segment from another. That constructs a path no water follows. Segment times must belong to one continuous route ending at the point of interest.

Flow regime boundaries matter. Sheet flow is usually limited to a relatively short distance in design methods; after flow concentrates, a different relation applies. The exact limit and method are not universal. Use the governing guidance rather than extending a convenient equation beyond its intended range.

The outlet is part of the definition. Moving the analysis point upstream removes downstream travel segments and can change the controlling source area. A nested catchment has different Tc values for different points of interest even though they share land.

Map evidence should include topography, drainage connections, barriers, culverts and field observations where possible. A digital elevation model can miss kerbs or pipes. A map-derived flow path is a hypothesis until checked against the drainage system and relevant event conditions.

Contents · Previous chapter · Next chapter

For one segment with length L and representative velocity v, travel time t = L/v. If L is in metres and v in metres per second, t is in seconds. Divide by 60 for minutes. This is the simplest transparent relation beneath many segmented methods.

Take a 600 m shallow concentrated segment with estimated mean velocity 0.50 m/s. Its travel time is 600/0.50 = 1,200 seconds = 20 minutes. A 1,800 m channel at 1.50 m/s also takes 1,200 seconds. Equal time does not imply equal length or flow regime.

Velocity depends on slope, roughness, hydraulic radius, depth and method. A value inferred for a substantial storm may not describe a small event that barely connects surface flow. Treat the estimate as conditional rather than a permanent property printed onto the landscape.

If velocity varies continuously, dividing total length by one guessed average can differ from integrating time along the path. Segmenting the route approximates that integral. More segments can improve representation, but only if their parameters are supported. False precision can increase when every uncertain patch receives a detailed number.

Check the direction of sensitivity. Increasing velocity decreases travel time. Doubling length at fixed velocity doubles time. If a spreadsheet shows the opposite, inspect cell references and units. These simple monotonic checks are valuable before comparing sophisticated formulas.

The calculation should retain source values. Store length, slope, roughness or velocity basis, flow type, unit conversion and result. A final Tc without the segment record is difficult to audit or update when the site changes.

Contents · Previous chapter · Next chapter

The fictional path has three segments: 120 m of sheet flow at 0.10 m/s, 600 m of shallow concentrated flow at 0.50 m/s and 1,800 m of channel flow at 1.50 m/s. Each segment takes 1,200 seconds or 20 minutes, giving total travel time of 60 minutes.

The arithmetic is exact for the stated constant velocities. The hydrologic estimate is not exact because the velocities are model inputs. Reporting Tc = 60.000 minutes would confuse arithmetic precision with physical certainty. A result such as about 60 minutes is more honest unless the governing method specifies reporting differently.

Now consider a second route: 300 m at 0.25 m/s followed by 2,400 m at 2.0 m/s. Its times are 20 minutes and 20 minutes, again totaling 40 minutes. Although the second route is longer overall, it is faster. The first route controls under this simplified comparison.

Change only the sheet-flow velocity on the first route from 0.10 to 0.05 m/s. Its first segment becomes 40 minutes and total travel time becomes 80 minutes. A modest-looking change in an early slow segment changes the controlling clock by one third.

If the sheet-flow length were incorrectly extended to 300 m at 0.10 m/s, that segment alone would take 50 minutes. The large result may reflect misuse of the flow-regime assumption rather than a genuinely slow catchment. Method limits belong in the calculation record.

The table makes each contribution visible. A reviewer can see that all three segments matter equally in the original example even though their lengths differ by a factor of fifteen. That is more informative than presenting only Tc = 1 hour.

Contents · Previous chapter · Next chapter

A hyetograph records rainfall through time. A loss model separates total rainfall from excess rainfall that contributes to the direct runoff representation. The centroid of excess rainfall can occur later than the centroid of total rainfall if early rainfall is largely lost to interception, infiltration or storage.

A hydrograph records discharge through time. Its peak may occur after excess rainfall centroid because runoff from different parts of the catchment is translated and stored. The interval between those reference points is a hydrograph lag under the stated definition.

Rainfall start to peak includes storm timing before the effective input centroid. In the fictional event, rainfall begins at 09:00, excess-rainfall centroid occurs at 10:05 and discharge peaks at 11:10. Start-to-peak is 130 minutes, while centroid-to-peak lag is 65 minutes.

Calling both values “lag” without qualification creates an apparent disagreement of 65 minutes. The catchment did not change; the reference points did. Every measured lag should name the rainfall quantity, centroid or start rule, discharge feature and time zone or clock basis where relevant.

Sampling resolution matters. A 15-minute rainfall record and hourly discharge record cannot support a peak time to the nearest minute without interpolation assumptions. Rounded timestamps can alter a short catchment’s estimated lag substantially.

The lesson transfers beyond hydrology: timing metrics remain meaningful only when their event definitions stay attached. A clean number cannot repair ambiguous clocks.

Contents · Previous chapter · Next chapter

The fictional table compares three clocks. Rainfall start to peak is 130 minutes, excess-rainfall centroid to peak is 65 minutes and segmented Tc is 60 minutes. Their proximity or difference should be interpreted through definitions, not as a pass-fail equality test.

Some hydrologic methods relate lag and Tc empirically. For example, NRCS guidance discusses a relation L = 0.6Tc for average natural watershed conditions and approximately uniform runoff distribution. That is a method-specific relation under stated conditions, not a universal identity between every measured storm lag and every Tc estimate.

If Tc = 60 minutes, the 0.6 relation gives L = 36 minutes, not the fictional observed 65 minutes. This mismatch can prompt questions about rainfall distribution, excess-rainfall estimation, storage, flow-path parameters, event selection and method applicability. It does not license changing a definition until the numbers match.

Alternatively, using measured L = 65 minutes in the inverse relation gives Tc ≈ 108.3 minutes. That inferred value answers the empirical-method question, while the 60-minute segmented estimate answers a path-velocity model. Their difference is diagnostic evidence about assumptions and uncertainty.

Do not average the two into 84.2 minutes without a rationale. Different models are not independent measurements of the same perfectly defined quantity. Compare inputs, calibration domain and intended use first.

The gold-standard explanation states which clock enters the final design or analysis and why. A model may require lag, Tc or time to peak. Supply the parameter consistent with that model rather than the number whose magnitude looks most plausible.

Contents · Previous chapter · Next chapter

Consider two storms over the same catchment with the same total rainfall. Storm A delivers most excess rainfall early; Storm B delivers it late. Their excess-rainfall centroids differ. Even if the catchment’s routing response is represented by the same unit hydrograph, the calendar time of the discharge peak can change.

For a linear time-invariant teaching model, shifting the entire excess-rainfall pattern by 30 minutes shifts the direct-runoff hydrograph by 30 minutes. The centroid-to-peak lag remains the same, while rainfall-start-to-peak changes if the start definition does not shift equivalently. This simple example shows why the chosen clock matters.

Now change storm shape rather than merely shifting it. Two pulses can create overlapping runoff contributions. The hydrograph peak may align with the second pulse even though the first contributed substantial volume. Measured centroid-to-peak lag can differ from a single-pulse event.

Nonlinear losses add another change. A dry catchment may absorb early rainfall, moving the excess-rainfall centroid later. A wet catchment may produce excess earlier. The physical flow paths can be unchanged while the observed event timing changes through effective input.

For short urban catchments, rainfall movement across the area can matter. A storm travelling downstream may synchronise local runoff differently from one travelling upstream. A single area-average hyetograph can hide that spatial timing.

Observed lag is therefore an event property shaped by catchment and storm. Estimating a stable design parameter from one event requires caution. Use multiple suitable events, consistent definitions and an analysis that matches the intended model.

Contents · Previous chapter · Next chapter

Urbanisation can shorten some surface routes by adding efficient drains, increase connected impervious area and alter loss processes. These changes can reduce travel time and sharpen peaks, but the outcome depends on the network and storage. A detention basin can delay and attenuate runoff even within an urbanised catchment.

Vegetation, surface roughness and microtopography affect shallow travel. Channel modification changes velocity and storage. Culverts can accelerate flow under one condition and restrict it under another. A single land-cover label does not determine Tc.

In the fictional segmented path, replace the 600 m shallow segment at 0.50 m/s with a drain at 1.50 m/s. Its time falls from 20 to about 6.67 minutes, and total travel time falls from 60 to about 46.67 minutes. If a downstream detention structure adds an effective 25-minute delay for the target hydrograph feature, observed peak timing can become later despite the faster conveyance segment.

That 25-minute value is not simply added to Tc unless the chosen method defines it that way. Storage routing changes hydrograph shape, not just a parcel clock. Keep translation and attenuation concepts distinct even when both influence time to peak.

Seasonal or maintenance changes matter too. Debris, vegetation growth or sediment can alter a channel. A design parameter from an old clean network may not describe current condition. Field evidence and maintenance records help interpret discrepancies.

The changed-context rule is to rebuild the route and response model. Do not assume every intervention that increases velocity reduces every lag metric, or every storage feature increases the same clock by a fixed amount.

Contents · Previous chapter · Next chapter

Time-of-concentration methods include segmented travel-time approaches and empirical equations based on catchment characteristics. Different methods were developed for different settings and data. Applying several equations to one catchment can produce a wide range.

The first response should not be to select the middle value automatically. Inspect the calibration range, variables, flow regime and definition. An equation developed for small rural basins may be unsuitable for a highly connected urban drainage area. A method that ignores an important storage feature can be systematically misleading.

Parameter uncertainty also propagates. In the simple t = L/v relation, a 20 per cent lower velocity gives a 25 per cent longer time because velocity is in the denominator. Uncertain slow segments can dominate total uncertainty. Sensitivity analysis should vary plausible inputs one at a time and together.

Calibration against observed events can improve fit but may not transfer to events of different magnitude or antecedent condition. Validate on independent events where possible. Report whether the parameter was measured, calibrated, inferred from a formula or adopted from guidance.

Avoid precision theatre. Two equations returning 47.3 and 88.6 minutes do not justify an average of 67.95 minutes. Their spread may indicate structural uncertainty. A clear range with reasons can be more useful than a single unsupported decimal.

For design, follow the applicable authority’s method and document professional judgement. For learning, compare methods as models: what each assumes, what evidence it uses and which observation would challenge it.

Contents · Previous chapter · Next chapter

Mistake one is using the geometrically farthest point. The repair is to compare complete hydraulic travel times. Distance is one input, while velocity and route determine time.

Mistake two is adding unrelated slow segments from different routes. The repair is to preserve continuous paths to the same outlet.

Mistake three is calling rainfall-start-to-peak, centroid-to-peak lag and Tc the same quantity. The repair is to name both reference points and the method.

Mistake four is assuming L = 0.6Tc is a definition. It is an empirical relation under stated conditions in a particular framework. Check applicability before use.

Mistake five is extending sheet-flow equations across distances or conditions outside guidance. The repair is to classify flow regimes and respect method limits.

Mistake six is using one event’s lag as a permanent catchment constant. The repair is to examine rainfall pattern, losses, antecedent condition and multiple events.

Mistake seven is reporting more precision than the inputs support. The repair is to retain exact arithmetic internally while communicating an appropriately rounded estimate and uncertainty.

Mistake eight is treating model disagreement as a calculation error without checking definitions. Two correct calculations can answer different hydrologic questions. Align the clocks before comparing values.

Feedback should begin at the first unsupported step. If a learner chose the longest line on a map, fix path logic before reviewing division. If the path is sound but units are mixed, repair the conversion before debating hydrologic method.

Contents · Previous chapter · Next chapter

A fictional catchment has Route A with 150 m at 0.10 m/s, 900 m at 0.75 m/s and 2,400 m at 2.0 m/s. Segment times are 1,500, 1,200 and 1,200 seconds: 25, 20 and 20 minutes. Route A totals 65 minutes.

Route B has 250 m at 0.20 m/s and 3,000 m at 1.5 m/s. Its times are 1,250 seconds, or about 20.83 minutes, and 2,000 seconds, or about 33.33 minutes. Total time is about 54.17 minutes. Route A controls despite being only slightly longer in mapped distance.

For one event, excess-rainfall centroid occurs at 14:20 and discharge peaks at 15:30. Measured hydrograph lag is 70 minutes. Rainfall began at 13:40, so start-to-peak is 110 minutes. State both rather than reporting an undefined lag.

Now increase Route A’s middle-segment velocity from 0.75 to 1.5 m/s. That segment falls from 20 to 10 minutes, and Route A totals 55 minutes. It still narrowly exceeds Route B’s 54.17 minutes and remains controlling. If the first segment velocity also rises from 0.10 to 0.15 m/s, its time falls from 25 to about 16.67 minutes, giving Route A about 46.67 minutes. Route B then controls.

The hydraulically most distant point can therefore switch after a targeted change. Updating only the old controlling route would miss that transition. Recalculate plausible routes after material changes.

Finally, suppose a later storm gives 50-minute centroid-to-peak lag. Do not conclude that Tc physically fell by 20 minutes without examining storm distribution, loss estimation and peak timing. The event clock changed; the path model requires its own evidence.

Contents · Previous chapter · Next chapter

Catchment timing becomes clear when every value carries its definition. Travel time follows a route. Tc selects the longest relevant route under a method. Hydrograph lag compares rainfall and runoff features. Time to peak belongs to a hydrograph formulation. They interact, but they should not be blended casually.

Continue with unit hydrographs and the limits of shift-and-add to see how rainfall pulses combine into runoff. Use rainfall time steps and equal storm totals for temporal aggregation, and recession constants and storage for the falling limb. Watersheds: equal volumes and different peaks connects timing with peak synchronisation.

The official USDA NRCS Part 630 Hydrology, Subpart F: Time of Concentration, amended June 2025, defines travel time, lag and Tc and states conditions for relationships used in its framework. The US Federal Highway Administration’s Urban Drainage Design Manual, HEC-22, fourth edition provides a wider design context. The teaching numbers here were created independently and do not represent a site design.

Questions readers often ask

Is Tc the time for every drop to leave? No; it is a response-time concept tied to the hydraulically most distant contributing path, not complete drainage. Is the longest stream the controlling path? Not necessarily; compare travel time. Can measured lag calibrate Tc? It can inform a model, but the relation and event definitions must match. Does urbanisation always shorten lag? Not automatically; conveyance, connected area, losses and storage interact.

For the final check, take one reported catchment time and ask four questions: from where, to where, measured or estimated how, and under which event or design condition? If the record answers all four, the clock can support a useful comparison.

One last transfer test is to move the outlet. Choose a point halfway down the original channel, remove the downstream segment and reconsider every candidate route. The old controlling point may no longer control, and the measured hydrograph feature belongs to a different drainage area. This small exercise prevents Tc from becoming a permanent catchment label. It is always attached to a point of interest, a representation of contributing flow and the conditions under which the estimate will be used.

Contents · Previous chapter · Return to the Hydrology guide