Hydrographs and Detention
Detention can't make runoff disappear; it can only make it wait.
Before this Lesson
Pave a meadow, and the storm that used to reach the creek at 20 cfs now arrives at 60 cfs. The town wants the peak held to what it was, so you'll build a pond. How big? Neither number can say, because a peak is a rate and a pond holds a volume. The answer is in the whole storm, start to finish, drawn as a graph. By the end of this Lesson, you'll size that pond in three lines.
Flow against time
Stand at a watershed's outlet through a storm and write down the flow every few minutes. Plot it against time, and you have a hydrograph. The flow climbs, tops out, and then sinks back more slowly than it rose.
Everything in this Lesson rests on one fact about that graph: the area under it is a volume of water. Flow is volume per time, so flow times time is volume. One cfs running for one hour delivers 3,600 ft³, an amount you'll see written as 1 cfs-hr. So write the balance first, and read it off the graph: water in, water out, and the difference is whatever got stored.
The Handbook prints no hydrograph diagram and defines none of these terms, so they're yours to carry:
- Baseflow is what the stream carries anyway, mostly groundwater seeping in.
- Direct runoff is what the storm adds on top of the baseflow. Spread its volume evenly over the watershed and you get the depth of excess rainfall, the rain left after the ground has taken its share. That's the runoff depth from Curve Number Runoff, under a new name.
- The rising limb is the climb, as water from farther and farther away starts reaching the outlet.
- The time to peak runs from the start of the excess rainfall to the highest flow.
- The lag time runs from the center of the excess rainfall to the peak, so the time to peak is half the burst's duration plus the lag.
- The recession limb is the long fall after the peak, as water held up in soil, channels, and puddles drains out.
One inch, one shape
Now for a bold claim. Drop exactly 1 in. of excess rainfall evenly across a watershed over some stated duration, say 1 hr, and measure the direct runoff. That hydrograph is the watershed's 1-hour unit hydrograph, with ordinates in cfs per inch of excess rainfall. The claim is that this one curve predicts the watershed's response to any storm.
The Handbook's Hydrology/Water Resources section gives it one sentence, as the direct runoff from "one unit of rainfall occurring uniformly in space and time over a specified period of time." Read that unit as 1 in. of excess rainfall. Since it's direct runoff, there's no baseflow in it.
The area under it is that 1 in., spread over the whole watershed. The claim rests on three assumptions:
- Linear. Twice the excess rainfall gives twice the flow at every moment, on the same time base.
- Time-invariant. The same burst an hour later gives the same hydrograph an hour later. The watershed responds the same way every time.
- Uniform, for a stated duration. The excess rainfall falls evenly over the watershed and steadily through the duration. A 1-hour unit hydrograph is built for 1-hour bursts.
None of this is strictly true, since wet ground answers faster than dry. It's a model, simple enough to use by hand and close enough to earn its keep.
Scale, lag, sum
The first two assumptions are all the machinery you need. A storm with several bursts of excess rainfall is just several unit hydrographs: scale each one by its burst's depth, start each one when its burst starts, and add them at every moment. Take a 1-hour unit hydrograph, , with 1.0 in. of excess in the first hour and 2.0 in. in the second.
| Hour | U (cfs/in.) | 1.0 in. | 2.0 in., 1 hr later | Direct runoff |
|---|---|---|---|---|
| 0 | 0 | 0 | 0 | |
| 1 | 50 | 50 | 0 | 50 |
| 2 | 150 | 150 | 100 | 250 |
| 3 | 100 | 100 | 300 | 400 |
| 4 | 60 | 60 | 200 | 260 |
| 5 | 30 | 30 | 120 | 150 |
| 6 | 10 | 10 | 60 | 70 |
| 7 | 0 | 0 | 20 | 20 |
| 8 | 0 | 0 |
In symbols, with excess depths , , and in consecutive periods of the unit duration , the flow at any time is:
Look where the peak landed: hour 3, though the unit hydrograph itself peaks at hour 2, because the bigger burst came second. So sum the rows around the peak instead of assuming it. If a problem asks about one time, you need just that row: at hour 4, cfs.
The sum is direct runoff only, so add the baseflow to the final sum, once. And the scale factor is always excess rainfall. When a problem gives a storm depth and a curve number, find the runoff depth first.
With storm depth in inches, the Handbook's NRCS (SCS) Rainfall-Runoff section gives
That is a depth in inches, and it's your excess rainfall. For more, see Curve Number Runoff.
What paving does to the hydrograph
Now build a subdivision on the watershed and think about the graph. Two physical things change. Less rain soaks in, because roofs and streets have replaced soil: that's the higher curve number from Curve Number Runoff. And the water that runs off travels faster, through gutters and pipes instead of across grass: that's the shorter time of concentration from Time of Concentration and Design Storms. The first changes how much water there is, which is the area under the curve. The second changes how soon it gets there. Run those two forward and see what they do to the peak.
Making the water wait
A detention pond takes that flood and passes it through a small outlet. To see what it does, picture a bathtub with its drain open, and turn the tap up and then back down. The deeper the water, the harder the drain runs. While the tap beats the drain, the tub fills and the drain speeds up, even after you start closing the tap. The level stops rising at the moment the two are equal. That's the deepest the water gets, so it's also the hardest the drain ever runs.
So a routed outflow peaks where it crosses the falling limb of the inflow hydrograph, the same instant the pond is fullest. It can't peak with the inflow, because at that moment the pond is still filling fast.
Before the crossing, the water between the curves is going into the pond. After it, that same water comes back out, and the two areas match because the pond has no other exit. So detention lowers the peak and delays it, and it doesn't remove a single cubic foot. The release depends on the outlet, usually an orifice or a weir, whose equations are in Tanks, Siphons, and Weirs. The Handbook has nothing on detention itself.
Retention and infiltration are a different deal. A retention basin holds water with no routine surface release, until it evaporates or soaks in, and an infiltration trench sends it straight into the ground. That water never reaches the creek as runoff, so the runoff volume really does drop. Detention works on the peak. Retention and infiltration work on the volume too.
Sizing the pond from two triangles
Now you can size the pond. Draw the post-development inflow as a triangle with peak and time base , its full length from start to finish. The outflow is capped at the pre-development peak, , so draw it as a triangle too: it rises in a straight line from the start, reaches on the falling limb, just as the bathtub says, and then follows that limb down. Both peaks are the kind you found with in Time of Concentration and Design Storms.
The storage is the area between them. The two triangles share a base, so subtract their areas:
The time to peak never appears: slide the inflow's peak left or right, and the area doesn't change. The time base does appear, because the difference in the peaks is only a rate, and it takes a time to make it a volume. A real outflow curves, so this is an estimate, and a fair one: the routed pond above has the same peaks and base, and it stores about 38 cfs-hr where the triangles give 40.
Development often shortens the hydrograph too, so a problem may give each triangle its own base. Then find each area with its own base:
That one is a rougher estimate. It counts pre-development volume that would leave after the pond's fullest moment, so a real pond can need more. Real designs route the storm step by step, which is beyond this Lesson.
Sizing the pond from a table
Sometimes the inflow comes as a table of readings, and the outlet as a constant allowable release rate, . The balance is the same: whenever the inflow is above the release, the pond stores the difference. When the inflow is below it, the pond is either still empty, with the water just passing through, or already past its fullest and draining. Neither adds to the size you need.
So for each reading above the release, find the excess, add those up, and multiply by the time step:
In the figure, the excesses are 50, 80, and 40 cfs, which add to 170 cfs. Each one lasts a 30-minute step, or 1,800 sec, so ft³, about 7.0 acre-ft. This is the Trapezoidal Rule from the Handbook's Numerical Integration section, with the excess at zero at both ends. If a crossing falls between two readings, the sum is an estimate, but the method is the same.