Curve Number Runoff
The ground soaks up rain until it can't, and the curve number says how soon.
Two inches of rain fall on a wooded hillside and on the parking lot at its foot. The lot sheds about 1.8 in. of it, nearly everything. The hillside sheds about 0.06 in., barely a film. Same storm, thirty times the runoff. All of that comes from one number describing the ground and one short equation, and before long you'll work out the hillside's depth yourself.
The ground is a sponge
Hold a dry kitchen sponge under a slow tap. At first nothing drips, because the sponge keeps every drop. Then a little starts to come through. Leave it there, and more and more of the water passes straight through, because there's less and less room left. Ground in a storm behaves much the same way.
The NRCS curve number method turns that picture into arithmetic. (The agency used to be the Soil Conservation Service, so you'll also see it called SCS.) Write the balance first: every inch of rain either stays on the ground or runs off. The method tracks three depths, all in inches:
- , the storm's total rainfall;
- , the size of the sponge, which the Handbook calls the maximum basin retention;
- , the runoff.
The first inch nearly all stays. By the sixth, almost 90 percent runs off. There's no moment when the ground flips from holding to full. It just lets through a growing share of each extra inch.
That needs a warning. It's a depth: the sheet of water you'd get if all the runoff were spread evenly over the watershed. It isn't the flow in cfs of , from The Rational Method, even though the Handbook's Hydrology/Water Resources section prints the two a few lines apart with the same letter.
The curve number is a friendlier scale
is the physical quantity, but it's an awkward scale to work with: it's zero for a sealed surface, and it has no top. So hydrologists describe the ground with a dimensionless dial instead, the curve number , which tops out at 100. The Handbook's NRCS (SCS) Rainfall-Runoff section links the two, with in inches:
At , : no sponge at all, so every drop runs off. At , it's a 10 in. sponge. The 1,000 and the 10 aren't physics; they just set where the scale lands. What matters is the direction: a higher curve number means a smaller sponge, and the sponge grows faster and faster as drops.
What sets the dial? Three things about the ground, and nothing about the storm:
- Hydrologic soil group. Soils are sorted from group A (sandy, takes water in quickly) to group D (clayey, barely takes it in).
- Land cover. Woods hold water back, lawns less so, and pavement hardly at all.
- Antecedent moisture. How wet the ground already is when the storm starts. A damp sponge has less room.
| Land cover | Group A | Group B | Group C | Group D |
|---|---|---|---|---|
| Woods, good condition | 30 | 55 | 70 | 77 |
| Residential, 1/4-acre lots | 61 | 75 | 83 | 87 |
| Commercial and business | 89 | 92 | 94 | 95 |
| Paved parking and roofs | 98 | 98 | 98 | 98 |
Source: adapted from NRCS TR-55, Urban Hydrology for Small Watersheds, not the FE Reference Handbook
Read across a row and the soil gets tighter. Read down a column and the surface gets harder. The bottom row is the one to remember: impervious cover sits near 98 no matter what's underneath. It isn't 100 because even pavement wets and puddles a little. The Handbook prints the equations but no curve numbers, soil groups, or moisture conditions, so a problem has to hand you , and this table is here for the pattern, not for lookup.
From sponge to equation
Now bring in the storm. Before anything runs off, the ground takes a first gulp: rain wets leaves, fills puddles, and soaks into the surface, and none of it reaches the outlet. The method sets that gulp, the initial abstraction, at . The 0.2 isn't a law of nature. It's a fit to measurements from real watersheds.
That leaves inches of rain to split between sponge and runoff. The share that runs off should be small when this leftover rain is small compared with the sponge, and close to 1 when it's large. The simplest fraction that behaves that way is the leftover rain over itself plus :
Multiply it out, and you have the equation the Handbook prints, with all depths in inches:
The is nothing new, only the from that denominator. And the first form hands you a free check: when the leftover rain equals , exactly half of it runs off.
The smallest storm that runs off
Runoff starts only when the rain gets past the first gulp, so is also the smallest storm that produces any runoff at all. Anything smaller, and the answer is zero.
The equation won't tell you that. The Handbook doesn't name the initial abstraction or state any limit: just sits inside the equation. And that numerator is squared, which turns a negative into a positive. Take , where the gulp is 1.33 in., and a 1.0 in. storm:
That's a precise-looking answer to a question that never came up. The ground had room to spare, and the runoff is zero. So compare with first, every time.
When the ground is mixed
Few watersheds are all one thing. A real one is a patchwork of woods, yards, and pavement, each with its own curve number. The standard move is to blend them into one composite, weighted by area, so a big patch counts for more than a small one:
The Handbook prints no composite formula, but this is its weighted arithmetic mean, from Dispersion, Mean, Median, and Mode Values, with area as the weight. The result has to land between the lowest and highest curve numbers, pulled toward whatever covers the most ground, which makes a quick check. Then run it through the same two equations.
One honest note: because isn't a straight-line function of , blending first isn't quite the same as running each patch alone and adding up the runoff. It's the standard shortcut, and it's what problems expect.
From a depth to a volume
A depth isn't an amount of water until you say what it covers. Picture the runoff as one sheet, deep, lying over the whole watershed. Its volume is the depth times the area, and the only trap is the units: comes out in inches, so make it feet first.
With in acres, comes out in acre-ft: a sheet 1 ft deep over 1 acre. The Handbook's Units and Conversion Factors table gives 43,560 ft² per acre, so 1 acre-ft is 43,560 ft³. It has no row for acre-ft or square miles, but gives 640 acres per square mile.
What development does to a storm
Now put it together. Build on a meadow and hard surface replaces soil, so the curve number rises, the sponge shrinks, and the same storm sends more water downhill. The chart below draws the runoff equation at six curve numbers. A storm is a vertical line on it, and the gap between two curves along that line is the extra runoff.
Each curve sits on the axis until the rain passes its , then bends upward as the sponge fills, until it runs almost parallel to the no-sponge line. nearly lies on that line, which is why the parking lot gives back almost all of its rain.
To get the increase, work out twice with the same , and subtract. There's no shortcut: isn't proportional to , so the ratio of the curve numbers doesn't scale the runoff.
Notice what you haven't found in all of this: a flow rate. The method says how much water runs off, not how fast it leaves. Hydrographs and Detention picks up this depth, calling it the excess rainfall, and spreads it out over time.