SCS Runoff Methods for NZ Flood Studies: Which Variant Should You Use?

The Soil Conservation Service (SCS) runoff method is the workhorse of NZ flood hydrology, but there are at least three variants in common use, and choosing the wrong one can produce Q100 estimates that differ by 30% or more. On the Tukituki River flood study at Mt Herbert Road, three SCS variants were run in parallel, producing Q100 estimates of 894, 1,129, and 1,186 m³/s. Understanding why those numbers differ, and which to adopt, is what separates a defensible methodology from a desk-top guess.

What the SCS method actually does

At its core, the SCS method converts rainfall depth to runoff depth using a single parameter: the Curve Number (CN). The CN encodes the catchment's soil type, land cover, and antecedent moisture condition into a value between 0 and 100. Higher CN means more runoff. The relationship is non-linear: small changes in CN near the upper end of the range produce large changes in runoff volume.

The method was developed by the US Soil Conservation Service (now the NRCS) in the 1950s and 1960s. It was designed for small agricultural catchments in the United States. New Zealand practitioners adopted it because it is simple, well-documented, and works reasonably well for ungauged catchments where there is no recorded flow data to calibrate against.

The three variants in common NZ use

The differences between SCS variants come down to how they handle three things: initial abstraction, storm temporal distribution, and the unit hydrograph shape. Here is what each variant does differently.

1. SCS Standard (US original)

The original formulation uses an initial abstraction (Ia) equal to 0.2S, where S is the potential maximum retention derived from the CN. The storm profile is typically the SCS Type II distribution (a 24-hour nested storm with a sharp central peak). The unit hydrograph is triangular with a peak factor of 484.

This variant tends to produce higher peak flows because the Type II storm profile concentrates rainfall intensity into a short burst at the centre of the 24-hour period. It was developed for US conditions and does not necessarily reflect NZ storm patterns.

2. SCS with NZ storm profiles

This variant retains the SCS loss model (CN and Ia = 0.2S) but replaces the Type II storm distribution with a temporally-distributed storm derived from HIRDS data or local rainfall records. The storm duration is matched to the catchment's time of concentration rather than defaulting to 24 hours.

By using locally-derived temporal patterns, this approach typically produces lower peak flows than the US standard, because NZ storm profiles tend to be less sharply peaked than the Type II distribution. The difference is significant: on the Tukituki study, switching from Type II to a locally-nested profile reduced the Q100 estimate by approximately 20%.

3. Modified SCS (Ia = 0.05S)

Research published in the early 2000s (notably by Woodward et al., 2003) found that the original Ia = 0.2S relationship consistently over-estimated initial abstraction when compared against measured rainfall-runoff data. The modified approach uses Ia = 0.05S, which means less rainfall is "lost" before runoff begins.

This variant produces higher runoff volumes and, consequently, higher peak flows than the standard SCS formulation with the same CN. On the Tukituki study, the modified initial abstraction increased the Q100 estimate by approximately 5 to 8% compared to the standard Ia = 0.2S with the same storm profile.

Why the Tukituki results differed by 30%

The three Q100 estimates from the Tukituki River study resulted from combining these variants:

• 894 m³/s: SCS with NZ storm profile and standard Ia = 0.2S. The locally-derived temporal pattern and higher initial abstraction produced the lowest estimate.
• 1,129 m³/s: Modified SCS (Ia = 0.05S) with NZ storm profile. Reducing the initial abstraction increased runoff volume substantially.
• 1,186 m³/s: SCS with US Type II storm profile and Ia = 0.2S. The sharply-peaked storm distribution drove the highest peak flow despite the higher initial abstraction.

The spread between the lowest and highest estimate is 33%. That is not a rounding difference. It is a methodological choice that directly affects whether a building platform is above or below the assessed flood level.

Which variant should you adopt?

There is no single correct answer, but there are defensible approaches. The key considerations are:

• Council expectations: Some councils specify which method to use in their district plan or engineering code of practice. Hawke's Bay Regional Council, for example, generally expects locally-derived storm profiles rather than US Type II distributions.
• Catchment scale: The SCS method was developed for small catchments (under 25 km²). For larger catchments like the Tukituki, it should be used as one of several methods, with results cross-checked against rational method estimates or regional flood frequency analysis.
• Sensitivity testing: Running multiple variants and reporting the range is better practice than selecting one variant and presenting it as definitive. The Tukituki study adopted the modified SCS (Ia = 0.05S) with NZ storm profiles as the primary estimate, then reported the full range for transparency.
• Calibration data: Where flow records exist, calibrate the CN and initial abstraction against observed events. This removes much of the uncertainty associated with choosing between variants.

Practical implications for flood hazard assessments

If you are commissioning a flood study for a resource consent application, the choice of SCS variant will directly influence the assessed flood level and, consequently, the minimum floor level and building platform constraints on your site. A 30% difference in Q100 translates to a meaningful difference in flood depth, particularly on flat terrain near river floodplains.

Ask your engineer which SCS variant they are using and why. If they are only running one variant without sensitivity analysis, that is a gap in the methodology. A robust flood study should present results from multiple approaches and justify the adopted value.