PRM Calculator

Open PRM Calculator

Estimate design floods for ungauged South African catchments using the Probabilistic Rational Method (PRM) — an adapted Rational formulation that replaces the subjective runoff coefficient with regionally-calibrated scaling factors. This guide covers the probabilistic scaling framework, the 41 homogeneous flood clusters, required inputs, the step-by-step computation, and how to use the results alongside other flood estimation methods.

Overview

The Probabilistic Rational Method (PRM) is an adapted Rational Method that uses regionally-calibrated probabilistic runoff coefficients to estimate design floods for ungauged catchments in South Africa. Developed by Calitz and Smithers (2020) as part of the NaFSAT v2.0 project, the PRM enhances the traditional Rational formulation by incorporating regional scaling factors derived from statistical analysis of observed flood data.

The method divides South Africa into 41 homogeneous flood clusters, each with a unique set of growth factors for seven standard return periods (2-year to 200-year). By combining these cluster-specific growth factors with catchment-specific scaling factors derived from area, rainfall, and distance to coast, the PRM produces probabilistic peak-discharge estimates that reflect the regional flood signature rather than a subjective coefficient choice.

Related methods

The PRM sits between the classical Rational Method (subjective $c$ values, preferred for small urban catchments) and the envelope-based RMF Calculator (Kovacs / Francou-Rodier upper-bound approach). For gauged sites, the Flood Frequency Analysis tool provides at-site quantiles that should generally take precedence over PRM estimates when a reliable annual-maximum series is available.

Methodology

Rational Method basis

The PRM is founded on the classical Rational formula, which relates peak discharge to rainfall intensity, runoff coefficient, and catchment area:

$$Q \;=\; \frac{C \cdot I \cdot A}{3.6}$$

Classical Rational formula — peak discharge from rainfall and area

where $Q$ is the peak discharge (m³/s), $C$ is the runoff coefficient (dimensionless), $I$ is the design rainfall intensity at the time of concentration (mm/hr), and $A$ is the catchment area (km²). The factor 3.6 performs the SI unit conversion from mm/hr · km² to m³/s.

The PRM extends this formulation by replacing the single deterministic runoff coefficient with a product of two terms: a 10-year scaling factor $\mathrm{SF}_{C10}$ derived from catchment geography, and a cluster-specific growth factor $C_T$ that adjusts the base estimate to the target return period. The resulting expression is dimensionally identical to the Rational formula but statistically grounded in observed flood frequency data.

Probabilistic scaling

Rather than relying on subjective runoff-coefficient selection from a land-use table, the PRM uses two statistically-derived scaling factors — $\mathrm{MAF}_{\mathrm{SF}}$ (mean annual flood scaling factor) and $\mathrm{SF}_{C10}$ — that explicitly account for three catchment descriptors: catchment area, mean annual precipitation, and distance to coast. These factors are combined with cluster-specific growth factors to produce return-period-dependent peak discharges.

Why probabilistic?

Traditional Rational Method coefficients are chosen subjectively from tables, introducing significant and hard-to-quantify uncertainty. The PRM’s regional scaling approach is calibrated against observed flood data within each homogeneous cluster, providing a more statistically rigorous estimate. The price paid for that rigour is reduced flexibility — the PRM cannot represent sharp local variations (e.g. a highly urbanised sub-area within a rural catchment) in the way a composite Rational $C$ value can.

Homogeneous clusters

South Africa is divided into 41 homogeneous flood clusters based on observed flood-frequency characteristics. Clusters were delineated so that sites within a given cluster share a similar standardised flood-frequency curve — that is, their dimensionless growth factors $C_T / C_{10}$ are statistically indistinguishable.

Each cluster has a unique set of regression coefficients $(a, b, c, d)$ for the two scaling factors, and a unique set of growth factors $C_T$ for the seven standard return periods. The cluster assignment for a catchment is determined automatically by a point-in-polygon spatial lookup on the catchment centroid coordinates.

Input parameters

The following parameters are required for a PRM calculation:

• DR10% (mm) — Design rainfall depth for the 10-year return period. Obtained from the Design Rainfall tool or entered manually.
• I(Tc,T) (mm/hr) — Design rainfall intensity for the catchment critical storm duration. Derived from the time of concentration and the design rainfall depth-duration relationship.
• Tc (hr) — Time of concentration, representing the time for runoff to travel from the most hydraulically remote point of the catchment to the outlet. See the Tc Calculator for supported methods.
• Catchment Area (km²) — The contributing drainage area. Can be entered manually, drawn as a polygon on the map, or imported from the Watershed Delineation tool.
• Location — Centroid coordinates (latitude and longitude) used for spatial lookups of the PRM cluster, MAP, and coast distance.

Input sources

Design rainfall parameters (DR10% and I(Tc,T)) can be sourced directly from the Design Rainfall tool within HydroDesign, ensuring consistency across your analysis. Where a full rainfall-runoff study is being assembled, use the same IDF surface for the Rational Method, SCS Method, and PRM so that differences between methods reflect the runoff logic rather than the rainfall input.

Computation

Spatial lookups

Using the catchment centroid coordinates, the tool performs three spatial lookups against South African reference grids:

• PRM Cluster — Point-in-polygon lookup on the 41-cluster layer to determine the homogeneous flood cluster (1–41). This cluster ID drives the choice of regression coefficients and growth factors.
• MAP — Mean Annual Precipitation (mm/yr) from the national rainfall grid. MAP is a proxy for the general climatic wetness of the catchment.
• Coast Distance (DC) — Distance from the catchment centroid to the nearest coastline (km). DC captures the maritime influence on flood-producing rainfall mechanisms.

Scaling factors

Two scaling factors are computed using cluster-specific regression coefficients $(a, b, c, d)$ — each scaling factor has its own coefficient set:

$$\mathrm{MAF}_{\mathrm{SF}} \;=\; \exp\bigl(a \cdot \ln A + b \cdot \ln \mathrm{MAP} + c \cdot \ln \mathrm{DC} + d\bigr)$$

Mean annual flood scaling factor

$$\mathrm{SF}_{C10} \;=\; \exp\bigl(a \cdot \ln A + b \cdot \ln \mathrm{MAP} + c \cdot \ln \mathrm{DC} + d\bigr)$$

10-year scaling factor

where $A$ is the catchment area (km²), $\mathrm{MAP}$ is mean annual precipitation (mm/yr), and $\mathrm{DC}$ is the distance to coast (km). Each scaling factor uses its own set of regression coefficients that are specific to the PRM cluster in which the catchment falls.

The log-linear regression form ensures positivity of the factor and reflects the typical multiplicative interaction between the three descriptors observed in flood-frequency data.

Peak discharge

The peak discharge for return period $T$ is calculated as:

$$Q_T \;=\; \frac{C_T \cdot \mathrm{SF}_{C10} \cdot I(T_c, T) \cdot A}{3.6}$$

PRM peak discharge for return period T

where $C_T$ is the growth factor for return period $T$ drawn from the cluster lookup table, $\mathrm{SF}_{C10}$ is the scaling factor described above, $I(T_c, T)$ is the design rainfall intensity (mm/hr) for the critical duration and target return period, and $A$ is the catchment area (km²). The division by 3.6 performs the SI unit conversion.

Growth factors

Growth factors $C_T$ are dimensionless multipliers that scale the base flood estimate to different return periods. They are derived from regional flood-frequency analysis of observed streamflow data within each homogeneous cluster. By construction $C_{10}$ is typically close to 1 — the 10-year event is the reference level against which the scaling factor $\mathrm{SF}_{C10}$ is calibrated.

Return periods

The PRM provides peak discharge estimates for seven standard return periods, each associated with a cluster-specific growth factor:

Return Period (yr)	AEP (%)	Growth Factor Key
2	50	$C_2$
5	20	$C_5$
10	10	$C_{10}$
20	5	$C_{20}$
50	2	$C_{50}$
100	1	$C_{100}$
200	0.5	$C_{200}$

Common engineering use

For typical stormwater drainage design, the 1:10 to 1:50 year return periods are most commonly used. The 1:100 and 1:200 year values are typically reserved for critical infrastructure such as bridges, large culverts, and small dams, and should be cross-checked against the RMF Calculator envelope and — where dam safety is in play — the PMP Calculator.

How to use

1. Open the PRM Calculator tool from the sidebar or start a new calculation from the dashboard.
2. Select a location on the map by clicking on the catchment centroid or entering coordinates manually. The tool only works within South Africa’s borders — a centroid outside the 41-cluster coverage will return an error.
3. Enter or import parameters. Provide the design rainfall depth (DR10%), rainfall intensity $I(T_c, T)$, time of concentration ($T_c$), and catchment area. These can be entered manually or imported from previous calculations via the Import from Rational / SCS buttons.
4. Click Calculate to run the PRM computation. The tool will automatically perform the spatial lookups, compute scaling factors, and calculate peak discharges for all seven return periods.
5. Review results in the results panel, which displays the peak discharge for each return period along with the intermediate parameters (cluster, MAP, coast distance, $\mathrm{MAF}_{\mathrm{SF}}$, $\mathrm{SF}_{C10}$).
6. Export the results to CSV, Word, or Excel for inclusion in the engineering report.

Best practice

Always compare PRM results with at least one other flood-estimation method (SCS, SDF, Unit Hydrograph, or Index Flood) to validate the magnitude of your design flood estimates. A PRM value that sits outside the envelope of the other methods by more than ±50 % usually indicates either an unusual input (check area and MAP first), or that the catchment sits near a cluster boundary and should be assessed against both adjacent clusters.

Limitations

• Geographic applicability. The PRM is applicable to South Africa only. The homogeneous clusters and regression coefficients are derived from South African flood data and must not be transferred to neighbouring SADC states or other regions.
• Ungauged catchments. The method is designed primarily for ungauged catchments. Where reliable observed streamflow data are available, at-site Flood Frequency Analysis will generally produce more defensible estimates.
• Comparison with other methods. Results should be compared with other flood-estimation methods (SCS, SDF, Unit Hydrograph) to assess consistency. Significant discrepancies may indicate issues with input data or with the applicability of one or more methods.
• Input data quality. The accuracy of PRM estimates depends heavily on the quality of input data — particularly the design rainfall depth, rainfall intensity, time of concentration, and catchment area delineation.
• Catchment area range. The regression coefficients were developed using catchments within a specific size range (broadly a few km² to several thousand km²). Application to very small (< ~5 km²) or very large (> ~10 000 km²) catchments outside this range may reduce reliability.
• Stationarity assumption. Like all regional methods calibrated against historical records, the PRM assumes that the underlying flood-generating processes are stationary. Climate change and landscape change effects are not explicitly modelled.

Engineering judgement

The PRM is one of several flood-estimation methods available in HydroDesign. Engineering judgement must always be applied when selecting design flood values, and results should be evaluated in the context of the specific project requirements and risk profile. For critical infrastructure, present the PRM alongside at least two independent methods and justify the adopted value explicitly.

References

• Calitz, J.P. & Smithers, J.C. (2020). Development and assessment of a probabilistic rational method for design flood estimation in South Africa. Water Research Commission Report, Pretoria.
• NaFSAT Version 2.0. National Flood Studies Assessment Tool. Water Research Commission, South Africa.
• Smithers, J.C. & Schulze, R.E. (2003). Design Rainfall Estimation in South Africa. Water Research Commission Report K5/1060. Pretoria.
• SANRAL. (2013). Drainage Manual (6th ed.). South African National Roads Agency, Pretoria. Chapter 3 — Hydrology.
• Alexander, W.J.R. (2003). Flood Risk Reduction Measures — Incorporating Flood Hydrology for Southern Africa. University of Pretoria, Department of Civil and Biosystems Engineering.

---

Open PRM Calculator