ReFH Method (ReFH Rainfall Runoff Method)

Data

Theory and Guidance

Introduction

The [ReFH*](Revitalised Flood Hydrograph) Method Boundary (ReFHBDY) derives an inflow hydrograph for a catchment or sub-catchment. The hydrograph then becomes a boundary condition, equivalent to a Flow Time Boundary, and can therefore form inputs to a hydrodynamic or routing model. This inflow can be applied at the upstream end of a reach or may be distributed over a reach in conjunction with the Lateral Inflow unit. When run in boundary mode, the resultant hydrograph may be viewed using the Graphical or Tabular postprocessors; furthermore, summary hydrograph (.zzh) and data input (*.zzb) files may be viewed prior to running a full simulation; this preview functionality is also available within the Flood Modeller interface.

Alternatively, the rainfall component of the ReFH Boundary can be used as a direct rainfall boundary (equivalent to an REBDY), when used in conjunction with a Lateral Inflow unit, by selecting the Hyetograph Boundary type option.

The ReFH Boundary is a rainfall-runoff model using procedures developed by UKCEH to update the FSR/FEH Rainfall Runoff Method, in response to concerns that the FSR/FEH design model tended to overestimate design floods. ReFH improves the way that design events are modelled and has a number of advantages over the FSR/FEH unit hydrograph and losses model. The key improvements are:

• a new baseflow model which provides a more objective method of separating total runoff into baseflow and direct runoff;
• a loss model based on the uniform PDM model of Moore (1985);
• improved handling of antecedent soil moisture conditions;
• a more flexible unit hydrograph shape,
• explicit reconciliation to statistical estimates of peak flow.

Users should note that in order to use local data with the ReFH Method, it is strongly recommended that model parameters are estimated using the full ReFH software developed by UKCEH and Wallingford Hydrosolutions (www.hydrosolutions.co.uk). The FEH supplementary Report No. 1 ('The Revitalised FSR/FEH Rainfall-Runoff Method') provides further guidance on the method and use of local data. In all cases users are advised to check the latest guidance on the use of local data.

Where the catchment is ungauged and there is no opportunity to use local data, then the ReFH Method Boundary should be used in preference to the FEH Method Boundary. This is because the catchment descriptor equations have undergone further calibration and improvement as part of the development of the ReFH method.

The ReFH hydrograph peak can be reconciled to a statistical estimate of the design peak and is now achieved by adjusting parameters in the loss model (C_ini and a_T). The reconciliation procedure is described further below.

It is strongly recommended that users of the ReFH Boundary read the Technical Report, produced by UKCEH and available from www.defra.gov.uk. The user should also be conversant with the methods described in the Flood Estimation Handbook and must be aware that design flow hydrographs produced by the ReFH method can still be highly inaccurate if the model parameters are based solely on catchment descriptors.

When using the ReFH Method users should also note that the use of this method for Scottish catchments has not been verified and as of July 2008 SEPA will not accept a design flood estimate based solely on the ReFH method, but will continue to accept the FSR/FEH rainfall-runoff method as an appropriate method for rivers in Scotland.

The Revitalised Method

As mentioned above, the ReFH model is composed of three sub-models, a loss model, a routing model and a baseflow model.

The loss model is based on the Probability Distributed Model (PDM). Effective rainfall is now evaluated sequentially during a storm and if soil storage is filled, runoff will equal 100%. The new model requires an estimate of the initial soil moisture content (C_ini) at time zero. C_ini is estimated on a seasonal basis (summer or winter) by the new model (i.e. using catchment descriptors - see equations below) or alternatively, the user can define a value (within the ReFH software C_ini can also be modelled from antecedent conditions). The model also requires an estimate for C_max which can also be obtained from catchment descriptors or defined by the user. A more complete description of the loss model can be found in section 3 of the Technical Report.

The routing model still uses an instantaneous unit hydrograph (IUH) and scales this to each catchment by area and time to peak (Tp). However, in ReFH the shape of the IUH has been changed by adding a kink at 2Tp along the descending limb. This gives the formerly triangular IUH a slightly more realistic profile, and introduces two new terms U_p and U_k (see equations for further details). Importantly, the ReFH model estimate of Tp is not equivalent to the FEH estimate of Tp, and users should be careful as to not confuse the two. The new ReFH model has led to a revision in the catchment descriptor predictor equation used to calculate (Tp). The new formula makes use of the same four catchment descriptors but the coefficients have changed (see equations for further details). A more complete description of the routing model can also be found in section 3 of the Technical Report.

The baseflow model in the FSR/FEH rainfall runoff model was a constant value, independent of direct runoff. The new ReFH baseflow model works on the assumption that the input to the baseflow reservoir is related to the rate of surface runoff. Model parameters are Baseflow Recharge (BR), Baseflow Lag (BL) and Initial Baseflow (BF₀). All three baseflow parameters can be estimated from catchment descriptors or the user can specify their own, using local data and the additional software from UKCEH.

FSR/FEH storm profiles continue to be used in the new ReFH model and the equations for calculating the Areal Reduction Factor (ARF) and Critical Storm Duration (D) remain unchanged (see equations). However, because the equation for estimating Tp from catchment descriptors has changed, the ReFH recommended storm duration (D) will be different from the value returned by using the FEHBDY.

The user should note that unlike the FEH method, there is always equivalence between the rainfall return period and the flood return period in ReFH. This has been achieved through introduction of a seasonal correction factor, where annual maximum rainfall statistics are converted to seasonal statistics (winter or summer) depending upon location, season (determined from catchment descriptors or specified by the user), duration and flood return period.

Peak Flow Reconciliation

The ReFH design (or catchment descriptor) method has been calibrated to ensure that flood frequency curves derived from the method correspond to the flood frequency curves derived through a pooled statistical analysis of AMAX events. For any given catchment at any given return period T, the difference between the peak flow estimate generated from the ReFH model (using T-year design rainfall) and the corresponding T-year estimate obtained from the pooled analysis was minimised by adjusting a free variable. In this case the free variable is termed a and the term will vary with return period. The ReFH method assumes that the a coefficient is equal to one for a 5-year return period (a₅ = 1).

Note that the method has been calibrated using URBEXT1990 values, and it is these that will be imported using the Flood Modeller "Import Catchment Descriptors" facility, rather than URBEXT2000 values.

Calibration of the ReFH design method proved problematic as the gradients of the flood frequency curves generated by the pooled statistical method and the ReFH model were quite different and so different correcting ratios would need to be applied to the 5 year flood from say the 100 year flood. This effectively meant that the ReFH model must lose increasingly more water at higher return periods than at lower return periods. The decision was taken to impose these extra losses on the design method by modifying the ReFH loss model. A catchment specific value of the initial soil moisture condition (C_ini) would be calculated when T = 5 year (and so a = 1), such that the 5 year estimate of peak flow from the ReFH method was reconciled with the 5 year estimate from the statistical pooled method. With C_ini fixed, the free variable a will then vary as the return period is increased. The ReFH method was calibrated up to the 150 year event and users should also note that only seven urban catchments were used in the calibration procedure. Thus there are a number of limitations associated with the method and the user is referred to section 6 of the Technical Report for further information.

The procedure adopted by UKCEH to calibrate the ReFH model is important to understand, and it is recommended that the user reconciles their ReFH design peak flow estimates to statistical pooled estimates using a similar method. Through trial and error, the user should first adjust C_ini for the 5 year return period until the ReFH peak flow corresponds to the statistical estimate of the 5 year flow. Once the 5 year peak flows have been reconciled by adjusting C_ini, the user can then reconcile peak flows at other return periods by adjusting the a factor (leaving C_ini at the value derived for the 5 year return period).

Using the ReFHBDY unit within the Flood Modeller interface

On opening an ReFHBDY unit within the Flood Modeller interface, the user is presented with five tabs: Catchment, Rainfall, Models, Options and Results. The first three tabs form the components of ReFH; the Options tab allows for some model adjustment and the Results tab displays the values for the generated hydrograph and its intermediate constituent components.

Prior to viewing the Results tab or running a simulation, the minimum input requirements are the Catchment Descriptors (Catchment Tab), Storm Duration, Data Interval and Return Period (Rainfall Tab). Although a recommended timestep range and storm duration are provided on the Results Tab, an estimate of these must still be provided initially.

Catchment descriptors may be imported from a csv file produced by the FEH CD-ROM (using the Import Catchment Descriptors button), or may be input manually. Note that only those descriptors used in the calculations are imported, in addition to the Easting and Northing grid reference, which are provided for information only (although may be used by postprocessing tools, e.g. TabularCSV, to extract coordinates).

All parameters have the option to use the 'ReFH design standard' or from 'Catchment Descriptors' as appropriate, and may be overridden by a user input ('User Defined'). By default, all options are initially set to 'ReFH design standard' or 'Catchment Descriptors'. Similarly, the season (Rainfall Tab) is initially set to 'DEFAULT', which will apply a summer or winter design, depending on the degree of urbanisation of the catchment (URBEXT > 0.125 indicates a Summer storm is recommended). The season can also be overridden explicitly. The "Reset all options" button will reset all of these options to 'ReFH design standard' or 'Catchment Descriptors' (or 'DEFAULT' in the case of season).

Donor correction factors (DCF), will be applied to all parameters whose method is set to 'Catchment Descriptors'. Thus, for a design standard calculation [uncorrected], the DCF should be set to 1 (the default).

Optimised model parameters may be imported from a CSV file produced by the ReFH software by using the "import parameters" button. The parameters that may be imported thus are: C_max, T_p, U_p, U_k, BL and BR. If using this facility, the appropriate methods will automatically be set to 'User Defined'.

The Options Tab allows the user to select the calculation method using the UKCEH-approved dynamic link library (DLL) calculation engine or the equations obtained from the Published Report. The 'Published Report' option uses equations and methods interpreted literally from the Published Report and is provided to allow greater flexibility, e.g. the Urban enhancements, a larger storm area to influence the calculated ARF value and to facilitate a user-input storm to be used. Furthermore, the dll is only available in 32-bit format and will not work with 64-bit versions of Flood Modeller. NB The storm area (if different from the catchment area) only affects the ARF calculation in 'Published Report' mode and has no effect on total rainfall volume.

The ReFH calculation engine DLL (FEHRRModel.dll) is produced by Wallingford HydroSolutions Ltd.

By clicking on the Results tab, the user is effectively launching a boundary mode simulation of Flood Modeller of a model containing the single ReFHBDY unit. This produces the output hydrograph and its constituents in tabulated form, and many of the intermediate calculated parameters. Recommended values for design storm duration and timestep are given, which can be subsequently checked against those originally entered. N.B. if subsequently adjusting the storm duration or time interval, these should be rounded so that the duration divided by the interval is an odd integer. The hydrographs and hyetographs may also be viewed graphically via the Plot button. The hydrograph (.zzh) and summary data (.zzb) output files are also produced and are able to be viewed via the appropriate button on the tab. Occasionally, the simulation may fail, e.g. due to bad input data or an inappropriate licence, and like any Flood Modeller simulation, will generate a run error and produce diagnostic output in such instances. The user is advised to check this output (N.B. The error dialogue box may appear partially hidden behind the "Please Wait..." dialogue box).

Equations

The following equations are those published in the Technical Report. These are the equations used when selecting the 'Published Report' calculation method.

C_MAX = 596.7 BFIHOST ^0.95 PROPWET ^-0.24

C_INI,winter = 0.5 C_MAX ( 1.2 - 1.7 BFIHOST + 0.82 PROPWET )

C_INI,summer = 0.5 C_MAX ( 0.9 - 0.82BFIHOST - 0.43 PROPWET )

T_p = 1.563 PROPWET ^-1.09 DPLBAR ^0.6 (1 + URBEXT) ^-3.34 DPSBAR ^-0.28

U_p = 0.65

U_k = 0.8

BL = 25.47 BFIHOST^0.47 PROPWET^-0.53 (1 + URBEXT)^-3.01 DPLBAR^0.21

BR = 3.751 BFIHOST^1.08 PROPWET^0.36

BF_0,winter = AREA (63.79 (CINI - 120.79) + 5.54 SAAR) × 10^-5

BF_0,summer = AREA (33.94 (CINI - 85.42) + 3.14 SAAR) × 10^-5

Recommended D = T_p(1 + SAAR/1000)

Recommended timestep is 10-20% of the time to peak, T_p.

Datafile Format

Line 1 - Keyword 'ReFHBDY' (case insensitive) #REVISION#1

Line 2 - Label

Line 3 - z, easting, northing

Line 4 - tdelay, ∆t, bfonly, SCFLAG, scfact, hymode, SCALING, minflow

Line 5 - CAREA, SAAR, URBEXT, SEASONFLAG, METHOD

Line 6 - STAREA, STDUR, SNRATE (SNRATE not currently used in ReFH)

Line 7 - ERFLAG, ARFFLAG, comment

Line 8 - P, T, arf, c, d1 ,d2 ,d3 ,e, f (P only used if ERFLAG = /'USER/'; the remainder only if ERFLAG = 'DESIGN')

Line 9 - RPFLAG, SCFFLAG, SCF

Line 10 - nrp

Line 10.1 to Line 17.nrp - rp

Line 11 - CMAXFLAG, CINIFLAG, ALPHAFLAG, comment

Line 12 - CMDCF, CMAX, CINI, ALPHA, BFIHOST

Line 13 - UHFLAG, TPFLAG, UPFLAG, UKFLAG

Line 14 - TPDCF, Tp0, TpT, DPLBAR, DPSBAR, PROPWET, Up, Uk (Tp only used if TPFLAG = 'USER')

Line 15 - nuh, units, uhfctr (nuh may equal zero if UHFLAG = 'DESIGN')

Line 15.1 to Line 15.nuh - uh

Line 16 - BLFLAG, BRFLAG, BF0FLAG

Line 17 - BLDCF, BL, BRDCF, BR, BF0