FLOOD01798 - Laserfiche WebLink

Home Browse Search

L.l1V"'U' U I I I I I I I I Ii I I I I I I I 3.19 (Book V), and as described in the RORB Manual (Laurenson and Mein, 1997). Other network-type models such as the Watershed Bounded Network Model and RSWMI RAFTS (Book V Section 3) normally employ a value of m a little lower than 0.8. The standard model value can be used, especially for catchments with small flood plains, but extreme floods are likely to be overestimated to some extent. For catchments with appreciable flood plains, more reasonable estimates of extreme floods would be obtained by changing the value of m in the model to a value in the range 0.8 - 0.85. However, it must be recognised that in all cases, Extreme floods beyond the credible limit of extrapolation cannot be estimated with high accuracy. The Piecewise Linear Model uses an asymptotically linear storage-discharge relation, and therefore probably better allows for the behaviour of real catchments. This model has been incorporated into a program package for the Watershed Bounded Network Model, with options for modelling different degrees of non-linearity (Boyd et aI., 1987). Kinematic wave models normally involve a power- type relation, and thus probably lead to some over- estimation of extreme floods. Some slight increase of hydraulic roughness coefficients or equivalent adjustment would be appropriate. (b) Ungauged catchments For ungauged catchments, the model parameter values must be estimated from caiibration on nearby catchments or from regional relationships (see Book V Section 3; Dyer et al. 1996). The regional relationships for k in Equation 3.2 (Book V) are for an m of 0.8 for RORB, and for standard exponent values of other models where available. They will thus be directly applicable to catchments with small flood plains. For catchments with appreciable flood plains, it may be possible to adjust the value of k from a regional relationship for the higher value of m by means of equation 3.19 (Book V) and as described in the RORB Manual (Laurenson and Mein, 1997). An estimate would be necessary of the magnitude of the floods used in deriving the data on which the regional relationship was based (this estimate represents the credible limit of extrapolation associated with the derived regional relationship). In view of the likely errors in regional relationships this adjustment is hardly justified, and direct use of k values from a regional relationship with an m value of 0.85 would also be satisfactory where appreciable flood plains are present, even if this gives a tendency towards conservative estimates. If possible, the designer should check the magnitudes of the floods from which the regional relationship is derived as a guide to the likely conservatism of the estimate. 4.3.6 Specific Recommendations for Unit Hydrograph Models (a) Gauged catchments Where data are available, unit hydrographs should be derived from several large floods on the catchment. It is again desirable that the floods should have been of sufficient magnitude for overbank flow to have occurred. Because of non-lInearily in catchment response, use of an average unit hydrograph probably would lead to an underestimate of extreme floods. Two allernative procedures are recommended to rectify this: (i) Use of an arbilrary adjustmenllo fhe unit hydrograph shape. StaJ'S iR the precelfure _: 1. Determine an average of the derived unit hydro- graphs, taking care to obtain a valid answer either I I I U............. '" - l.-~UlllClUUIl VI LClI~e: ~v L.^~I'l;lI'O:;' .............'" by averaging the peaks in magnitude and timing, or by aligning the peaks of the individual unit hydrographs before averaging (see Book V Section 2). If any trends are evident in the derived unit hydrographs, a trend unit hydrograph corresponding to the maximum calibration floods should be determined rather than an overall average of all unit hydrographs. The average or trend average unit hydrograph is more appropriate than the extreme of all the derived unit hydrographs, as the latter may result from uneven areal distribution of rainfall, inadequate sampling of the rainfall, or other untypical conditions. 2. Keeping the time to peak unchanged, adjust the shape of the average unit hydrograph by increasing the peak flow and ordinates of the rising limb by the following percentages: , 20% for catchments with V-shaped valleys and small or no flood plains; and, . 15% for catchments where many of the valleys have appreciable flood plains. If only small floods with mainly in-bank flow are used in deriving unit hydrographs,' an increase in the peak of the average unit hydrograph of 30- 40% may be desirable in this step. 3. Sketch the shape of the falling limb to maintain the same volume as for the original unit hydrograph. Volume is easily checked as it is proportional to the sum of the unit hydrograph ordinates spaced at equal intervals. Shortening the base length by approximately 10 or 7% for unit hydrographs with small or large positive skewed shapes respectively is appropriate and helps in maintaining the correct volume. (ii) Use of the trend diagram of Body (1962). The log-log relation implies a power function, and the PMF adjusted by this procedure may thus be an overestimate. To minimise but not eliminate this tendency, it is recommended that the value obtained directly from the trend line should be used, and not the value from the upper 95% confidence limit. (b) Ungauged catchments For ungauged catchments, a synthetic unit hydrograph must be estimated from regional relationships, and similar considerations apply as discussed above for a derived average unit hydrograph. Suitable regional prediction equations are provided in Book V Section 2. 4.4 Baseflow The hydrograph models generally only give the direct storm runoff, and some baseflow must be added to obtain the total hydrograph. However for most of Australia, baseflow is small compared with direct runoff, especially for extreme floods. In some regions, it may be neglected. In the south west of Western Australia (Harvey, 1982) and possibly in some catchments in other regions, baseflow is an appreciable proportion of observed floods; care is then needed in its separation in anail'sis. Where baseflow is added, an approximate value is generally satisfactory. For the 1 in 50 AEP and 1 in 100 AEP design floods the average baseflow for the catchment could be used. as recommended in Book V, Section 2. Where there is clear evidence that initial baseflow increases with flood magnitude a constant base1'low 20% to 50% greater than the maximum value estimated in observed floods may be appropriate. If the difference