FLOOD01798 - Laserfiche WebLink

Home Browse Search

U~IU 1 I I I 1 I I 1 I I I I , I I individual location and the desiQu values denved <ISing the generalised procedures detailed in Book II Section 2. Accordingly, it is recommended that skilled hydrometeor- ological advice be sought before using revised design rainfalls in this AEP range. (b) Areal delliglt rainfall deplhs Point design rainfall depths are converted to areal rainfall depths by selecting a representative average point rainfall depth for the catchment and multiplying by an areal reduction factor corresponding to the total catchment area. In some circumstances it may be necessary to accommodate significant variation in design rainfall characteristics by dividing the catchment into two or more sub-catchments (see Section 3.10b). The areal reduction factors presented in Book II Section 1 are based on an analysis of a limited number of rainfall stations in the Chicago area and Anzona, U.SA A number of recent studies have yielded appreciably different areal reduction factors (e.g. NERC, 1975; Masters and Irish, 1994), and there is some evidence to suggest that the current areal reduction factors are also not applicable to arid regions (Kemp and Daniell, 1997). The most extensive study in Australia to date has been undertaken by Siriwardena and Weinmann (1996) using daily rainfall data from over 2000 rain gauges located across Victoria. They derived areal reduction factors for durations of 18 to 120 hDurs, catchment areas of between 1 and 10000 km', and for AEPs from 1 in.2 to 1 in 2000 (Grayson et aI., 1996). Their factors are up to 10% lower than the values presented in Book II Section 1. Similar work is currently underway in other states to determine specific regional values; tt is recommended that Ihese revised factors be used in lieu of the values presented in BODk II Section 1. Where such revised factors are not available practitioners should use their own judgement as to whether the factors derived for other regions of Australia are mDre appropnate than the values presented in Book II Section 1. (c) Design rainfalls for longer durations In some instances it is necessary to derive design rainfalls with durations longer than the upper limit of 72 hours provided by the procedures in Book II Section 1. The most common application requiring long duratiDn rainfalls is the determination of maximum outflows from reservDirs, though long duration events should be considered whenever the performance of a hydrological system is appreciably affected by hydrograph volume as well as peak. Longer duration design rainfall estimates for Rare to Extreme events are available from some regional methods; thus to construct frequency curves for all durations Df interest tt may be necessary to extrapolate from the design information provided in Book II, Section 1. Unfortunately there is no established procedure for denving long duration rainfalls from the design information provided in Book II Section 1. One possible approach is to undertake a frequency analysis of local gauged data for 72 hour events as well as for the longer durations required. The shape of the frequency curves of the longer duration events can be assumed to be the same as the shape of the 72 hDur curve, where the location of the curves for lower AEP events is determined by the procedures described in the following sections. As discussed above, it will be necessary to adjust the 72 hour design rainfalls derived usi~g the local gauged data to conform with the design est,mate derive,,! using the Book II procedures. Alternatively, if regional estimates are available for Large to Rare events, the extrapDlation procedure described in Section 3.6.3 could be used to extend the design estimates to longer durations. I I I I I DOOK VI - t:.stlmatlon OT Large to t:.xtreme r-IOoaS 3.3 Extrapolation of f)e$ign Rainfalls to the Credible Limit 3.3.1 Areas Without Specific Regional Design Rainfall Information For those areas where no specific regional analyses have been undertaken, the credible limit of extrapolation is an AEP of 1 in 100 or, in selected cases, 1 in 500. The n~ce~s~ry infor~ati~n required to derive estimates up to thIS Iom.t,s prOVided In Book II, Section 1. If design rainfalls for rarer events are required then the procedures descnbed in Section 3.6.1 ShDUld be used to obtain approximate but generally conservative estimates. 3,3.2 Areas With Regional Design Rainfall Information (a) Regional frequency estimation methods The fundamental premise of regional methods is that the frequency distributions of standardised data from a number of sites are similar to that for the site of interest Thus, data from a large number of sites Can be pooled to Increase the accuracy of the design rainfall estimates at a single site. This pooling of regional information relies upon the Identification of a homogeneous region within which certain statistical characteristics of Rare rainfalls can be assumed to be similar. Many methods have been proposed for the analysis of Rare rainfalls (e.g. see National Research Council, 1988; Cunnane, 1989; Stedinger et aI., .1992), and historically perhaps the most popular approach IS based on derivation of a dimensionless regional frequency curve that is made site-specific by scaling using a location measure such as the at-site mean or median. Considerable recent research has demDnstrated the benefits to be gained from applying regional techniques to "real wo~id' data (e.g. Potter and Lettenmaier, 1990; Lettenma,er et aI., 1987). Considerable interest has been shown in regional frequency analysis procedures based on L-moments (Hosking and Wallis, 1997) and there are a number of published studies describing their application to the analysis of Rare rainfalls (e.g. Schaefer, 1990; American Meterological Society, 1992; McConachy et al. 1997). ' (b) Constraints on extrapolation Recent advances in regional estimation methods nDW enable design rainfalls to be derived for very low exceedance probabilities. Table 1 lists the optimistic limit of extrapolation for a range of different data sources but at present it is considered that the practical limit in Australia for rainfall frequency estimation is unlikely to be beyond an AEP of 1 in 10000. This limit is associated with the generally low density and short record of rainfall data though it should be stressed that these issues are th~ subject of ongoing research and the limit of extrapolation will depend upon the region under consideration. A further practical constraint on the limit of extrapolation is that the estimates should be consistent with Extreme design rainfalls based on PMP estimates (Section 3.6). The estimatiDn of design rainfalls for very low exceedance prDbabilities requires the investment of considerable resources. It is expected that such analyses will only be undertaken by hydrometeorological specialists using data from a wide region, where the costs of deriving the design estimates can be offset by the subsequent benefits accruing from their use in a large number of studies.