Project 6

Losses for Design Flood Estimation

There are two distinct and different components to this project which are the development of ¿statistical loss models¿ for catchment simulation and the development of ¿deterministic loss models¿ for catchment simulation.

For this first subproject, there is a need to develop further the statistical interpretation of loss models that was proposed originally by Walsh et al. (1991), ie a loss model that results in transformation of rainfall frequency into flood frequency. This model was developed further by Rahman et al. (2000). At present there is some information available for New South Wales and for Victoria (available from a CRC Catchment Hydrology project) and for southern Queensland (QUT project). However, this information is not consistent and the various studies used different approaches and assumptions in determining the information. Furthermore, values from some of these studies will be supplanted by new rainfall data and hence these previous studies will need replication, consolidation and expansion. As part of this subproject, there is a need also to define the influence of alternative process models and software implementations of these on statistical loss model parameters.

For the second subproject, there is a need to develop appropriate loss models when maintenance of frequency is not required as part of the catchment simulation. There is a need to correlate information about soil type and infiltration rates (see Tayfur, 2001) with available information on soil classifications and soil horizons as available from soil maps (note that many of these soil maps are available in a digital format). Additionally, there is a need to consider the utility (or otherwise) of the rainfall excess model and whether this form of loss model needs further development. As part of this second subproject, there is a need to develop measures of the parameter variability and the statistical distribution. This information will be needed for application with Monte Carlo approaches for estimation of flows with a given exceedence probability; see for example Blaikie and Ball (2005).

ARR Project 6 ¿ Loss models for catchment simulation ¿ consists of four phases of work as defined in the outcomes of the workshop of experts in the field held in 2009.

Phase 1 ¿ Pilot Study for Rural Catchments. Involves a pilot study on a limited number of catchments that trials potential loss models to test whether they are suited for parameterisation and application to design flood estimation for ungauged catchments.
Carried out by SKM ¿ Stage 2

Phase 2 ¿ Collate Data for Rural Catchments. Streamflow and rainfall data for a large number of catchments across Australia will be collated for subsequent analysis.
Carried out by SKM ¿ Stage 2

Phase 3 ¿ Urban Losses. The phase involves analysis of losses for urban areas and estimation of impervious areas.
Carried out by Cardno ¿ Stage 2

Phase 4 ¿ Analysis of Data for Catchments across Australia (currently being undertaken by SKM in Stage 3). Loss values will be derived in a consistent manner from the analysis of recorded streamflow and rainfall from catchments across Australia. The results will then be analysed to determine the distribution of loss values, correlation between loss parameters and variation with storm severity, duration and season. Finally, prediction equations will be developed that relate the loss values to catchment characteristics.

PROJECT START: Stage 2 July 2009 PROJECT FINISH: Late 2014 CONTRACTOR: Discussion Paper Prepared by SKM, Rural losses Phases 1, 2 and 4 -SKM, Urban losses Phase 3 - Cardno