Search

International audience; Confronted with poor model performance, the Hydrologist has blamed data errors, non-Gaussianities, model nonlinearities, parameter uncertainty, and just about everything else from Pandorra's box. Moreover, recent work has suggested astonishing numerical artefacts may arise from poor model implementation. Yet progress in hydrology requires reducing predictive errors and disentangling individual sources of uncertainty. How can this be accomplished? First, robust and efficient numerical methods are needed to avoid unnecessary artefacts. Second, the formidable interaction between data and structural errors, irresolvable in the absence of independent knowledge, can be approached using statistical analysis of rain- and stream- gauge networks. Structural errors, the key unresolved challenge, can then be explored using flexible model configurations, paving the way for more stringent hypothesis-testing. Importantly, informative diagnostic measures are available for each component of the analysis. This paper surveys several recent developments along these research directions and indicates areas of ongoing and future interest.

This study tests the 1D/2D coupled modelling approach on its ability to emulate the âroadâ portion of runoff which is conveyed in the kerb/gutter alignment. The event modelled is an approximately 1 in 1 year ARI event in magnitude. Given itâs small ARI it may be supposed that it is mainly comprised of runoff from impervious surfaces (10% roof and 90% road and others). In the event flow from the road moves as sheet flow to the kerb where it is routed to inflow pits. Via the pits it enters the pipe and at the downstream end of the pipe it is gauged. The modelling system is found to be capable of matching the gauged hydrograph for the relatively small event very well. What is also found however is that model parameters that might, in a 1D model, be relatively well informed by field inspection, are instead required to be altered outside recommended ranges in order to optimise the fit between gauged and modelled flow. That is, there is a degree of conceptualisation in the 1D/2D system that was unexpected.

A number of areas associated with current practices for floodplain modelling have been highlighted during the revision of ARR as requiring further investigation. One of these areas and hence the reason for one of the revision projects is the need for guidance on 2D modelling in urban areas. This need for guidance has increased as 2D models have become the tool of choice for prediction of flood behaviour within urban floodplains throughout Australia. These 2D models also are being used to simulate rainfall-runoff processes that traditionally have been modelled using conceptual lumped storage rainfall-runoff routing models. The revision project implemented to satisfy this need adopted a non-traditional approach. Instead of inviting a small group of authors to write the project report, a team of authors that represented model suppliers, experienced users, academics and clients from across Australia was assembled. The team used an online wiki tool to develop content collaboratively. Outlined in this paper are the process, experiences learned and details of the final report arising from this project.

Since ARR 1987 improved statistical techniques have been presented which are ideal for producing regression relationships relating different catchment characteristics to peak flow. A unique feature of the quantile regression technique is that as different catchment characteristics become more or less important to peak flow their varying significance can be incorporated. For example the importance of losses declines for larger longer events where previous areas become saturated and start to contribute runoff. As a first step towards carrying out the quantile regression work, a database of available gauged data has been compiled. The data base is still being finalised however at the end of the proofing process, whereby gauges are checked for length of record, completeness, quality, ratings, stability of upstream development, it is expected that only a handful of gauges will be found to be suitable. Given the paucity of data it is not possible to establish relationships, by regression, between various catchment characteristics and peak flow for urban catchment. Thus the paper calls for a focus on urban gauging in Australia. This paper suggests that a goal be established, for the establishment of urban gauging sites across Australia (and the maintenance of existing sites) with the suggestion being that such work can be funded as part of climate change research.

Since publication of ARR87, available computer resources have increased thus enabling the analysis of a pluviograph record. This analysis typically involves extracting the significant storm events and determining the temporal patterns of these storms. These extracted temporal patterns will then be scaled for the design rainfall depth and filtered if necessary. Presented herein are the results of a study into the observed and filtered temporal patterns derived from the Sydney Observatory Hill pluviograph record. It was found that filtering changed the ranking of periods within the storm, changed the depth of antecedent rainfall prior to the intense part of the rainfall event, and changed the percentage of the storm occurring in the core of the storm (most intense part of the storm). Furthermore, if aggressive filtering of the observed temporal pattern is required, the user is probably better to not filter at all as the pattern will no longer resemble the original pattern.

An important component of estimating the flood flow is the temporal pattern associated with the design rainfall burst. The purpose of this paper, as part of the current revision of ARR, is to develop a new methodology for developing rainfall temporal patterns which will be presented together with a discussion of the implication for design flood flow estimation. Historical storms at two locations, namely Sydney and Richmond (New South Wales, Australia), were divided arbitrarily by meteorological origin into frontal or convective storms. These categories of historical storms were subdivided further by the location of the centre of mass of the storm. The relevant statistical properties for these storms were extracted for each of these categories and used in the development of the design temporal patterns. The proposed approach is based on a non-dimensional conditional random walk to generate an indefinite number of synthetic storms that preserve the internal structure of real storms. This approach allowed better representation of the temporal variability than the AVM method however, some problems with internal burst frequency were observed. Finally, rather than developing a single rainfall temporal pattern, the proposed approach develops a suite of synthetic design temporal patterns which enables the designer to assess the sensitivity of the peak flow estimate to the temporal pattern of rainfall.