The State of Queensland, Australia, initiated the Brisbane River Catchment Flood Study in response to the devastating floods in January 2011 and subsequent recommendations of the Queensland Floods Commission of Inquiry. The goal of the hydrology phase of the study is to produce a set of competing methods for estimating design floods in the Brisbane River catchment, followed by an extensive reconciliation process to identify the most reliable design flood estimates for various locations and a wide range of return periods. The work is being delivered by a consortium of consultants: Aurecon, Royal HaskoningDHV, Deltares and Don Carroll Project Management.
Three approaches are used to estimate return periods of peak discharges and flow volumes:
- Flood Frequency Analysis,
- Design Event Approach; and,
- Monte Carlo Simulations.
With the Flood Frequency Analysis method, statistics of peak flows and flow volumes are estimated directly from observed flow records. The Design Event Approach and Monte Carlo Simulation (MCS) methods are both referred to as referred to as ‘rainfall based methods’, as they both rely on rainfall statistics in combination with a rainfall-runoff model to compute peaks flows and flow volumes at locations of interest.
The Design Event Approach, a more traditional rainfall-based method, relies on a number of simplifications. A fundamental assumption is that the return period of the flood is equal to the return period of the causal rainfall, which is not necessarily correct. Furthermore, the method assumes a single temporal pattern and spatially uniform rainfall distribution across the catchment. MCS removes many of the limitations common to Design Event methodologies. MCS is particularly advantageous in capturing the probability of flooding from larger river systems with many tributaries and catchments were additional uncertain factors, for example sea water level, antecedent moisture conditions and reservoir volumes, have a significant influence on the flood magnitude as well.
To estimate return periods of ‘moderate’ flood events, the Flood Frequency Analysis method is usually considered to be most accurate, as return periods are directly estimated from measurements. Extrapolation to large and rare events, on the other hand, is usually unreliable for this method and strongly influenced by the presence (or lack of) extreme events in the data record. For these types of events, the rainfall based methods are considered more reliable.
In the reconciliation process, significant effort was spent on obtaining mutual consistency in results for the three methods and on the validation of the methods using data records. As an additional step in the validation process, estimated design floods were verified extensively for mutual consistency between results of 22 locations in the catchment. The project is in its final stages and final design flood estimates have not yet been endorsed by the client so can’t be presented here.