Calibration of Runoff – Routing Models for the Port Hills Catchments
Stormwater Conference
❄
Christchurch City Council developed the city-wide flood models to assist with response to the city’s increased flood risk following the Canterbury earthquakes. The project developed new hydraulic models for Avon and Heathcote Rivers and Sumner Stream.
The steep, mostly non-urbanised Port Hills catchments of the Heathcote River comprise around 4,000ha of the total Heathcote River catchment area of around 12,300ha, generally with loess soils overlying volcanic rock. The hillside catchments are a significant element of the river hydrology. Historically, the Port Hills response has been heavily influenced by antecedent conditions, with a dry state resulting in large initial losses and a wet state leading to a significant runoff response. Flow records also show an extended tail following rainfall events. This understanding was tested during development of a reliable hydrological model of the system.
Previous hydrological calibration efforts for the hillside catchments were mixed due to the limited available data and the methodologies applied.
The objectives of the city-wide modelling demanded resolution of the urban drainage network, which is predominantly in flat areas of the city. Direct rainfall (rain on mesh) within the hydraulic model was considered to be the most practical hydrological approach to support the level of detail of the hydraulic model, and the rain on mesh methodology was initially considered in detail for the hillside catchments. Initial calibration trials with a wide range of rain on mesh model parameters for the Bowenvale and Hoon Hay Valley gauged catchments showed a poor correlation between recorded and modelled flow for the June 2013 and March 2014 large rainfall events.
A trial of XP-RAFTS and RORB non-linear runoff-routing models was subsequently undertaken using the Bowenvale and Hoon Hay gauged catchment data. The XP-RAFTS and RORB models more successfully predicted stream flow at the monitoring sites than the rain on mesh approach. The models were calibrated using a set of catchment storage parameters common to the two events and the two gauged catchments. The RORB model was adopted in the city-wide project as the preferred methodology for deriving stream flow estimates from the hillside catchments.
The XP-RAFTS and RORB runoff-routing models evaluated in the study are not widely used in New Zealand, but are well established and widely applied in catchment studies Australia-wide. Accordingly, there are limited published regional parameters or methods for parameter estimation available for NZ catchments.
The available flow data at the Bowenvale and Hoon Hay monitoring sites allowed derivation of catchment storage parameters for the RORB model applied in the city-wide modelling project. Good calibration was achieved using those models for the steep, non-urbanised gauged catchments in Christchurch, where calibration of previous hydrologic models has proven challenging.
The use of non-linear runoff-routing models, including XP-RAFTS and RORB models investigated in this study, offer additional tools that may be used to resolve rainfall runoff response in NZ catchments. Further investigation of gauged catchments in NZ would support the understanding of regional relationships for parameter estimation, and may support wider use of the models for estimation of rainfall runoff in ungauged catchments.