Water Under the Bridge: Flood Modelling of the Ngarara Stream, Waikanae

Abstract

The Ngarara and Waimeha Streams lie just north of the Waikanae River on the Kapiti Coast, approximately 50km from Wellington. Just downstream of their confluence is the Field Way Bridge. The streams are included in the Greater Wellington Regional Council MIKE 11 hydraulic model for the area, which has indicated that the bridge is under-capacity during high flow events. Greater Wellington has identified the need to refine the ungauged Ngarara branch of the Waimeha-Ngarara MIKE 11 model as a priority. This will enable better informed decisions regarding flood management, and expensive engineering works on the Field Way Bridge. The hydrological inputs to the MIKE 11 model were considered the area of greatest uncertainty. The incorporation of sub-catchment-level design hydrographs are a key model refinement. A series of 100-year ARI design hydrographs was synthesised using the HEC-HMS implementation of the SCS rainfall loss model and Clark unit hydrograph. The process was documented to allow evaluation of the resulting hydrographs. Analysis of the effect of HEC-HMS rainfall inputs showed that the 'balanced storm approach', significantly overestimates design flows. The development of synthetic hydrographs for the Ngarara catchment presented a number of challenges. These included accurate catchment delineation, determination of suitable hydrological parameters and rainfall inputs, and obtaining check figures for the design hydrographs. Some of these challenges are common to all ungauged catchments. Others are specifically related to low-lying, low relief, sub-catchments on the coastal plain, which also have permeable soils. ESRI GIS functionality was used as the general support tool for the study. It was used to create a distributed rainfall surface, necessary for the allocation of design rainfalls, and to automatically delineate the sub-catchments. This latter process allowed evaluation of the efficiency and effectiveness of ESRI automated catchment delineation tools when applied to high-resolution LIDAR data-based DEMs, of low-relief topography. The choice of design rainfalls, and flow check figures, was based on comprehensive hydrological analyses of all available rainfall and flow records. These were combined with the results of regional analyses from other studies. Existing regional design rainfalls overestimate inputs to the Ngarara catchment. Design rainfalls were therefore assigned from the Waikanae Water Treatment Plant hydrometric record. Analysis of the flow record for the nearby Mangaone Stream showed that previous flood frequency analysis was incorrect. Continued monitoring and re-evaluation of current design figures is therefore essential. Mangaone rainfall and flow data were used to identify relationships between flow magnitude and SCS Curve Number; and flow magnitude and runoff coefficients. A correlation was found between event magnitude and runoff coefficients, but any calibration of the Curve Number was inconclusive. As in previous studies, both analyses displayed a high degree of scatter. Despite the best efforts to assign suitable HEC-HMS parameters, the values used were still largely empirical, and uncalibrated for the Kapiti area. Local flow records were unavailable so the synthesised design hydrographs are also uncalibrated. The selection of HEC-HMS parameters involved considerable subjectivity. To confirm the actual hydrological response of the Ngarara catchment matches that synthesised, flow monitoring must be carried out at strategic catchment locations. In the absence of these data any future development of the Greater Wellington MIKE 11 model will be compromised by inaccurate hydrological inputs.