Simulation of the upper Waimakariri River catchment by observed rain & radar reflectivity : a dissertation submitted in partial fulfilment of the requirements for the degree of

Abstract

ModClark and Clark’s Unit Hydrograph (Clark’s UH) within HEC-HMS software are distributed and lumped models, respectively. Clark’s UH simulates the transformation and attenuation of excess precipitation, and requires time of concentration (Tc) and Storage Coefficient (R) parameters. ModClark transformation accounts for variations in travel time to catchment outlet from all regions of a catchment, and it additionally requires gridded representation of a catchment and Gridded cell-based input files. Four cases (three from observed rain, and one from radar reflectivity) of three chosen events were specifically chosen and examined for the comparison of simulation results with the same estimated initial parameters apart from different rainfall inputs. The Upper Waimakariri River Catchment was divided into ten subcatchments, and the HEC-HMS basin model parameters were estimated by using the physical/hydrological characteristics. However, ModClark transformation was unavailable because of an output error from converting ASCII to gridded Soil Conservation Service Curve Number (SCS CN) format by the conversion tool – ai2dssgrid.exe. Therefore, Mean Aerial Precipitation (MAP) for each subcatchment was calculated by Thiessen polygon method combined with an overlay analysis for grid-cell-based rainfall estimation from radar with geographic information system (GIS) tools. The automated calibration/optimisation procedure included in HEC-HMS package was applied to the cases which showed a deviation between simulation and observed flows. The purpose is to ‘optimise’ the initial estimates of parameters only in a mathematical-fit manner based on the observed flows from the only discharge gauge at Old Highway Bridge (OHB). The TC values calculated from the five equations vary in a relatively narrow range apart from the one from Bransby-Williams equation. Therefore, the values from all the other four equations were averaged and used as the initial TC input. The simulation results showed that there was a notable difference between observed and simulated hydrographs for some case studies even though TC, R, CN, and lag time were calibrated/optimised separately. Also, radar estimated rainfall and grid-based data storage system (DSS) need more investigations.