Hydrological modelling with weather radar data in urban drainage systems
The management of large scale strategic urban combined drainage systems is becoming increasingly dependent upon weather radar systems which can provide quantitative precipitation information to improve the overall efficiency of a system's operational performance. Thus, there has been an increasing requirement for a more detailed knowledge of the radar rainfall data accuracy and the development of a mathematical rainfall-runoff model that can be used to analyse and control a system in real-time. Within this context, several important factors including signal attenuation, temporal and spatial data resolutions and rainfall quantisation schemes that determine the accuracy of radar rainfall estimates were examined in this thesis. In order to facilitate real-time flow simulation and forecast, a Conceptually Parametrised Transfer Function (CPTF) model has been developed based on Dynamic Linear Reservoir theory. The model is structurally simple and operationally reliable. It can be easily identified and robustly updated following a pulse response-to-CPTF procedure in which Genetic Algorithms play a key role. Using the model, the accuracy of areal rainfall estimates obtained by the Hameldon Hill radar has been assessed, firstly by comparing the radar rainfall estimates with `ground truth', and then by comparing the simulated hydrographs with the actual flow observations. Finally, a case study was conducted using radar rainfall data to highlight the potential benefit of real-time control for the strategic urban drainage system in the Fylde Coast. The major achievements documented in this thesis are: 1) A rule for determination of an appropriate input data resolution for hydrological models; 2) A general probability density function for describing the sampled radar rainfall intensities; 3) An efficient quantising law (ß-Law) and an associated adaptive rainfall quantisation scheme; 4) Three general conceptual pulse-response functions developed based on Dynamic Linear Reservoir theory; 5) CPTF model; and 6) A case study on the potential benefit of real-time control in the Fylde urban drainage system.