Historic methods for time series predictions on ungauged sites in the UK have tended to focus on the regionalisation and regression of model parameters against catchment characteristics. Owing to wide variations in catchment characteristics and the (often) poor identification of model parameters, this has resulted in highly uncertain predictions on the ungauged site. However, only very few studies have sought to assess uncertainties in the predicted hydrograph. Methods from the UK Flood Estimation Handbook, that are normally applied for an event design hydrograph, are adopted to choose a pooling group of hydrologically similar gauged catchments to an ungauged application site on the River Tyne. Model simulations are derived for each pooling group catchment with a BETA rainfall-runoff model structure conditioned for the catchment. The BETA rainfall-runoff model simulations are developed using a Monte Carlo approach. For the estimation of uncertainty a modification of the GLUE methodology is applied. Gauging station errors are used to develop limits of acceptability for selecting behavioural model simulations and the final uncertainty limits are obtained with a set of performance thresholds. Prediction limits are derived from a set of calibration and validation simulations for each catchment. Methods are investigated for the carry over of data from the pooled group of models to the ungauged site to develop a weighted model set prediction with pooled prediction limits. Further development of this methodology may offer some interesting approaches for cross-validation of models and further improvements in uncertainty estimation in hydrological regionalisation.

This work is devoted to investigation of the flow interaction between above and below ground drainage systems through gullies. Nowadays frequent flood events reinforce the need for using accurate models to simulate flooding and help urban drainage engineers. A source of uncertainty in these models is the lack of understanding of the complex interactions between the above and below ground drainage systems. The work is divided into two distinct parts. The first one focuses on the development of the solution method. The method is based on the unstructured, two- and three-dimensional finite volume method using the Volume of Fluid (VOF) surface capturing technique. A novel method used to link the 3D and 2D domains is developed in order to reduce the simulation time. The second part concentrates on the validation and implementation of the Computational Fluid Dynamics (CFD) model. The simulation results have been compared against 1:1 scale experimental tests. The agreement between the predictions and the experimental data is found to be satisfactory. The CFD simulation of the different flow configurations for a gully provides a detailed insight into the dynamics of the flow. The computational results provide all the flow details which are inaccessible by present experimental techniques and they are used to prove theoretical assumptions which are important for flood modelling and gully design.

Past studies of overbank flow and sediment deposition have identified the importance of basin scale and reach scale controls. In addition, floodplain topography has been identified as an important control on overbank flow sequences and associated sediment transport and deposition. Recently, the adoption of combined numerical modelling and field-based approaches have provided an effective means of quantifying reach scale processes operating in floodplain environments. However, there is a need to investigate these processes both at the reach and catchment scale. In this study, a twostage procedure has been developed in an attempt to investigate these processes at a range of spatial scales. First, an existing two-dimensional hydraulic model that solves the depth-averaged shallow water equations was used to predict distributed flow depths and velocities over high-resolution topographic grids representing each of the study reaches. These grids were generated from detailed field survey data of floodplain topography collected using a Global Positioning System (GPS) within three study catchments where between six to eight contrasting reaches were chosen to investigate basin and reach scale morphological controls on overbank flows. The results of the hydraulic model indicate three stages in floodplain inundation. First, initial inundation of low-lying areas adjacent to channel margins. Second, a rapid increase in inundation extent via interconnected low-lying areas as threshold discharges are exceeded. Third, complete inundation where floodwater is conveyed as a single unit in a downstream direction. These patterns were also observed using ground and oblique aerial photography of floodwater inundation patterns. The second stage in the procedure utilised a sediment transport and deposition model that was developed using these hydraulic data. This model was used to estimate patterns and rates of floodplain sedimentation at a reach scale and then extended to the catchment scale. Parameter combinations used in this stage were investigated using a Generalised Likelihood Uncertainty Estimation (GLUE) framework and predictions were compared with estimates of medium-term sedimentation rates derived from 137Cs analysis of floodplain sediment cores. The results of the sediment transport and deposition model indicate that within reach variability of floodplain sedimentation is influenced by small-scale local topographic controls. Typically, 15 % of the inundated floodplain area receives ~50% of the total sedimentation amount and reflects the low-lying areas adjacent to the channel. In contrast, 40% of the inundated floodplain area receives < 20% of the total sedimentation, reflecting distal floodplain areas that are inundated for short periods of time.

Most approaches for modelling flood inundation and depth in lowland settings involve laborious parameterisation of stream channel and floodplain hydraulic characteristics and most require intensive inputs to set up and maintain. Such an approach is unrealistic in Bangladesh with its highly complex and dynamic hydrological network and sparse data collection system. This needs-driven research develops techniques that could be used in an operational, national system for monitoring and assessing floods and for monitoring riverbank erosion and dynamics of river channel morphology. Satellite synthetic aperture radar (SAR) images are used in a consistent, multitemporal data set for the 1998 monsoon flood season. The SAR images are processed and analysed with digital elevation models (DEM) and other spatial data for monitoring floodplain water levels. Compared water levels for the 24-day image acquisition cycle compare with field data by average absolute value difference of 33 cm and with regression coefficient of 0.86. Daily floodwater elevations are simulated through regression relation of the floodplain with conventional river gauge recordings. Flood depth is derived by difference with a DEM and used to create a series of daily flood depth maps that compare with field data by less than 40 cm absolute value difference. Because of the unusually high water levels during the 1998 flood, it is possible that there may have been an unusually direct hydrologic connectivity between the river and the floodplain. Such a relation has yet to be extended to other flood seasons or to other floodplain areas. A series of spatial data products demonstrates the uses of these daily flood data for flood risk and vulnerability mapping, flood damage assessment, river erosion monitoring, and other applications. Before the methods can become operational, data processing, modelling, and analysis will need to become more quantitative and more efficient. Furthermore, institutional linkages and multi-stakeholder co-operation and involvement will need to be ensured. Bangladesh lacks the financial resources for implementing and maintaining an operational system, however, it has the technical capability and the commitment of development partners.

This research evaluated the application of artificial neural networks and hydrodynamic models for river flood modelling and prediction. The study has been completed in three main parts: First part focused on the application of artificial neural networks for flood prediction in ungauged catchments. Catchment descriptors were used as input data and the index flood was the output of the model. Different types and numbers of catchment descriptors (17 descriptors and more than 1000 catchments) were used to choose those that gave the best relationship with the hydrological behaviour and flood magnitude. ANN models with different architectures were developed and applied to training and validation sets of data to find the best type of ANN for this application. Selection of pooling groups of catchments either randomly or according to geographical proximity did not produce desirable results. Therefore hydrologically similar catchments were clustered using the FEH-Software before entering descriptors into the ANN model. This improved the accuracy of predicted floods. The second part of the research aimed to model river flow in a multi-gauging station catchment and provide real-time prediction of peak flow downstream. Three types of ANN (Multi-Layer Perception (MLP), Recurrent and Time Lag Recurrent) were adapted to evaluate the applicability of this technique. The study area covers the Upper Derwent River, a tributary of the River Trent in the UK. River flow was predicted at the subject site with lead times of 3, 6, 9 and 12 hours. Tests were completed using different lengths of input data to evaluate the effect of input data size in model outputs. The number of gauging sites to be used as data sources in the model as also evaluated. In the final part, the application of artificial neural networks (ANN) to optimise the results obtained from a hydrodynamic model of river flow as evaluated. The study area is Reynolds Creek Experimental Watershed in southwest Idaho, USA.

This thesis describes the implementation of general Computational Fluid Dynamics (CFD) techniques to laboratory and natural channels under flood flow conditions. Two commercially available codes, TELEMAC and CFX4, have been used in this work. The assessment of CFD for the calculation of flooded channel flow dynamics is carried out by simulating one laboratory test case from the Flood Channel Facility (FCF) Series B. This test case is that of a meandering two-stage channel with a depth ratio of 25% on the flood plain. Results from a computer simulation of experiment B23 are presented with a detailed quantitative comparison of the measured velocity, turbulence and bed shear stress. It supports the conclusion that CFD is able to account for the different flow mechanisms arising from the interaction between inbank and overbank flows in meandering channels. The maximum error in the prediction of the velocity is 10% and the comparisons show the calculations of bed shear stress to be reasonably accurate as well. Numerical tests indicate that the numerical solution is relatively independent of the boundary conditions, and confirm that turbulence transport is of minor importance in the experiment simulated. Numerical results from the simulation of flood flow mechanisms in natural rivers are also presented. It is hoped that these are of value to practitioners. Two 1-km reaches on the River Severn and River Ribble are modelled. They permit the investigation of two-stage channel flow dynamics at a larger scale. The numerical verification process establishes that the scale and the complex nature of the geometry are limiting factors, particularly for the numerical discretization of the domain and the calculation of the variables at the walls. It is however possible to estimate a priori part of the error such constraints generate. Away from the walls, the flow main features seem well predicted. The parallel between the velocity fields observed in river flood flows and those observed in the FCF is evident. Validation against field data suggests that the models are able to reproduce the flow mechanisms and account for bed shear stress variations correctly. Yet a significant level of uncertainty remains when the model predictions are compared against measured point data; more validation work is therefore required.