K. Risva, Development of a distributed hydrological software application employing novel velocity-based techniques, MSc thesis, 166 pages, Department of Water Resources and Environmental Engineering – National Technical University of Athens, November 2018.
The aim of this study is the development of an event-based distributed hydrological model, incorporating novel methodologies for estimating the effective rainfall and representing the routing processes. First, we distinguish the effective from the gross rainfall, at a cell basis, thus extracting the spatial distribution of surface runoff during the simulation period. The underlying model is based on an improved NRCS-CN scheme, which uses a spatially-varying CN (different for each cell) and two lumped dimensionless parameters, i.e. one for representing the antecedent soil moisture conditions (AMC) of the basin at the beginning of the storm event, and one for estimating the initial rainfall abstraction. Key modelling novelty is the adjustment of the so-called reference CN value (i.e. the value that refers to average soil moisture conditions and 20% abstraction ratio) against the two aforementioned lumped parameters. For the propagation of runoff to the basin outlet two flow types are considered, i.e. an overland flow across the catchment’s terrain, and a channel flow along the river network. These are synthesized by employing a velocity-based approach, to determine the flood hydrograph. This approach implements an original methodology for assigning realistic velocity values along the river network. These use macroscopic hydraulic information as well as the time of concentration of the basin, which is considered function of runoff intensity. The proposed approach takes advantage of regional relationships and literature values for assigning appropriate values to all model attributes, except for the two lumped parameters of the rainfall-runoff transformation, which are either manually assigned or inferred through calibration. In the last case, it is essential to extract the sub-surface flow component (interflow) from the total hydrograph, which may be done through several approaches of varying complexity. Here we propose an empirical method, requiring the fitting of a lumped hydrological model the observed hydrograph, which explicitly accounts for the contribution of interflow to total runoff. An alternative, more integrated approach, aims at running the distributed model with additional functionalities, in order to obtain the full hydrograph at the basin outlet. In this context, we have also developed a more generic version of the modeling framework, in which the NRCS-CN procedure is combined with a continuous soil moisture accounting scheme, thus generating both the surface (overland) runoff as well as the interflow through the unsaturated zone. Apparently, this augmented version requires few additional parameters, since more processes are accounted for within the simulation procedure. For the schematization of the model domain, the user needs to formulate two spatial layers, i.e. a grid-based partition of the basin to equally-dimensioned (squared) cells, and a graph- based configuration of the hydrographic network, comprising junctions and interconnected river segments. In the context of model development, we used the high-level programming language, Python, to build a GUI interface, for data management and visualization, and to run simulations and optimizations. The two modeling versions and the software are successfully tested in the representation of two flood events across Nedontas river basin.