Modeling the hydrological responses of a lysimeter simulating an experimental green flat roof with Hydrus-1D

In this master's thesis, experimental lysimeter data were used to simulate various configurations of extensive green terrace roofs constructed at the Centre for Building Science of France (CSTB, Nantes) under the research program "VegDUD" (The role of vegetation in sustainable urban development: an approach based on climatology, hydrology, energy management and atmospheric issues) funded by the National Research Institute of France (ANR). These experimental green terrace roof systems were subjected to atmospheric weather conditions of the city of Nantes (France), where their hydrological response to rainfall episodes was measured. The aim of this thesis is to model the hydrological response of the lysimeter with 12 cm growing soil layer and "Sedum album" vegetation by solving the 1-dimensional Richards Equation with the finite element technique. For this purpose, Hydrus 1D software, which can solve the Richards Equation by finite element method in a numerical environment and is available for free license use, was used. In the thesis study, the data recorded in the summer of 2011 in a 12 cm layer with "Sedum album" vegetation on it: instantaneous rainwater amount, water flow rate percolating through the system, meteorological station measurements and TDR sensor data measuring the water content inside the layer were used. The problem was solved using the Penman-Montheith model for potential evapo-transpiration, which is the upper boundary condition of the problem. The van Genuchten-Mualem (vGM) model was preferred for the water retention curve and water hydraulic conductivity functions, which are the hydraulic characteristic functions of the growing layer. For estimation of the parameters of these functions, the soil hydraulic parameters were obtained by using inverse method of infiltration experiments conducted within the scope of VegDUD program. In the first stage, the growing layer was tried to be simulated by following a simple (homogeneous) flow model (single permeability). However, it was observed that there was a significant difference between the actual and modeled amount of runoff water through the lysimeter. In particular, it was observed that the amount of rainwater leaving the system was actually very low. Therefore, in a second stage, a double flow model (double permeability) was chosen. In this case, the modeled "runoff" flow rate was observed to be closer to reality. However, since the double-permeability model requires more parameters, it is necessary to increase the replicability of the infiltration data used for the determination of these parameters and to develop inverse method methods for the determination of soil hydraulic parameters. Furthermore, the preferred 1D dimensional approach may be insufficient to represent the behavior of the 3D system, especially the effect of the slight slope could not be configured. As a result, it was seen that the homogeneous (single permeability) model is not well suitable for modeling the hydrological behavior of green roof systems. Instead, a double permeability model was found to be more appropriate. As a research perspective, it is envisaged that the double permeability model will be validated using more infiltration experiments and improved BEST-2K method to more accurately determine the soil hydraulic parameters and model the flow in a 3D environment. Key words: Double permeability model, evapotranspiration, green roof, Hydrus-1D, Richards Equation