Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Civil Engineering

Regional models are commonly constructed such that they reach out to well-defined boundaries and then they are used as reference models for detailed studied for local area. For such detailed study, it becomes essential to use a high resolution numerical model to simulate the local features that may have significant impact on the system dynamics. This high resolution level in groundwater model is not feasible as the regional model area is very large to reach out to well-defined boundaries. So, in regional models coarse grid size will be sufficient; however, local models may require more detailed three dimensional modeling with fine grid size. The main objective of this research is to link the two different discretization models and transfer boundary conditions from the coarse regional model to the fine local one. Two approaches for mapping the regional information over the local domains are developed. The two approaches are the tri-linear interpolation approach (TIA) and artificial neural networks (ANN) approach. Both of them are tested to assess the efficiency of each one in transferring the information from coarse grid model to the fine grid one. The overarching conclusion is that it is better to use ANN technique with generating several scenarios than relying on the TIA in linking the two different discretization models.

Issued also as CD