Comparison of surface water flow simulation over structured and unstructured grids

Structured grids have drawbacks such as cell distortion and artefacts leading to numerical accuracy reduction during simulations. Spatial support size had proven its importance in various estimation, and this is where unstructured grid comes into action. Performing geostatistical simulation on unstructured grid requires accounting of support size effect. In this article, two methods have been used for accounting the support size effect, one is the classical fine-scale simulation approach and the other approach is using Discrete Gaussian Model (DGM). Each method is applied to generate simulated Digital Elevation Model (DEM) (as used in this study) and the resultant DEMs are studied to understand the effect of spatial support size on surface flow estimation of water. It was observed that due to regularisation in the data of unstructured grid the flow velocity and water surface elevation of the unstructured DEM gave a similar behaviour than the structured DEM. The minimum RMSE for water surface elevation is 0.83 m for unstructured DEM while the minimum RMSE for flow discharge is 0.38 m3/s. The maximum coefficient of determination of flow channel velocity and water surface elevation is 0.709 and 0.86 respectively. The results suggested that the unstructured DEM generated using DGM approach shows a high correlation to reference DEM used in this study than the simulated structured DEM. The surface flow output it was inferred the flow variation in both structured and unstructured DEM is affected not only by the vertical resolution of DEM but also by the horizontal resolution.