Grid Generation

A variety of orthogonal grids may be generated using the integration of several software packages. These grids may be rectangular, geographic rectangular (a lat/long grid on the spheroid, centered on the latitude the grid is created) or false-pole geographic rectangular (a lat/long grid on the spheroid , centered on the equator). Polar grids may also be created, where grid size increases with distance from the origin. These types of grids may be constructed within PLUM (see Visualisation for a description of PLUM). Finally general orthogonal curvilinear grids may be created using dedicated software which is accessed through a graphical user interface within PLUM, allowing the user to easily define the shape of the curvilinear grid with simple point, click and drag operations. The user interface is illustrated in Figure 1; the blue dots may be added or deleted to define the shape of the grid. Red dots denote the corners of the grid. In general, practice is required to create the required resolution in the required location for these general curvilinear grids. Internal or external branches may be appended to the core curvilinear grid, to allow for geographies such as a winding river entering a bay. An example of such is grid is shown in Figure 2; this example uses a 2-D laterally averaged grid to represent the upper reaches of the main river, so that the cross-river dimension is eliminated to result in a less restrictive model time-step. Additionally, the curvilinear grid can be used in a coastline-following capacity, as in the case in the upper river. The branching used in these grids typically renders much of the grid as undefined space (i.e. locations that cannot be attributed a geographic lat / long position), and it is not uncommon for the wet cells in these grids to number less than 20% of the total grid size. This makes the use of the sparse coordinate system in SHOC attractive, since these undefined cells are eliminated from everything (memory and computation). In all these grid generation routines, land is identified using a coastline mask, and bathymetry may be interpolated onto the grid from user supplied databases (netCDF or ascii) using a variety of schemes (averaging, bilinear interpolation, inverse distance, weighted area, Sibsonian and non-Sibsonian natural neighbours). PLUM has grid editing capability to define land cells or water cells, identify open boundaries, create cyclic grids and add, delete or manually adjust grid lines. The masking menu in PLUM is shown in Figure 3. Bathymetry may also be edited. Once a grid is generated, it may be exported in a format compatible with input into the hydrodynamic model. Additional functionality is also available, for example the creating of arbitrary partitions for multi-processing, where grid statistics are supplied to optimise load balance and information transfer volumes. The grid generation routines available in EMS allow rapid production of grids, ultimately decreasing the time and effort required to produce pilot models.


Figure 1 : PLUM grid generation interface.


Figure 2 : Example of a complex curvilinear grid.



Figure 3 : Masking grid generation menu.


Legal Notice and Disclaimer | Copyright | Privacy