Updates & Feature Requests

The GPU version of the model is very inefficent at running 1D (horizontal) transects - there are simply not enough grid points to make good use of 1,000's of GPU cores. However, the GPU is efficient at running many 100's to thousands of transects at once. In this case, effectively, we are running a 2D simulation, where the x-dimension is the same (the distance along the transect), but the y-dimension is the "transect dimension." So, for example, y=1 is the first transect, y=2 the second, and so on. The physics model is reduced to 1D, so there is no hydrodynamic information being passed along the "transect dimension." This should allow for thousands of 1D transects to be run in near real time, the output of which could then be used to mine statistics, understand parameter sensitivity / uncertainity, etc. with a large number of transect samples. For example, lets say I…

You can now run simulations using the fully nonlinear extended Boussinesq equations. The implementation follows the equation and numerical model presented in Kim et al., 2009, minus the horizontal rotational terms. The equations are the conservative form of the Wei et al. model, also presented in Shi et al, 2012. The numerical scheme is a hybrid finite-volume / finite-difference approach, with leading order terms solved in a similar way as the standard Celeris solver, but using a 4th-order MUSCL-TVD scheme and an HLLEM flux solver, with a 4th order predictor-corrector time stepping integration. The implementation is essentially the finite-volume version of COULWAVE - very similiar to FUNWAVE-TVD. Tests show accuracy similar to these models.

The fully nonlinear / high order model should be considered experimental, and is likely not as stable / robust as the standard Celeris solver, particularly when interacting with the simulation (changing depth on the fly, etc.). The…

You can now add velocity vectors onto the rendered wave field:

In the "Modify Visualization" panel, you will now see three new options related to the vector arrow overlay:

You can choose to include vectors from either the instantaneous velocity field or the time-averaged velocity (current) field (as shown here). You can change the relative length of the arrows with the "scale factor" or increase/decrease the number of arrows displayed with the "density factor." By changing these two factors, you are able to display vector field detail on any scale of interest. For example, if I was interested in seeing the current profile in the immediate vicinity of one of the breakwaters in the above images, I could increase the density to a large value (e.g. 10 to 10x the number of arrows in both the x and y direction), increase the scale factor to larger value as well (e.…

It is a very experimental feature, but you can now add 3D boxes of arbitrary size, orientation, and location into the Explorer (fly-through) view. Any number of boxes can be added to the view. The box properties are contained in a json file (e.g. model.json), and are loaded through the interface - there is a new user input box for this. The json structure is straightforward, for example: {

"models": [

{

"id": "osm_188709102",

"type": "box",

"center": [