Bathymetry smoothing using adaptive and spatially variable LP method

[jivica]

Dear ROMS users,
I took me some time and effort to updated our old linear programming (LP) smoothing method into modern Python.
It doesn't depend anymore on lp_solve linux library, slow writing of multiple tmp files on disk, no Matlab and it is even faster ~1-2 orders of magnitude (in my test cases ~40x). It can use multiple threads if available, is quite easy and straightforward to install and only depends on 3 standard Python libraries: Numpy, netCDF4, Scipy/HiGHS (optionally if you want to plot results you'll need Matplotlib). There are working 2 examples on how to use it. In normal case it loads hraw from the grid file you specify. Some useful features are, for example, you can apply spatially variable smoothing i.e. more strict rx0 values in deep regions and higher values in shallow, smoother values around the boundary - sponge regions - where you melt with parent model with minimal intervention.

https://github.com/ivicajan/roms-bathy-smooth

Let me know if any problems or you find it useful.

Regards,
Ivica

[arango]

Thank you, Ivica, for this essential contribution. Many users ignore this important aspect when building their ROMS grid application!

[jivica]

I just published the package on PyPi so it is even easier to install

All you need is a simple command:

Code: Select all

 pip install roms-bathy-smooth 

and there you go.


Cheers,
Ivica

[jpringle]

Thank you so much for doing this; getting the old package to work (and showing students how to do it) was getting increasingly cumbersome!

Thanks
Jamie Pringle