Here is an idea that people may find a bit radical but it can get you back on the road. This is just my opinion and lots of people have lots more experience than me. They are welcome to point it out if I am missing something. I am being a bit of a "devil's advocate" here.
First off, check if the harmonics from the global tidal model are accurate at a nearby port. Chances are, they are. If they are, then think about whether including tides in your circulation model is likely to lead to improvement. Chances are, it will not. Chances are, adding the global model harmonics to your circulation model solution
post facto will be more accurate. I personally prefer the latter whenever possible. It is also faster and cleaner once you get your systems in place.
There are trade-offs to either approach. You have already learned about the downsides to adding tides at the ocean boundaries of a coastal circulation model (beginning with hair loss / hair greying). There are limits to the accuracy of all the boundary schemes (Flather etc.) and in my experience there is always some energy reflected. There is also incomplete undestanding of bottom geometry, friction, etc., which are critical to tides due to the relatively high velocity of tidal currents. Nowadays the global tidal models are tuned to both satellite and port data worldwide, and there are no ocean boundaries. It is difficult to improve on them except in small embayments and wide mudflats etc. (and only then if you have really good bathymetry data).
Nowadays I would only add tides to coastal circulation models if I have reason to believe there is a nonlinear interaction between the tidal and wind-driven currents. (I should add that nowadays, I would only run a coastal circulation model if I have a profoundly real reason to do so, because the CMEMS/Mercator system is so good, it is really hard to improve on, and I get a result in seconds as opposed to days or weeks. And by the way you can download a global model forecast with tides already in it! I just prefer to leave the modelling to the experts, and get on with the part that interests me.)
If anyone wishes to contact me directly, please send it to
john.luick@austides.com, as I will see it immediately, rather than in days or months (as in the case of DMs sent to this ROMS forum).
