Filtering tides

Frequently Asked Questions about ROMS usage

Moderators: arango, kate, robertson

Post Reply

Filtering tides

#1 Unread post by dobbins »

Hi All

I'm working with Kate Hedstom on a model of the Coastal Gulf of Alaska that includes tidal forcing. Now, I want to analyse the output without tidal aliasing. I'd like to filter the model output with a lopass filter on the fly. Has anyone done this with ROMS? Is it on the developers' priority list?

Thanks, Liz

Elizabeth L. Dobbins

User avatar
Site Admin
Posts: 1295
Joined: Wed Feb 26, 2003 4:41 pm
Location: DMCS, Rutgers University

#2 Unread post by arango »

Well, this depends on the kind of filter. If all what you want is an specific time mean, this can be done in the average file by just averaging over the number of time steps required for a particular tidal period. You just need to compute the number of time steps required and set NAVG
to that value.

Other kind of filters, will required writing a new routine in ROMS and compute then directly in ROMS. Some special switches can be added to ROMS to write that kind of output. This kind filter can be added with added with a CPP option.

Any volunter? I can help with the IO logistics and the parallelization.

Best, H

Hernan G. Arango

Posts: 1
Joined: Tue Jul 01, 2003 12:14 am
Location: Deltares | Delft Hydraulics, Netherlands

#3 Unread post by blaas »

The UCLA & IRD Brest users (e.g., Penven, Marchesiello and me) have a similar case. Up to now we use FFT on the output fields to get the tidal constituents and tidal mean. Since you then need a series that is long enough and of sufficiently high temporal resolution to separate the components, it's a lot of history data you generate. We have been thinking to try to see if one could do a tidal harmonic decomposition or filtering in runtime and output unaliased subtidal averages. However we haven't got the time to build anything yet.

So if someone has made or is interested to take some steps into this we would really like know.


Meinte Blaas | <br>


Tidal filtering

#4 Unread post by hunter »

Hello All,

A useful trick I learned from Robin Pingree many years ago was to slightly adjust the frequencies of the main tidal constituents so that they are periodic in 14 days. The frequencies are (cycles/14 days):

M2: 27

S2: 28

K1: 14

O1: 13

These are all orthogonal over a 14 day period and so are easily separated by a simple Fourier analysis. The forcing is also exactly periodic over 14 days, so that it is easy to test when a tidal model is at steady state.


John Hunter

Post Reply