Filtering Oceanographic Current Meter Data for Tidal Removal

AI Thread Summary
To filter tidal components from oceanographic current meter data, users are advised to utilize the Fourier transform to identify and remove the tidal frequency spike. Designing a notch filter with a center frequency of 12 Hz is recommended for better results, as simply zeroing out coefficients in an FFT can yield poor filtering. For those without MATLAB, Octave is suggested as a free alternative that can perform similar functions. Users are encouraged to explore filter design basics and consider using Excel or ExcelStat for FFT and spectral analysis. Understanding these filtering techniques is essential for effective data analysis in oceanography.
Glowfish
Messages
2
Reaction score
0
Hi
I am working with current meter data (oceanographic) and have to filter my timeseries to remove the tidal component (frequency 12h). I have not worked with filters before and would appreciate any help.
Thanks
 
Mathematics news on Phys.org
You need to take the Fourier transform of your data; this will give you a representation of your data as a spectrum of frequencies. I suppose there should be a spike at the tidal frequency (twice per day, or 1/12 per hour, or whatever units you use). Then simply erase that spike, and transform your data back into the time domain.
 
Ben Niehoff said:
You need to take the Fourier transform of your data; this will give you a representation of your data as a spectrum of frequencies. I suppose there should be a spike at the tidal frequency (twice per day, or 1/12 per hour, or whatever units you use). Then simply erase that spike, and transform your data back into the time domain.


Better yet, design a notch filter with good characteristics, and a center frequency of 12 Hz. Taking an FFT, zeroing out coefficients, and then inverse transforming gives a terrible filter.

Probably this thread should be moved to EE or something like that?
 
If you have access to MATLAB, it makes filtering datasets quite easy. Let us know what tools you have to work with.

- Warren
 
quadraphonics said:
Better yet, design a notch filter with good characteristics, and a center frequency of 12 Hz. Taking an FFT, zeroing out coefficients, and then inverse transforming gives a terrible filter.

Probably this thread should be moved to EE or something like that?

Think it's a simple typo -- quadraphonics meant (1/12) Hz.
 
berkeman said:
Think it's a simple typo -- quadraphonics meant (1/12) Hz.

Indeed; thanks for that.

A simple highpass filter might also be a good choice, if there isn't any interesting information below 1/12 Hz.
 
berkeman said:
Think it's a simple typo -- quadraphonics meant (1/12) Hz.

Actually, this is also a typo. The frequency is 1 / 12 hours, or 2.31481481 x 10-5 Hz. :biggrin:

- Warren
 
chroot said:
Actually, this is also a typo. The frequency is 1 / 12 hours, or 2.31481481 x 10-5 Hz. :biggrin:

- Warren

Ack! I'd better get back to work... :redface:
 
berkeman said:
Think it's a simple typo -- quadraphonics meant (1/12) Hz.

Okay, I got a lot of advice but still not sure what to do. Unfortunately I don't have matlab. At the moment I'm working with excel and excelstat which allows me to do FFT and spectral analysis.
How do I design a notch filter or a highpass filter in excel with a frequency of 1/12 hr?
Baby steps please.:redface:
Lisa
 
  • #10
If you're not familiar with any kind of filter design, we cannot really help you. It's too sophisticated a problem to walk you through on a forum.

If you'd like, you can probably use the free program Octave to perform your actual filtering. (Octave is a free MATLAB clone.) I'd be happy to design your filter coefficients for you.

- Warren
 
  • #11
It will be great if someone could give a basis lesson on filter. I understand that discrete wavelet transform can also be efficiently implemented using filters.

Thanks chroot for the information on Octave. It's new to me. Do you recommend it? What are the advantages of this software? I have some little knowledge about Matlab. Do Octave have good toolbox, in particular for signal or image processing?
 
  • #12
What are the advantages of this software?
It's a Matlab clone and it's free.
 
  • #13
That's good start. Hopefully there is no copyright violation when using free clone software. Do you recommend any good website that I can download the Octave.

At the moment, however, I'm trying to understand https://www.physicsforums.com/showthread.php?t=32703" first. Not yet on MATLAB programming. Hopeful later I will be able to write program on wavelet transform using filters.
 
Last edited by a moderator:
  • #15
I have already download Octave version 3.0.1. Thanks Science Advisor CRGreathouse.

Have try simple commands and it works :smile: I will figure out later how to run a m-file.
 

Similar threads

Back
Top