[Scilab-users] FFT - upward continuation of a 2D grid

[Rafael Guerra]

Hi Lester,

The spatial resolution of 0.016 m seems unrealistically too-finely sampled... Could you confirm the units?
Also, it is possible that your upward continuation requires an angular wavenumber:   exp(-2*%pi*k*uc) ?

Regards,
Rafael

-----Original Message-----
From: users <users-bounces at lists.scilab.org> On Behalf Of arctica1963
Sent: Saturday, July 28, 2018 2:23 PM
To: users at lists.scilab.org
Subject: Re: [Scilab-users] FFT - upward continuation of a 2D grid

Hi Rafael,

Thanks for the pointers. Still having some issues getting it to work on real data. Upward continuation, should act like a low-pass and give a smoother view of the data.

dimx=size(xt);
dimy=size(yt);

nx=dimx(1);
ny=dimy(1);

dxy = 0.016666666;
dkx=1/(nx*dxy);
dky=1/(ny*dxy);

kx=dkx*(-(nx/2)+1:nx/2);
ky=dky*(-(ny/2)+1:ny/2);

[KX,KY]=meshgrid(kx,ky);
k=sqrt(KX.^2 + KY.^2);

uc=4000; // continue up 4000m

//fft_Boug_corr=fft2(Boug_corr);
//shift_f=fftshift(fft_Boug_corr);
//fft_clean=clean(shift_f);

fft_clean=clean(fftshift(fft2(Boug_corr))); // merge commands

Fup=fft_clean.*exp(-k*uc);

F_real=real(ifft(Fup)); 

It does generate output, but certainly not as expected for upward continuation of data. Not sure what I'm missing, assuming I have done the wavenumber thing correct. Normally I would do this via GMT, but want to get it going under Scilab for convenience.

Clearly, before doing a proper fft, we need to pad the data to expand the dimensions over which the fft is done to avoid edge effects. Just running it with the grid dimensions should still produce a recognisable result.

Matbe testing with a synthetic grid would help.

Cheers
Lester



--
Sent from: http://mailinglists.scilab.org/Scilab-users-Mailing-Lists-Archives-f2602246.html
_______________________________________________
users mailing list
users at lists.scilab.org
http://lists.scilab.org/mailman/listinfo/users