You are on page 1of 33

Oasis montajTM 1D-FFT

1 Dimensional Fast Fourier Transform


Filtering and Interpretation System
for Potential Field Data

TUTORIAL and USER GUIDE

www.geosoft.com
The software described in this manual is furnished under license and may
only be used or copied in accordance with the terms of the license.

Manual release date: 7/24/02.

Written by Chris Musselman and Nancy Whitehead. Please send comments


or questions to info@geosoft.com

Copyright Geosoft Inc. 2002. All rights reserved. No part of this


publication may be reproduced, stored in a retrieval system or transmitted, in
any form, or by any means, electronic, mechanical, photo-copying, reading,
or otherwise, without prior consent from Geosoft Inc.

Program Copyright Geosoft Inc. 2002. All rights reserved.

Geosoft and Oasis montaj are registered trademarks of Geosoft Inc.


GEOSOFT, Oasis are trademarks of Geosoft Inc.

Windows, and Windows NT are either registered trademarks or


trademarks of Microsoft Corporation.

Geosoft Incorporated
8th Floor
85 Richmond St. W.
Toronto, Ontario
M5H 2C9
Canada
Tel: (416) 369-0111
Fax: (416) 369-9599

Web Site: www.geosoft.com


E-mail: info@geosoft.com
Contents

Geosoft License Agreement 1

Year 2000 Date Considerations 1

Chapter 1: System Capabilities and Concepts 3

Understanding 1-Dimensional Fast Fourier Transform Filtering 3

Summary of Fourier Filtering Capabilities 4

Introduction to Filtering Concepts 5

Understanding the Mathematical Framework for Transforms 5

Relating Potential Fields to Transform Spectra 5

Understanding the 1D-FFT Filtering Process 9

Finding More Help Information 10

Contacting Technical Support 10

Chapter 2: Quick Start Tutorial 11

Create a Workspace 11

Load the 1D - FFT menu 11

Open a Database 12

Apply Filter to Data Channel 13

Chapter 3: Preparing Data and Applying Filters 15

Preparing Data for Filtering 15

Applying Standard Filters 16

Analytic Signal (FFTAS GX) 16

Apparent Density (FFTDENS GX) 16

Apparent Susceptibility (FFTSUSC GX) 17


Bandpass Filter (FFTBANDP GX) 18

Butterworth Filter (FFTBUTTW GX) 18

Downward Continuation (FFTCONT GX) 19

Cosine Roll-off Filter (FFTCOSN GX) 20

Gaussian Regional/Residual Filter (FFTGAUSS GX) 21

Highpass Filter (FFTHIGHP GX) 21

Horizontal Derivative (FFTHZDRV GX) 22

Horizontal Integration (FFTHZINT GX) 22

Lowpass Filter (FFTLOWP GX) 22

Magnetic Pole Reduction (FFTRPOLE GX) 23

Upward Continuation (FFTCONT GX) 23

Vertical Derivative (FFTVDRV GX) 24

Vertical Integration (FFTVINT) 24

Chapter 4: Applying Custom Filters 25

References 27

Glossary 28
1

Geosoft License Agreement


By using the Program(s), you agree to sign and be bound by the terms of Geosofts License Agreement.
One copy of the signed agreement must be returned to Geosoft Inc. or an authorized representative.
The terms of the agreement are as follows:

Grant of License. This License Agreement allows you to use one copy of the Program(s) for which
you purchased a license, which may include user documentation in printed or online (electronic) form,
on any single computer at any one time. If you have multiple licenses for the Program(s), you may have
as many copies in use as you have licenses. The Program(s) is in use when loaded into Temporary
memory (i.e. random access memory) or installed into the permanent memory (e.g. hard disk or other
storage device). If the anticipated number of users will exceed the number of applicable licenses, you
must ensure that the number of people using the Program(s) does not exceed the number of licenses.

Upgrades. If the version of the Oasis montaj Program(s) you purchased is an upgrade from an Oasis
for DOS Program(s), you may use the Oasis montaj Program(s) only in conjunction with the Oasis for
DOS Program(s) for which you are licensed.

Copyright. The Program(s) is owned by Geosoft Inc. or its suppliers and is protected by Canadian
copyright laws and international treaty provisions. Therefore, you must treat the Program(s) like any
other copyrighted material (e.g. a book) except that you may make one copy of the Program(s) solely
for backup or archival purposes.

You may not rent or lease the Program(s), sub-license or otherwise transfer or encumber the rights and
obligations hereunder without the prior written approval of Geosoft.

You understand and acknowledge that this license is strictly limited to its terms and, in particular,
grants no rights to any other person and is restricted in its application solely for use by you. You may
not transfer, transmit, disclose, modify, translate, adapt, reproduce, decompile, reverse engineer or
disassemble the Program(s) or any documentation without the prior written approval of Geosoft.

You shall indemnify and save harmless Geosoft from and against any and all claims, suits, actions,
costs (including reasonable legal fees and disbursements) or other liabilities to any person whatsoever
arising in any way out of or in respect of any breach by you of your obligations or the restrictions set
out in this License Agreement.

Year 2000 Date Considerations


The Licensed Programs have been tested conform to DISC PD2000 1:1998 Year 2000 Conformity
Requirements (www.bsi.org.uk/disk/year2000/2000.html), with the exception of clause 3.3.2,
paragraph b. Section 3.3.2 paragraph b) requires that inferences for two-digit year dates greater than or
equal to 50 imply 19xx, and those with a value equal to or less than 50 imply 20xx. The Licensed
Programs will recognize all two-digit years as 19xx. This is to prevent errors importing historical data
that pre-dates 1950. All dates that follow 1999 must use four digit dates in the Licensed Programs.
Chapter 1: System Capabilities and Concepts 3

Chapter 1: System Capabilities and Concepts


This User Guide is targeted to Earth Science professionals who are familiar with
geophysical data processing methods and who want to apply Fourier domain linear
filters to process and interpret one-dimensional (line) potential field data from
magnetic and gravity surveys.

To use this package effectively, you should be familiar with Fourier domain filtering
concepts and methods including Fast Fourier Transforms. You should also have an
appreciation of how to apply filters for numerical analysis and interpretation of line
data.

Understanding 1-Dimensional Fast Fourier Transform


Filtering
Fourier filtering techniques are an important tool to geophysicists who work with
potential field (gravity and magnetic) data. Fourier filtering techniques provide an
effective means of processing and interpreting data that is broadband in nature (i.e.
each observation contains the effects from all magnetic or gravity sources that
produce a response above the noise level at the point of observation).

By transforming data to the Fourier domain (through the use of a Fast Fourier
Transform, or FFT), we can deal with the data as a function of wavenumber, or
wavelength. In this form, there are a number of manipulations that can be applied to
enhance information that is of interest, remove information that is not of interest, or
transform the data (see Bhattacharya, 1966, and Spector and Grant, 1970).

For example, in the Fourier domain we might apply a Butterworth low-pass filter to
data to remove short wavelength features that are not of interest. We might also
upward continue or downward continue data to a new level that is more sensitive to
the information of interest, or we might transform magnetic data from an observed
geomagnetic latitude to the magnetic North pole. All of these procedures can be
carried out using Fourier techniques.

The Oasis montaj 1D- FFT Processing System is a suite of Geosoft eXecutables
(GX applications) that you can use to apply Fourier domain filters to one dimensional
(line) data stored in an Oasis montaj database. Designed primarily for potential field
data, Oasis montaj - 1D- FFT provides you with the tools required to:
Perform profile-based processing for interpretation and modeling purposes
Interactively select filtering parameters, display power spectra and identify
optimal filters for processing and interpretation
Rapidly apply filters to one or more lines of data
4 Chapter 1: System Capabilities and Concepts

Summary of Fourier Filtering Capabilities


Generally, you can think of Fourier filters as falling within one of the following
groups:

1. Sharpening filters that enhance the shorter wavelength features in the data. These
include high-pass filters, downward continuations and vertical and horizontal
derivatives. Such filters are normally used to enhance information from shallow
geologic features.

2. Smoothing filters that enhance longer wavelength features in the data, normally by
removing or attenuating the shorter wavelengths features. These include low-pass
filters, upward continuations and integrations. Smoothing filters are normally
used to remove short-wavelength noise in the data or to remove the effects of
shallow geologic features.

3. Geophysical transformations that convert data from one physical form to another.
These include reduction to the pole for magnetic data and the calculation of
apparent magnetic susceptibility or apparent density.

Smoothing and sharpening filters are often combined to meet the needs of a specific
problem. For example, a vertical derivative (sharpening filter that enhances near-
surface geology) might be combined with an upward continuation (smoothing filter
that reduces the effect of noise in the data).

Filters can further be described as geophysical filters -- those whose outcome has a
geophysical basis or mathematical -- those that have a mathematical definition only.
Since geophysical filters are based on the physics of potential fields, they are ideally
suited to gravity and magnetic data whereas mathematical filters can be applied to any
kind of data.

Examples of geophysical filters are vertical continuations, vertical derivative or


integration, reduction to the pole and apparent magnetic susceptibility or density.
Mathematical filters include horizontal derivatives and the high and low pass filters --
- Butterworth, Gaussian, cosine and simple high and low pass cutoff filters.

In addition to filters, power spectra are often required. The Oasis montaj - 1D-FFT
system enables you to create a power spectrum of the data and to place the real and
imaginary components of the Fourier transform into channels of the database. You
may then apply your own mathematical algorithms to the data using the math
functions of Oasis montaj.
Chapter 1: System Capabilities and Concepts 5

Introduction to Filtering Concepts


For mathematical convenience, the Oasis montaj 1D-FFT Processing System applies
filters in the Fourier or wavenumber domain. If you are not already familiar with the
application of filters to one-dimensional data using Fourier Domain techniques, you
may want to read the following summary of filtering concepts.

Understanding the Mathematical Framework for Transforms

Mathematically, the Fourier transform of a space domain function f(x) is defined to


be:

f (w ) = f ( x ) e iwx dx
-

The reciprocal relation is



1
f (x ) =
2p
f (w ) e - dw
iwx

where w is an angular wavenumber in radians per ground_unit (for x in ground_units).


The wavenumber in cycles per ground_unit (r) is simply w /2p.

A line of data in the space domain can be thought of as a sequence of magnetic or


gravity values at points along a straight line, each point separated by a constant
distance. Such a line can be transformed to and from the wavenumber domain by use
of a discrete Fast Fourier Transform (FFT). The equivalent data set in the
wavenumber domain is commonly called a Transform. A Transform of a line is
consists of real and imaginary amplitude values as a function of wavenumber in
cycles per original distance unit.

In addition to these basic mathematical concepts, there are a number of other


generalized properties in Fourier domain processing, including:
A given potential field function in the space domain has a single and unique
wavenumber domain function, and vice-versa.
The addition of two functions (anomalies) in the space domain is equivalent to the
addition of their Transforms.
The energy spectrum is a 1-D function of the energy relative to the wavenumber.

Relating Potential Fields to Transform Spectra

The Fourier transform of the potential field due to a prismatic body has a broad
spectrum whose peak location is a function of the depth to the top and bottom
surfaces and whose amplitude is determined by its density or magnetization. You can
6 Chapter 1: System Capabilities and Concepts

relate the peak wavenumber (w') to the geometry of the body according to the
following expression:

ln( hb ht )
w = where
hb - ht

w' is the peak wavenumber in radians / ground_unit


ht is the depth to the top
hb is the depth to the bottom

For a bottomless prism, the spectrum peaks at the zero wavenumber according to the
expression:

f (w ) = e - hw

where w is the angular wavenumber in radians / ground_unit and h is the depth to the
top of the prism. For more information, refer to Bhatacharia, 1966.

For a prism with top and bottom surfaces, the spectrum is:

f ( w ) = e - htw - e - hbw

where ht and hb are the depths to the top and bottom surfaces respectively. As the
prism bottom moves closer to the observation point at surface, the peak moves to
higher wavenumbers as illustrated in the following figure:
1

no bottom
top = 4

bottom depth
36

20

12

0
0 1
wavenumber

Considering the spectrum of a fixed size prism, as the prism depth increases, the peak
of the spectrum shifts to lower wavenumbers (the space domain anomaly becomes
broader) and the amplitude of the spectrum decreases:
Chapter 1: System Capabilities and Concepts 7

1
thickness = 4

top = 4

16

0
0 1
wavenumber

When looking at the spectrum above, it is important to note that the amplitude of a
deep prism does not exceed the amplitude of the same prism at shallow depth at any
wavenumber. The effect of increasing the depth is to shift the peak to lower
wavenumbers. Because of this characteristic, there is no way to separate the effect of
deep sources from shallow sources of the same type by using wavenumber filters.
You can only distinguish sources if the deep sources have greater amplitude or if the
shallow sources have less depth extent.

When considering a line that is long enough to include many sources, you can use the
log spectrum of this data to determine the depth to the tops of a statistical ensemble of
sources using the relationship (see Spector and Grant, 1970):

log E ( k ) = 4phk

where h is the depth in ground_units and k is the wavenumber in cycles / ground_unit.

You can determine the depth of an 'ensemble' of sources by measuring the slope of the
energy (power) spectrum and dividing by 4p. A typical energy spectrum for magnetic
data may exhibit three parts -- a deep source component, a shallow source component
and a noise component.
8 Chapter 1: System Capabilities and Concepts

The following figure illustrates the interpretation of an energy spectrum into these
three components:
ENERGY SPECTRUM

deep sources
Nyquist

shallow sources

noise

wavenumber (cycles/ground unit)


Chapter 1: System Capabilities and Concepts 9

Understanding the 1D-FFT Filtering Process


When applying Fourier domain filters to a line of data the following steps are
typically applied:
1. Trend Removal A first order trend is calculated by least-square fitting a straight sloping
line to the data. This trend is removed and the trend coefficients are
saved as part of the zero wavenumber component of the transform. This
means that changes to the zero wavenumber in the Transform will also
effect the first order trend.
2. Line Extension The line length is extended by a minimum of 10% up to the next power of
two, which is required by the FFT process used in Oasis. The minimum
10% is to provide a minimum extension within which to predict data that
makes the function periodic.
3. Prediction The extended line and any holes in the line are filled in by predicting data
using maximum entropy prediction (Burg, 1975). Maximum Entropy
Prediction insures that the predicted data has the same spectral character
as the real data. The prediction further insures that the function has the
same value and slope at each end. The line is then considered smoothly
periodic because copies of the line can be place end to end without a
discontinuity.
4. FFT The data is that transformed from the space domain to the wavenumber
domain using an FFT. The wavenumber increment of the resulting
transform will be 1/(line_length). For example, if the extended line
length is 4096 points, and the data fiducial increment is 10 ground_units,
the wavenumber increment in the Fourier domain will be 1/(4096 * 10) or
0.00002441 cycles/ground_unit. The maximum wavenumber will be at
the Nyquist frequency, which is 1/(2*data increment). In the previous
example, this would be 1/20, or 0.05 cycles/ground_unit. The FFTOUT
GX will place the real and imaginary components of the transform into
channels of the database. The FFTPSPEC will calculate the power
spectrum and place the result in a power spectrum channel. The other
filters will carry on to the next step.
5. Filter Application The specified filter is applied to the data in the Fourier domain.
6. Inverse FFT The filtered wavenumber data is transformed back to the space domain
using an inverse FFT.
7. Trend The zero wavenumber amplitude is used to determine the trend to replace
Replacement in the data. Many filters will set the zero wavenumber to 0.0, which
results in no trend being replaced in the data.
8. Masking The original data channel is used as a mask to return the filtered data to
the same length and to place dummies in the final filtered data. The result
is placed in the output channel specified by the GX

The Fourier domain filtering process (as outlined above) is built in to every ID-FFT
filter.
Note: You can use the FFT1D|Advanced Usage menu item to go through the step-
by-step Fourier domain filtering process. This step-by-step method enables
you to apply a user defined filter formula or math expression.
10 Chapter 1: System Capabilities and Concepts

Finding More Help Information


There are several other functions included in the basic Oasis montaj help system that
may be useful to your work. The entire documentation for the system is available
through the online help system. This electronic library of information enables us to
constantly update the information and provide you with the most up-to-date
information available.
The best way to find information in this system is to use the Search tab to perform a
full-text search of all help topics. If you still cant find the information youre looking
for, the Online Books help system contains complete Geosoft manuals and tutorials in
Adobe PDF format.

Contacting Technical Support

The list below provides contact information for Geosoft Technical Support around the
world.
North America Europe and North Africa
Geosoft Inc., Geosoft Europe Ltd.
85 Richmond St. W., 8th Floor 20/21 Market Place, First Floor
Toronto, Ont., Wallingford, Oxfordshire
Canada United Kingdom
M5H 2C9 OX10 OAD
Tel. (416) 369-0111 Tel: 44 1491 835 231
Fax (416) 369-9599 Fax: 44 1491 835 281
Email: tech@geosoft.com Email: tech.eu@geosoft.com
South America Australia and Southeast Asia
Geosoft Latinoamerica Ltda. Geosoft Australia Pty. Ltd.
Av. Rio Branco, 156 - Gr. 2615 350 Hay Street
CEP 20043-900, Centro Subiaco, WA 6008
Rio de Janeiro, RJ, Brasil Australia
Tel: (55-21) 2532-0140 Tel. 61 (8) 9382-1900
Fax: (55-21) 2532-7197 Fax 61 (8) 9382-1911
Email: tech.sa@geosoft.com Email: tech.au@geosoft.com
South and Central Africa
Geosoft Africa Ltd.
Buren Building, Second Floor
Kasteelpark Office Park
c/o Nossob & Jochemus Streets
Erasmuskloof X3, Pretoria
Tel: 27 12 347 4519
Fax: 27 12 347 6936
Email: tech.za@geosoft.com
Chapter 2: Quick Start Tutorial 11

Chapter 2: Quick Start Tutorial


This tutorial is designed to explain how to apply 1D-FFT filters in Oasis montaj .
The procedures are designed to show you how you can apply 1D-FFT utilities and
filters to channel data in a database.
Create a workspace
Load the 1D-FFT menu
Open a database
Apply filter to data channel
These files are provided on the CD-ROM in the Oasismontaj\data\1d_fft directory.
Before you begin the tutorial, you will need to copy the database file (fft_mag.gdb) to
your working directory.

Create a Workspace
In order to access the 1D-FFT menu in Oasis montaj, you must have a workspace
open. The procedures below describe how to do this.

T O C REATE A W ORKSPACE :

1. Start Oasis montaj.


2. On the File menu, click New Workspace. The program displays the New
Workspace dialog box.
3. Specify a name and directory for the workspace. For example, name the
workspace 1D-FFT and specify the working directory as C:\tutorial\1D-FFT.
Oasis montaj will automatically look for your data in the directory containing this
workspace. Make sure you copy the sample data file (fft_mag.gdb) and the
sample grid files (fft_mag.grd and fft_mag.grd.gi) for this tutorial to your
workspace directory (C:\tutorial\1D-FFT).
4. Click [Save]. The system saves the workspace with the corresponding name.

Load the 1D - FFT menu


To begin working with the 1D-FFT system you must load the 1D-FFT menu
(1D_FFT.omn).

T O LOAD A M ENU :

1. On the GX menu, click the Load Menumenu option or the menu button ( ).
The system will display the Load Menu dialog box.
2. Select the 1D_FFT.omn file and click [OK].
12 Chapter 2: Quick Start Tutorial

3. The system will display the FFT1D menu on the main toolbar.
Butterworth Filter FFTBUTTW.GX
Gaussian Filter FFTGAUSS.GX
Cosine Roll-off Filter FFTCOSN.GX
Lowpass Filter FFTLOWP.GX
Highpass Filter FFTHIGHP.GX
Bandpass Filter FFTBANDP.GX
Vertical Derivative FFTVDRV.GX
Vertical Integration FFTVINT.GX
Horizontal Derivative FFTHZDRV.GX
Horizontal Integration FFTHZINT GX
Downward/Upward Continuation FFTCONT.GX
Magnetic Pole Reduction FFTRPOLE.GX
Magnetic Susceptibility FFTSUSC.GX
Density for Gravity Data FFTDENS.GX
Analytic Signal FFTAS.GX
Power Spectrum FFTPSPEC.GX

Preprocessing previewer FFTPREP.GX


Re-fid channel to distance base DISTFID.GX
Transform from space to wavenumber domain. FFTIN.GX
Transform from wavenumber to space domain. FFTOUT.GX
Re-fid channel to fiducial base REFID.GX

Open a Database
You must have a database open in your working directory to apply the 1D-FFT filters
to a data channel.

T O O PEN A D ATABASE :

1. On the Data menu, click Open Database. The system will display the Open
Database dialog box.
2. Select the fft_mag.gdb file and click [Open].
3. The system will display the fft_mag.gdb database.
Chapter 2: Quick Start Tutorial 13

Apply Filter to Data Channel


You must have a database open in your working directory to apply any 1D-FFT filters
to a data channel. In this example, we will apply a Butterworth filter to the mag
channel.

Note that, the entire Fourier domain filtering processes (see Understanding the 1D-
FFT Filtering Process page 9) is built in to every 1D-FFT filter.
Note: All 1D-FFT GXs (utilities and filters) have their own dialog boxes with
unique parameters specific for each GX. To find out detailed information on
each GX. Click the dialog [Help] button.
T O APPLY F ILTER TO D ATA C HANNEL :

1. On the FFT1D menu, click Butterworth. The system will display the FFT
Butterworth Filter dialog box.

2. Select the channel to filter (mag) from the drop-down list. Specify the name of the
output channel (butter_filt). You can specify values for central wavelength
14 Chapter 2: Quick Start Tutorial

cutoff, and degree of filter and select either a low or high pass filter. For more
information on these parameters click the [Help] button.
3. Click the [SetSampling] button. The Set FFT Sampling Parameters dialog box is
displayed.

4. Here you can specify the required distance increment and the interpolation method
to use. For more information on these parameters click the [Help] button.
5. Click [Back] to return the FFT Butterworth Filter menu. Click [OK] to apply the
Butterworth filter and display the output butter_filt channel in the current
database.
Chapter 3: Preparing Data and Applying Filters 15

Chapter 3: Preparing Data and Applying Filters


The following chapter explains how the Oasis montaj 1D-FFT system prepares the
data for filtering and details the application and mathematics of each of the 1D-FFT
filters.

Preparing Data for Filtering


The distance unit used by all filters is the fiducial. This means that if you upward
continue the data, for example, you must specify the continuation distance in
fiducials, or if you calculate a vertical derivative of total magnetic field data, the units
for the result are nT/fiducial.

It is normally more convenient to work with your data in a real distance unit such as
metres (or your ground_units). To do this, you must make sure the fiducial
numbering of the channel you want to filter is based on distance. If your fiducial
numbering is not based on a real distance unit, we recommend that you re-sample
your data to a real distance base before applying filters.

To resample the data to a distance base, select the AdvancedUusage|Re-fid to distance


option in the FFT1D menu. The corresponding GX (DISTFID) looks for a channel
named "_Distance" (containing the along-line distance of each point in the channel).
If the "_Distance channel does not exist, the system creates one based on the X and
Y channels in the database. The GX then resamples the input data channel (using
the distance channel as the new fiducial base) and creates a new data channel.

The DISTFID GX also creates a channel with the same name as the original channel,
but with extension _fid appended. This channel contains the original fiducial
values for the new data channel. If you want to return the filtered channels back to the
original fiducial base for other processing, you will need the original fiducial values.

To return channels to their original values, select the AdvancedUusage|Re-fid back to


fiducial option under the FFT menu and specify the filtered channel to convert back to
a fiducial base. As the output channel, you may wish to specify the same input
channel so that the system simply replaces the channel with the new data. As the
fiducial base channel, you must specify the _fid channel (for example, "mag_fid")
that you created from the original data. Finally, leave the start fiducial blank and
specify the original fiducial increment.
16 Chapter 3: Preparing Data and Applying Filters

Applying Standard Filters


This chapter describes the use of the individual filter GXs included in the Oasis
montaj 1D-FFT system. Each of these filters starts with a line of data in the space
domain and produces a new line of data in the space domain that is the result of
applying the filter. If you would like to apply your own custom filters, refer to the
Applying Custom Filters chapter.

This chapter describes the application and mathematics of each of the Oasis montaj
1D-FFT filters. The mathematical expressions use the following terms:

w Angular wavenumber in radians / ground_unit

k Wavenumber in cycles / ground_unit (w = 2pk )

N Nyquist wavenumber [1/2d where d is the original data increment]

Analytic Signal (FFTAS GX)

This filter calculates the analytic signal of a channel. The analytic signal can be
useful for locating the edges of remanently magnetized bodies and in areas of low
magnetic latitude (Macleod et al., 1994).

The analytic signal (as) of a profile is defined as:

as = dz dz + dx dx

where

dz is the vertical derivative

dx is the horizontal derivative

The vertical derivative is calculated using the FFT process described later in this
chapter. The horizontal derivative is calculated by applying a space domain
convolution filter. The analytic signal is then evaluated from these two sets of data.

Apparent Density (FFTDENS GX)

This filter calculates the apparent density of the ground that would give rise to the
observed gravity profile. The density assumes that the gravity profile is due to a set of
rectangular prisms with a top at the level of observation of the gravity profile, a
bottom at depth t, and infinite strike length.

You may wish to downward continue the profile to be close to the tops of the
assumed geologic model of interest before calculating the apparent density.
Chapter 3: Preparing Data and Applying Filters 17

You must supply the thickness of the earth model and the background density.

w
L( w ) =
2p G(1- e- tw )

Parameters:

t Thickness, in ground_units, of the earth model.


d Background density in g/cm3
G Gravitational constant.

Apparent Susceptibility (FFTSUSC GX)

The susceptibility filter calculates the apparent magnetic susceptibility of the


magnetic sources using the following assumptions:
The IGRF has been removed from the data.
There is no remanent magnetization.
All magnetic response is caused by a collection of vertical prisms of infinite depth
and strike extent.
A susceptibility filter is, in fact, a compound filter that performs a reduction to the
pole, downward continuation to the source depth, correction for the geometric effect
of a vertical prism, and division by the total magnetic field to yield susceptibility.

1
L( w , q ) =
2pF H ( w ) G( q ) K ( w )
H ( w ) = e - hw
G( q ) = (sin I a + i cos I cos( D - q ) ) 2
sin( aw )
K(w ) =
aw
H(w) Downward continuation to h

G(q) Reduction to the pole

K(w) Geometric factor of a vertical prism -- (2a in dimension)

Parameters:

h Depth in ground_units, relative to the observation level at which to calculate


the susceptibility.
18 Chapter 3: Preparing Data and Applying Filters

Ia Pole reduction amplitude inclination. Inclination to which to use the phase


component only in the reduction to the pole. The default is 20. If |Ia | is
specified to be less then |I |, it is set to I.

I Geomagnetic inclination

D Geomagnetic declination

F Total geomagnetic field strength

Bandpass Filter (FFTBANDP GX)

You can use the Bandpass filter to pass or reject a range of wavenumbers from the
data. However, applying such a simple cutoff filter to an energy spectrum almost
invariably introduces a significant amount of ringing (otherwise known as Gibb's
Phenomena). We recommend that you use a smoother filter such as the Butterworth
filter (FFTBUTTW).
1.0

reject pass reject


L(k) 0.5

0.0
k0 k1
Wavenumber (cycles/ground_unit)

L( k ) = 0, for k < k 0
L( k ) = 1, for k 0 k k1
L( k ) = 0, for k > k1

Parameters:

k0 Low wavenumber cutoff in cycles/ground_unit.

k1 High wavenumber cutoff in cycles/ground_unit.

0/1 If 1, pass the defined band. Otherwise, reject the defined band. The default is
to pass the band.

Butterworth Filter (FFTBUTTW GX)

The Butterworth filter is excellent for applying straight forward high-pass and low-
pass filters to data because you can easily control the degree of filter roll-off while
leaving the central wavenumber fixed. If ringing is observed, you can reduce the
Chapter 3: Preparing Data and Applying Filters 19

degree until you are satisfied with the result. A common, but more complicated
alternative is the Cosine filter (FFTCOSN).

1.0
16
4 8
n= 2

L(k) 0.5

0.0
0 kc N
Wavenumber (cycles/ground_unit)

1
L( k ) =
k n
1 + kc

Parameters:

k0 Central wavenumber of the filter.

n Degree of the Butterworth filter function. By default, we assume a degree of


8.

0/1 Residual/Regional flag to specify if a residual high pass or a regional low pass
is required. By default, the system applies a regional filter.

Downward Continuation (FFTCONT GX)

Downward continuation enhances the responses from sources at a shallow depth (by
effectively bringing the plane of measurement closer to the sources). Note, however,
that it is not possible to continue through a potential field source. If the data contains
short wavelength noise, this noise can appear as very shallow sources in the
continuation. Such noise should be removed before attempting to downward continue
the data. A Butterworth low pass filter set to between 1 and 1.5 times the depth can be
very effective to remove noise before continuation.

You should make a plot of the energy spectrum to determine the wavenumber at
which sources (noise) appears to be more shallow than the depth of continuation. The
energy spectrum is also a good guide for determining the depth to which you can
continue data downward.
20 Chapter 3: Preparing Data and Applying Filters

20.0

h = 16
h=8
L(k)
h=4
h=2
1.0
0 N
Wavenumber (cycles/ground_unit)

L( w ) = e hw

Parameter:

h Distance in ground_units, to continue down relative to the plane of


observation.

Cosine Roll-off Filter (FFTCOSN GX)

Because this filter has a smooth shape, and it does not alter the energy spectrum
below the start of roll-off (or after the end of roll-off in high-pass mode), it is
commonly used for simple high-pass or low-pass operations. To reduce ringing, you
can increase the separation between k1 and k0.

1.0

L(k) 0.5
n=2 1 0.5

0.0
0 k0 k1 N
Wavenumber (cycles/ground_unit)

L( k ) = 1, for k < k0
p k - k0
L( k ) = cos n , for k 0 k k1
2 k1 - k0
L( k ) = 0, for k > k1

Parameters:

k0 Low wavenumber starting point of the filter. (Cut-off wavenumber for high-
pass or start of roll off for low-pass.)

k1 High wavenumber end point of the filter. (Start of roll off for high-pass or
cutoff wavenumber for low-pass.)
Chapter 3: Preparing Data and Applying Filters 21

n Degree of the cosine function. The default is a degree of 2 for a cosine


squared roll-off.

0/1 0 for residual (high-pass) filter; 1 for regional (low-pass) filter. The default is
a low-pass filter.

Gaussian Regional/Residual Filter (FFTGAUSS GX)

The Gaussian filter is another smooth filter often used for low-pass or high-pass
applications.

1.0

L(k)
0.5

0.0
0 ko 2ko 3ko N
Wavenumber (cycles/ground_unit)

-( k )
2 2
2 k0
L( k ) = 1 - e

Parameters:

k0 Standard deviation of the Gaussian function in cycles/ground_unit (similar to


a cutoff except that the function amplitude at this point is only 0.39).

0/1 Specify 0 or 1 for residual or regional, respectively.

Highpass Filter (FFTHIGHP GX)

As with the band-pass filter, you should use this filter selectively because it can suffer
from Gibb's Phenomena (ringing).

1.0

reject pass
L(k)
0.5

0.0
ko
Wavenumber (cycles/ground_unit)
22 Chapter 3: Preparing Data and Applying Filters

L( k ) = 0 , for k < k0
L( k ) = 1 , for k 0

Parameter:

k0 Cutoff wavenumber in cycles/ ground_unit. Removes all wavenumbers below


this value.

Horizontal Derivative (FFTHZDRV GX)

You can use the horizontal derivative for identifying geologic boundaries in profile
data.

L( w ) = ( w i )
n

Parameter:

n Order of differentiation.

Horizontal Integration (FFTHZINT GX)

The Gradient Inversion (Horizontal Integration) filter calculates the horizontal


integral of the input transform. This is the inverse of the horizontal derivative. The
zero wavenumber is set to 0.

L(w) = (i w) -1

Lowpass Filter (FFTLOWP GX)

As with the band-pass filter, you should use this filter selectively because it can suffer
from Gibb's Phenomena (ringing).
1.0

pass reject

L(k) 0.5

0.0
k0
Wavenumber (cycles/ground_unit)

L(k) = 1, for k k0
L(k) = 0, for k > k0

Parameter:
Chapter 3: Preparing Data and Applying Filters 23

k0 Cutoff wavenumber in cycles/ground_unit. All wavenumbers above this value


are removed.

Magnetic Pole Reduction (FFTRPOLE GX)

Reduction to the pole is used in low magnetic latitudes to change an anomaly to its
equivalent as would be observed at the north magnetic pole. This transformation
simplifies the interpretation and visualization of anomalies from low magnetic
latitudes.

The reduction to the pole is:

1
L( q ) =
(sin I a + i cos I cos( D - q ) ) 2

where

I Geomagnetic inclination

Ia Inclination for amplitude correction (never less than I)

D Geomagnetic declination

Parameter

Ia Inclination to use for the amplitude correction. Default is 20. (Ia = 20, if I
>0; Ia = (-20), if I <0 ). If |Ia | is specified to be less then |I |, it is set to I.

Reduction to the pole has an amplitude component (the sin(I) term) and a phase
component (the icos(I)cos(D-q) term). When reducing to the pole from equatorial
latitudes, North-South features can blow-up due to the strong amplitude correction
(the sin(I) term) that is applied when D-q is /2 (i.e. a magnetic east-west
wavenumber). By specifying a higher latitude for the amplitude correction alone, this
problem can be reduced or eliminated at the expense of under-correcting the
amplitudes of North- South features.

An amplitude inclination of 90 causes only the phase component to be applied to the


data (no amplitude correction), and a value of zero causes phase and amplitude
corrections to be applied over the entire range.

Upward Continuation (FFTCONT GX)

Upward continuation is considered a clean filter because it produces almost no side


effects that may require the application of other filters or processes to correct.
Because of this, it is often used to remove or minimize the effects of shallow sources
and noise in grids.
24 Chapter 3: Preparing Data and Applying Filters

Also, you can interpret upward continued data numerically and with modeling
programs. This is not the case for many other filter processes.

1.0

h=2
L(w)
h=4
h=8
h = 16

0.0
0 N
Wavenumber (cycles/ground_unit)

L( w ) = e - hw

Parameter:

h Distance in ground_units, to continue up relative to the plane of observation.

Vertical Derivative (FFTVDRV GX)

The vertical derivative is commonly applied to total magnetic field data to enhance
the most shallow geologic sources in the data. As with other filters that enhance the
high-wavenumber components of the spectrum, you must often also apply low-pass
filters to remove high-wavenumber noise.

L( w ) = w n

Parameter:

n Order of differentiation.

Vertical Integration (FFTVINT)

This filter calculates the vertical integral of the input transform. This is the inverse of
the vertical derivative. The zero wavenumber is set to 0.

L(w) = w -1
Chapter 4: Applying Custom Filters 25

Chapter 4: Applying Custom Filters


The Oasis montaj 1D-FFT system also enables you to work with the transform
directly so that you can apply your own custom filters.

The Advanced Usage menu provides users with step-by-step 1D-FFT data processing
sub-menus. For information on the 1D-FFT filtering process see the Understanding
the 1D-FFT Filtering Process page 9. These menu items are designed for advanced
users who want to apply custom filters or math expressions to their data.

Pre-processing pre-viewer FFTPREP.GX


Re-fid channel to distance base DISTFID.GX
Transform from space to wavenumber domain. FFTIN.GX
Transform from wavenumber to space domain. FFTOUT.GX
Re-fid channel to fiducial base REFID.GX

For detailed information on these or any of the 1D-FFT GXs, click the dialog [Help]
buttons.

The FFTIN GX (FFT1D|Advanced Usage|FFT space --> Fourier menu item)


transforms a channel to the Fourier domain and creates a real and imaginary channel
and a wavenumber channel. You can then apply filters with the real, imaginary and
wavenumber channels to create new real and imaginary channels. The FFTOUT GX
(FFT1D|Advanced Usage|FFT Fourier --> Space menu item) transforms the filtered
real and imaginary channels back to the space domain and masks the result against the
original channel.

The channels created by FFTIN have the same name as the original input data channel
with the suffixes _r, _i and _w added. These are the real, imaginary and
wavenumber channels respectively. The wavenumber channel is in units of
radians/fiducial (2p cycles/fiducial). The fiducial numbering of the Fourier domain
channels is in cycles/fiducial.

Note that the FFTIN process does not remove a trend from the data. If your data
contains an undesired trend, you must remove it before running FFTIN. It is also your
responsibility to replace a trend in the data after processing.

The process of applying your own filter is best illustrated through an example. Say
we have an existing channel named Mag which is already sampled on a distance
base (the fiducials are ground_units). We would like to calculate the first vertical
derivative using the expression:

L( w ) = w n
26 Chapter 4: Applying Custom Filters

where n is the order of differentiation (in this case, n=1). To apply this expression we
must multiply the Fourier transform by the wavenumber:
1. On the FFT1D menu, select Advanced Usage|FT space -->Fourier and select
Mag as the channel to transform. This creates channels Mag_r, Mag_i and
Mag_w.
2. Select the entire Mag_r channel in the database (click three times on the
Mag_r channel header).
3. Press the = (equal sign) to enter an expression, and enter:
Mag_r*Mag_w
The Mag_r channel is replaced by the filtered real data.
4. Select the entire database for the Mag_i channel. Press = to enter an
expression, and enter: Mag_i*Mag_w. The Mag_i channel is replaced by the
filtered imaginary data.
5. Select FFT Fourier --> space... and choose the Mag_r and Mag_i as the
input channels, Mag as the original reference channel, and dz as the output
filtered channel. The dz channel will contain the vertical derivative in
nT/ground_units.
References 27

References
Bhattacharya, B. K., 1966, Continuous spectrum of the total magnetic field anomaly
due to a rectangular prismatic body. Geophysics, Vol. 31, p.97-121.

Burg, J. P., 1975, Maximum Entropy Special Analysis. Unpublished doctoral


dissertation. Stanford University. 168p.

Claerbout, J.F., 1988. Fundamentals of Geophysical Data Processing with


Applications to Petroleum Prospecting, Blackwell Scientic Publications, Palo Alto,
California, U.S.A.

Gupta, V. K., and Grant, F. S., 1985, Mineral exploration aspects of gravity and
aeromagnetic survey in Sudbury-Cobalt area, Ontario. SEG; The Utility of Regional
Gravity and Magnetic Anomaly Maps, W. J. Hinze (Editor) p.392-411.

MacLeod, I. N., Vierra, S. and Chaves, A. C., 1993, Analytic signal and reduction -
to-the-pole in the interpretation of total magnetic field at low magnetic latitudes.
Proceedings of the Third International Congress of the Brazilian Geophysical Society,
November, 1993.

McClellan, J. H. and Nawab H., Complex General-N Winograd Fourier Transform


Algorithm (WFTA), Programs for Digital Signal Processing, IEEE Press, pp. 1.7-1 -
1.7-10, 1979.

Spector, A. and Grant, F. S., 1970, Statistical models for interpreting aeromagnetic
data. Geophysics, Vol. 35, No. 2, p.293-302.

Wiener, N., 1949, Extrapolation, interpolation, and smoothing of stationary time


series. Cambridge, M.I.T. Press.

Winograd, S. On Computing the Discrete Fourier Transform, Mathematics of


Computation, Vol. 32, N0.141, pp. 175-199, Jan. 1978.

Claerbout, J.F., 1988. Fundamentals of Geophysical Data Processing with


Applications to Petroleum Prospecting, Blackwell Scientic Publications, Palo Alto,
California, U.S.
28 Glossary

Glossary
Array channels Special channels in columns of the spreadsheet that contain multiple
channels of data. Array channels are represented graphically by a curve in
the spreadsheet. See also channels, sub-array channels.

channels In the Oasis montaj spreadsheet, a channel is essentially a column that


contains a specific type of data.

column In the Oasis montaj spreadsheet, a vertical line of cells that contain data.

database See Oasis database

desktop Background area in the Oasis montaj workspace. You can open and
display a virtually unlimited number of Spreadsheet, Profile and Map
windows in this area.

fiducials Points accepted as fixed bases of reference. Marks indicating the order in
which each reading or sample reading was taken.

Graphical User Interface Interactive software environment where functions are performed by
selecting graphic objects.

grid Collection of points along rows and columns that define a two-dimensional
rectangular area on some plan, usually a ground plan.

Grids or Grid file Files containing location (X and Y) and data (Z) values. Values are
typically interpolated to create a regular and smoothly sampled
representation of the locations and data.

groups A set of graphics elements that make up a graphic component of the map.
For example, a line path plot, a contour plot or a profile plot would all be
separate graphics groups within the Data View.

GX or Geosoft eXecutable Programmed process (identified by the *.GX file extension) used to process
data in Oasis montaj.

Images or Image file Files containing location (X and Y) and color values. The values are not
interpolated. Standard PC file types created using imaging or electronic
photo-editing techniques.

line Linear array of observation points.

Main window Primary tool used to create and maintain databases, display data and
process data. Oasis montaj is a Graphical User Interface (GUI) system that
provides all functionality required to process and display virtually any type
of Earth Science data.

Map (*.MAP) Geosoft-developed file that integrates all graphics elements (lines, polygons
and text) and layers (base maps, data, grids, plots and images) constructed
in Oasis montaj.
References 29

map description file (*.MDF) File used when making a standard map that has a sheet size and margins
inside which there is a map area with a specified origin and scale. Usually,
MDF files are created with the MAPSET GX. MDF files may also be
created manually by editing an ASCII file with extension .MDF.

menu (*.OMN) Graphical list of commands or functions that a user may choose from.

Oasis database Object-oriented database that stores Earth Science data in a form that
delivers fast access to data and efficient storage required for applications
that deal with very large volumes of data.

Oasis montaj Oasis montaj is the core software platform that provides all functionality
required to manage, manipulate, visualize and map spatially located Earth
Science data.

random XYZ data Located XYZ data that cannot be grouped naturally into separate lines. For
example, regional gravity surveys or geochemical surveys are commonly
considered random data because the locations appear somewhat random in
nature.

row In the Oasis montaj spreadsheet, a horizontal line of cells that contain data.

Sub-array channels Individual channels of data from an array channel.

template A file used to manage data in a particular format.

View Stored snapshot of the screen settings, current line, displayed channels and
displayed profiles in Oasis montaj.

Workspace (*.GWS) Binary file that defines a desktop environment. It is essentially an


electronic briefcase that helps organise data and the activities related to a
data processing project.