Gradient Enhanced Gridding

The Gradient Enhanced Gridding dialog appears if you have selected "Yes" in response to the "Do you have gradient data?" prompt in the Advanced Gridding Options dialog.

Refer to the Application Notes below for additional information on gridding with gradient data.

Gradient Enhanced Gridding dialog options

Transverse gradient data channel	Select the data channel that contains the transverse gradient data. The gradient should be expressed in data units/units of distance measurement in your X,Y coordinate system. For example, gradient magnetic data in a UTM coordinate system must be un units of nT/m. Script Parameter: BIGRID.GRAD
Gradient sensor separation	You can optionally specify an apparent sensor separation that will be used to reconstruct original data from the gradient information. In the case of an airborne magnetometer survey, this would be the wing-tip separation if the sensors are located in the tip of each wing. If the data is not well levelled, a wide sensor separation may introduce noise, counteracting the gradient gridding. In this eventuality, it is recommended to decrease the sensor separation to a fraction of the cell size. If not specified, a separation of 1/20 the grid cell size is used. Script Parameter: BIGRID.GSEP
Gradient direction	"parallel to line (forward)" or " perpendicular to line (right)" This option controls if the gradient data will be corrected for line direction. If you have preprocessed and leveled your gradient data yourself, or if it has been prepared by a survey contractor, it is likely already oriented to be "parallel to line (forward)". You can check this by plotting the gradient profiles on a flight plan map. You may also have gradient data that is simply the difference between the right and left sensors, in which case the gradient will be "perpendicular to line (right)". In this case, the gradient will be corrected for the line direction to be in the gridding direction. The direction of each line is determined from the first and last point in each line, and the along-line gradient is used to calculate the correct gradient in the gridding direction indicated by the TRA parameter. Script Parameter: BIGRID.GCOR: 0 – parallel, 1 - perpendicular
Correct gradient levels	If your gradient data contains "heading" error, you may choose to correct each line by leveling the mean of the measured gradient data to match the gradient calculated from the total field data. Note that if the gradient data is already sufficiently well corrected and compensated, correcting the gradient levels can actually introduce noise because of the possible inaccuracy of the average calculated gradient background levels. This can be more of a problem with relatively short lines. Script Parameter: BIGRID.GLEV
Gradient noise level	Specify the nominal gradient noise level to be used to blend more accurate calculated gradient. Raw measured gradient data can have quite poor resolution at very long wavelengths, which tend to have very low absolute gradients. This can introduce noise in very smooth parts of a survey area. If you specify the noise level of your gradient data, the gradient calculated from the total field will be blended together with the measured gradient to minimise this problem. The blending algorithm is:  G = (Gg + Gm * SN) / (1 + SN)  SN = abs(Gm/N) where  G merged gradient to use in gradient gridding.  Gg gradient calculated from gridded total field.  Gm measured gradient.  SN signal to noise ratio at each point  N measured gradient noise level Script Parameter: BIGRID.GNOISE
Cancel	Return to the main advanced gridding parameters dialog.
OK	Begin gridding.

Application Notes

Gridding without the cross-line gradient will interpolate the smoothest surface that honors the data along each line. Gridding with the cross-line gradient will produce a surface that both honors the data and the measured gradient at each line. This can produce very much-improved gridded surfaces, especially for features that approach the line separation in size.

Figure 1: Gridded magnetic data: a) without cross-line gradient; b) with cross-line gradient enhancement.

The gradient enhanced method implemented in bigrid is the pseudo-line gridding method developed by Hardwick ¹.

Although gridding with gradients will work with any type of data, the principle application is in the preparation of grids from aeromagnetic surveys that use two wing-tip sensors to measure the transverse gradient. The technique used in BIGRID uses both the measured total field data and the gradients to model the magnetic field over the grid area. This requires that both the total field and the gradient data be well leveled using conventional leveling techniques. This method also ensures accurate resolution of low-amplitude, long-wavelength features in the data, which are poorly resolved in the gradient measurements alone.

Following are potential sources of error when working with gradient data.

1. Aircraft orientation error. Because of the orientation of the survey aircraft in flight, gradients may not be measured on a true horizontal plane. However, provided the survey is conducted to minimize aircraft maneuver, this noise is normally minimal.

2. Aircraft control surface error. The simple movement of the aircraft control surfaces – notably the ailerons, can introduce magnetic noise to wing-tip sensors. Surveys should be flown with this in mind, and survey pilots should be more concerned with minimizing the use of ailerons over attempting to maintain a very accurate flight path and elevation. Low-pass filters can be used to remove aileron noise if necessary, though this noise often overlaps signal of interest, so such data will be degraded.

3. Actual horizontal gradients can be very small, often less than the noise level in the measured gradient. If long-wavelength, low-amplitude features are of interest, you should specify the noise level of the gradient data so that this information can be accurately recovered from the total field data.

4. Gradient data that is not properly compensated for aircraft orientation, notably for heading error, can contain base level differences that vary relative to the line direction. In this case, the measured gradient data can be corrected so that the mean matches the mean of the calculated gradient along each line. This should only be done if the data has not been properly compensated as this technique may introduce noise where the calculated gradient is inaccurate.

References

1. C.D.Hardwick, "Gradient-enhanced total field gridding", SEG Technical Program Expanded Abstracts, 1999.