b'Minerals geophysics A government agency that undertakesvarious flavours of the implementation,gridded data and thus could tempt acquisition of many geophysical surveysare first order errors, second order, or thirdthe unwary processor to use an can also undertake a very thoroughorder errors, when judged from a finiteinappropriately smaller cells size than effort to keep up with the qualitydifference perspective. The actual widththe line spacing could ever justify.control aspects while the data are beingof this operator, each time it is applied collected and before acceptance. needs, at some point, to access originalHigher dimension signal observationsThe private sector, especially in minerals,observations that are being honoured toGradient gridding and tensor gridding is not always as organised, as theconstrain the curvatures to observations.change this processing practice quite a activity of getting a new survey is not asIt is for this reason, as much as anythingbit, as greater than 70% of the measured common an activity in any one company. else, that scalar airborne surveys havesignals can make it into the grid, come to use 4 cells between lines. producing a higher resolution grid of the Post-acquisition A grid contains the summary informationfield. In this case the quality control issue content of a survey, whereas an image isshifts even more critically to the gridding Most of the work done after flying,a reduction of that information using aalgorithm and its ability to honour locally and before final data delivery, involveslook up table colour stretch. The coloursall the trends that have been measured. a correction for systematic errors orand their transitions do not signify aThe technical objective here for instance, adjustments for known factors suchchange of geology or a boundary, butis to produce from magnetic gradient as magnetic compensation, altitudeare an arbitrary assignment so that thetensor data a grid with a cell size that corrections, spike removals, radoneye can better detect the informationis one tenth of the line spacing while corrections, de-stripping, de-corrugationcontent. It is important to note thathonouring the observations. This leads to and cross-over analysis. while honouring the gridding cell sizetwice the resolving power, as a minimum, chosen, most commercially availableto traditional TMI surveys. For instance, a Terrain gridding packages actually display5-metre cell size or less can routinely be In almost all cases, terrain effectson the screen an over interpolated,achieved with this technology. Figure2 are the dominant component of anyaesthetically pleasing version of theshows a model study of a dyke-like newly recorded airborne geophysical survey. It follows that a terrain dataset of appropriate resolution should be available before assessing the quality of the newly recorded survey. Up to 80% of the recorded signal can be directly attributed to terrain effects for gravity gradiometry for instance.GriddingGridding of an acceptable final version of the measured field, after all the conditioning steps and production of a residual anomaly estimate, to reveal how the local geology is influencing the measures, is the principal deliverable to clients. This follows from the reality that the ultimate client is usually an interpreting geologist, not a geophysicist. The very act of gridding immediately rejects, or does not use, around 80% of the acquired data due to the fact that flight line data are aliased in the direction of acquisition and poorly sampled between the lines. Honouring the 20% of the data that is being used, while a representation of the field in the grid is created, is critical. The common practice is to have just 4 cells between lines. As an aid, the physics that the measured field is known to obey can be used to constrain the gridding algorithms, hence the use of minimum curvature and bi-cubic splines for potential field componentsFigure 2.Magnetic Tensor gridding, synthetic data trials. (a) The forward model of a highly folded dyke or scalar measures. It is a limitation ofbody and its magnetic tensor default visual enhancement Cube Root of the Determinant, without any the minimum curvature smoothingnoise. The four flight lines at right angles to its strike are shown. Sample observation is made back to the convolution operator that the 11, 25,flight lines. (b) A tensor grid created, in this case with 12 cells between each pair of lines, recovering the and 49 terms that might be used here incorrect geology object signature. The top edge is unconstrained, so some artefacts are coming in.37 PREVIEW APRIL 2020'