ArchaeoPhysica Ltd on the WWW

Data Treatment (Geophysics)

Basic preparation

All types of geophysical data require some basic preparation prior to analysis & can be expected to require one or more of the processes below. The result at this stage is simply 'clean' field data without distortions introduced by the measurement process & if done properly it contains the same amount of information as originally collected.

Once this preparation is complete processing streams are then defined by the type of data collected. At this stage it is usual to generate a set of images (see presentation page) to provide an initial impression of content. Sometimes surveyors work up an interpretation straight from this image but it is usually preferable to deploy specialist processing as further information is usually available. Examples include conversion to apparent resistivity for electrical resistance data & inversion of resistance profiles, potential field analysis for magnetic data, statistical analysis & reference to absolute values for magnetic susceptibility & stack formation for radar data.

Data from radar surveys tends to pass through a slightly different set of processes (pre-stack) including filtering in time (down a trace) to remove high frequency ringing & along profiles to reduce the amplitude of spurious but ubiquitous reflections from the surveyor. Corrections for topographic variation are applied where necessary & then 3D stacks formed by interpolating & merging the individual profiles. From this timeslices can be extracted & viewed in plan form. Processes based on Fourier transforms cam be applied to model deposits by different dielectric permittivity (as opposed to the reflections from & within them) which can give a clearer idea of some buried structures.

Process Name	Target Data	Process Definition
Reduction of Diurnal	Magnetic & Gravity	Non-gradiometric data needs to have the temporal component removed from it, usually diurnal. Sometimes this is approximated through filters but the best way of achieving this is to use a stationary second instrument (base station) to record the variations. If all the instruments are synchronised in time the data from the base station can be subtracted from the rest.
Destripe / Heading Error Reduction	Magnetic & Electromagnetic	Reduction of a constant-amplitude offset between lines in opposite directions. It is caused by directional sensitivity within the sensor(s) & sometimes the presence of the surveyor. For GPS-tracked survey the process requires a heading rose to be surveyed.
Destagger / Shear Reduction	Magnetic & Electromagnetic	If the surveyor changes direction then their reaction time & the method of sampling in the case of some instruments can cause the data to be shifted slightly along each line. This tends to be common to the whole line & is easily reduced, however, the effects of variations in speed along a line are much more complex to remove.
Despike	All types of data	It is inevitable that measured data will contain isolated point defects & where these are of amplitudes similar to useful data they should be removed. A spike is the deflection of a single datum only, not a dipolar anomaly as is sometimes considered the case. The data can either be removed completely or replaced with an interpolated value.
Geolocate	All types of data	Many geophysical instruments collect data on local co-ordinate systems or simply as lines & samples. These have to adjusted to suite the project co-ordinates, sometimes by simple linear offsets, at other times by more complex geometric transformations.
Rubber Band / Interpolate	Wherever the quantity of data is different along lines or it was not collected along lines	Where data has been sampled in time & allocated times rather than offsets along lines these have to be interpolated to a regular grid for imaging & further processing. For GPS-tracked surveys the interpolation process is 2D to produce a regular grid from the point scatter.

Potential Field Processing

A potential field is any that exerts a force from a point & in geophysics examples are the magnetic & gravitational fields. The physical characteristics of these fields are well known & advanced processing can be applied to great effect, including to data from environmental & archaeological surveys. For magnetic data the processing is primarily applicable to total field measurements & not those from fluxgate gradiometers. The benefits are substantial although not easy to realise due to the complexities of the anomalies from archaeological sources in particular. One primary benefit is that magnetometer data collected on a single horizontal plan can easily be converted into the form apparent from a vertical gradiometer (pseudogradient).

This is useful because it allows magnetic data to be analysed in two parallel ways. In its original form the data contains simple magnetic anomalies without the complication of gradient & anomaly intensity is directly related to susceptibility & depth; for features at the same depth, the primary variable becomes susceptibility. At the same time, there is no loss of resolution of weaker broader anomalies from deep-buried sources & natural variations due to changes in the deeper soil horizons are easily detected & accounted for. In gradient form the data is effectively sensitised to near-surface changes & anomalies that are very close together are in theory better resolved. For archaeological projects these changes are all usually fairly relevant. The advantage therefore is that interpretation can be more thorough. With careful processing it is possible to split the shallow from the deeper component & in this form the data becomes extremely informative.

Raw total magnetic field	Shallow component of total magnetic field	Synthetic total field vertical gradient (1m pseudogradiometer)

Processing for visualisation

Processing for visualisation is intended to produce clearer images by altering the character of the data. Data processed in this way cannot be used for subsequent geophysical analysis though the information gained from the visualisation process is often invaluable when revisiting analysing the original data. There are a large number of algorithms in use, mostly designed for image processing, e.g., box car filters & one or two more specialised processes like the Wallis filter, illustrated below. It is worth noting that it is not always necessary to process data in this way; often good quality algorithms for the initial preparation followed by careful imaging is sufficient to guide analysis & interpretation. Analytical information still needs to be generated through application of relevant geophysical algorithms however.

The Wallis filter below is designed to alter the spread of amplitudes at each location to match a preset variance & thus reduce the effect of local trends that present areas of uniform colour or colour gradient in an image. In this example the fine magnetic structure associated with archaeological features in deep alluvium has been highlighted by suppressing the more dominant form of the palaeochannels.

Shallow component of total magnetic field	After application of a 5m diameter median Wallis filter