Geophysical Survey Methods
The methods most often employed by Archaeo-Physics are electrical resistance, magnetic field gradient, and ground penetrating radar (GPR) survey. These methods have proven to be the most useful tools for archaeological prospection in most situations. We also have a variety of other methods which we may employ under special circumstances, including: EM conductivity survey; "total field" magnetometry; soil compaction survey; and laboratory testing of field samples.
These methods provide excellent resolution of many types of archaeological features, and are capable high sample density surveys of very large areas and of operating under a wide range of conditions. When two or more methods are used in combination the different instruments provide complimentary data sets, which can give greater insight into feature composition than could be gained from a single method.
A brief introduction to our principal methods is presented here as an introduction to archaeologists. An in-depth study of these and other methods as applied to archaeology can be found in Clark (1996).
A GPS-controlled magnetometer survey
The general procedure followed to perform most surveys is to divide the survey area into a series of square or rectangular survey "grids." Each grid is surveyed by taking readings at regular intervals along regularly spaced transects. Ropes marked at regular intervals are used to control transect spacing and position along each transect. Successive transects are surveyed in a zigzag pattern until the grid is completed. The value and position of each data point is automatically recorded in digital format, and is later downloaded to a portable computer. For some large-scale reconnaissance surveys, spatial control may be provided by an integral GPS. Occasionally, these instruments are also used to record selected individual transects or for less formally "scanning" areas of interest
The general procedure followed to perform most surveys is to divide the survey area into a series of square or rectangular survey "grids." Each grid is surveyed by taking readings at regular intervals along regularly spaced transects. Ropes marked at regular intervals are used to control transect spacing and position along each transect. Successive transects are surveyed in a zigzag pattern until the grid is completed. The value and position of each data point is automatically recorded in digital format, and is later downloaded to a portable computer. For some large-scale reconnaissance surveys, spatial control may be provided by an integral GPS. Occasionally, these instruments are also used to record selected individual transects or for less formally "scanning" areas of interest.
Archaeo-Physics uses a variety of instrumetation, and considers both research questions and the physical context of the survey in instrument selection and survey design. While we may use a number of other instruments for special applications, resistance, magnetic, and radar methods are generally the most versatile and useful. Most of our surveys involve at least one of these instruments, and these three are discussed in greater detail below.
Electrical resistance survey using ropes
for spatial control
Electrical resistance (also called resistivity) surveys introduce an electrical current into the soil and measure the ease (or difficulty) with which this current flows within the soil. Variance in measured resistance values across a site can be interpreted as variance in the relative resistivity of materials composing the matrix in the vicinity of each reading. Resistance surveys respond to a combination of soil moisture, soluble ion concentration and physical soil type. Moist soils have lower resistivity than dry soils. Fine soils (clay) have lower resistivity than coarse soils (sands or gravels), and high salinity soils have low resistivity.
At a more fundamental level, the resistivity is governed by the number and mobility of free charge carriers available in the soil. The principal sources of these free charge carriers are soluble ions. Thus, the simultaneous availability of soil moisture and soluble salts determines the free charge carrier concentration in the soil. The mobility of these carriers is also an important parameter in soil resistivity. The mobility of the soluble ions is governed by soil moisture content, soil grain size, temperature, soil compaction, as well as the surface chemistry of the soil grains. These variables govern soil resistivity at the low frequency used in these surveys. At higher frequencies soil resistivity becomes a more complex issue.
Twin electrode surveys respond to the soil resistivity in the immediate vicinity of the sampling probes. The depth of the response is roughly proportional to a semi-spherical volume with a radius equal to the spacing of the current and potential probes. The actual radius of "response" is a complex function of feature contrast, feature size, feature geometry, and depth below the surface. In general, a greater probe spacing will result in increased depth of penetration, at the expense of resolution of small, low-contrast features.
Archaeologically useful surveys result when the resistivity contrast between the archaeological record and the surrounding soil matrix is great enough to be detected. The recorded data are an average made up of contributions from the surrounding soil matrix and the archaeological record. Therefore, it is clear that for detection, the contribution from the archaeological components to the measured average must be greater than the statistical uncertainty in the survey data. Appropriate survey design (instrument selection, instrument configuration, data sample density, and field methods) is necessary for a successful survey.
Site geology is seldom uniform and the spatial variability of resistance data associated with the geology will also be present in the survey data. Geology can usually be distinguished in the resistance data by its scale and geometry.
Electrical Resistance Data Processing
Resistance survey can map a great variety of features. Shown here are adobe walls of the Spanish Mission at San Marcos Pueblo, New Mexico
Resistance survey data are processed with Geoplot software, which is provided by the manufacturer of the survey instruments. The data quality is usually excellent and very little "clean up" processing is required. Extreme outliers are typically removed and the data are clipped to the appropriate range by examining the data statistics and histogram characteristics. The data are also interpolated (up or down) to a data sample density that is uniform in the X and Y directions for viewing and export to graphic software.
A convolutional highpass filter is often used to suppress large-scale geologic variation and to enhance small low-contrast anomalies. It is implemented by calculating the local mean in a moving window and subtracting the mean from the original map data. The size of the moving window is adjustable. It is typically set in the range of 5 - 10 meter radius. The result is a new map in which the average "background" resistance of the site has been subtracted. The mean value of the new map is zero.
Highpass filtering of resistance data offers a number of benefits to the mapping and interpretation of resistance data. In addition to enhancing the visibility of small low-contrast features, it also creates a resistance map with zero mean. This zero mean map can be thought of as a resistance map containing features which are greater than the local average (the positive values) and features which are less than the local average resistance (the negative values). The zero data regions correspond to areas of no resistance deviation from "background" or local mean.
With this insight, it is convenient to interpret positive data as features with "greater than average resistance." For example, stone architecture or pits filled with sand or gravel might appear as high-resistance anomalies. In like manner it is convenient to interpret negative data as features with "less than average resistance." Pits and trenches containing organically enriched fill, clays, and high salinity soils might appear as low-resistance anomalies.
Magnetic survey with the EM36 fluxgate gradiometer
Magnetic surveys - collectively referred to as "magnetometry" - may use different configurations of magnetometers. Most commonly used in archaeology is a gradiometer configuration, which measures the vertical gradient of the Earth's magnetic field. For the purpose of this type of survey and in the absence of archaeological and geological contributions, the earth's magnetic field near the surface of the earth is uniform and the gradient of this field is zero. When there is an archaeological or geological magnetic field, it adds to the earth's magnetic field and the magnetic field gradient is no longer zero.
Magnetic field gradient surveys measure this deviation from uniformity and report it as positive data when the deviation is in the direction of the earth's magnetic field and as negative data when the deviation is in the direction opposite the earth's magnetic field. In these surveys, the more "magnetic" the archaeological record the greater the magnetic field distortion and the greater the feature contrast in the survey map.
The archaeological record has two basic properties or mechanisms by which it distorts the earth's magnetic field. These are called the remanent magnetization (a permanent magnetic effect) and the magnetic susceptibility (a bulk magnetic property similar to density). Both mechanisms are dependent on the presence of iron (e.g., iron oxides in soils, sherds, and hearths) and both mechanisms alter the magnetic field at the surface of the site. They are thus mapped as distortions of the earth's magnetic field.
Remanent magnetization is the familiar "permanent magnet" effect and is associated with iron and steel objects (including rust) as well as with ceramics, hearths, fire pits, fire-altered soils and stone. In these materials, the remanent magnetization originates from heating the iron oxides found in most soils above a critical temperature (565 to 675 degrees C), called the Curie temperature. When the soil cools, the temperature-induced changes in the iron oxide crystals are "frozen" and become permanent. It is this change in the magnetic state of the soil (ceramic, hearth, etc.) which creates a remanent magnetic field. This thermally created magnetic field adds vectorially to the earth's magnetic field to cause a local distortion. Thus, most cultural objects and processes associated with heating are potential archaeo-magnetic survey objects of interest.
The magnetic susceptibility alters the earth's magnetic field directly in a manner roughly analogous to the way porosity alters the flow of water through a solid. That is, where the magnetic susceptibility is large (high porosity), the magnetic field is increased and where the magnetic susceptibility is low (low porosity), the magnetic field is decreased. Many cultural objects and processes (thermal, chemical, biological and biochemical, physical and mechanical) locally increase the magnetic susceptibility of the soil. The mechanism for this increase is also associated with changes in the iron oxide crystal structures within the soils. Local changes in site magnetic susceptibility alter the earth's magnetic field and it is this distortion which is mapped. In magnetic surveys, remanent magnetization (permanent magnet) effects are usually somewhat greater than susceptibility effects.
Many magnetic highs are a combination of induced field and remanent magnetization. The observed magnetic field strength is the result of the total magnetization of an object. The total magnetization is a vector sum of the induced magnetization and the remanent magnetization (Sharma 1997).
Magnetic Field Gradient Data Processing
All magnetic field gradient data are processed with Geoplot software, which is provided by the manufacturer of the survey instruments. Typically the data are "cleaned up" using a "Zero Mean Traverse" algorithm which removes scan to scan instrument and operator bias defects. A gaussian lowpass filter is used to remove high-frequency spatial detail, or smooth the data. The data are also interpolated (up or down) to a data sample density that is uniform in the X and Y directions for viewing and export to graphic software.
As with highpass filtered resistance data, magnetic field gradient data are also a zero mean bipolar data set. Magnetic field gradient maps can be thought of as containing features which increase the field gradient by locally adding to the earth's field and features which decrease the field gradient by locally subtracting from the earth's magnetic field. The zero data regions correspond to areas of uniform or undisturbed magnetic field.
Thus, all positive data can be interpreted as features with increased magnetic field due to increased susceptibility or remanent magnetization oriented in the same direction as the earth's magnetic field (e.g., hearths, fire-altered soils, bricks, sherds, and iron). All negative data can be interpreted as features with decreased magnetic field due to decreased susceptibility or remanent magnetization oriented in the direction opposite the earth's magnetic field (e.g. bricks, sherds, and iron).
Ground penetrating radar survey through a limestone temple pavement (Mahram Bilquis case study)
The GPR functions by sending high-frequency electromagnetic waves into the ground from a transmitter antenna. Some of these waves are reflected back to the surface as they encounter changes in the dielectric permittivity of the matrix through which they are traveling and are detected by a receiver antenna. The amplitude and two-way travel time of these reflections is recorded on a portable computer. This information is then used to construct a two-dimensional plot of horizontal distance versus travel time. Data collected in the field are stored on a portable computer for later analysis. A more complete and technical discussion of the method can be found elsewhere (Annan and Cosway 1992; Conyers and Goodman 1997).
The effectiveness of GPR is controlled by the local soil conditions. GPR is most effective in locating buried objects in a homogenous soil matrix with a high electrical resistivity. GPR is least effective in a heterogeneous environment with high electrical conductivity. A heterogeneous environment contributes to signal scattering and can result in insufficient depth of penetration and a "noisy" reflection (poor signal-to-noise ratio). A conductive environment can seriously inhibit depth of penetration due to conductive losses. Conductive loss is the result of the electromagnetic wave creating a conductive current in the soil medium. This conductive current loses energy in the form of heat and can also set up what is often referred to as "eddy currents." Eddy currents are secondary electromagnetic waves created by the conductive currents in the soil. These waves can obscure reflections of interest with strong horizontal banding, a phenomenon known as "ringing."
Although GPR survey can be performed in a number of ways, the method we generally employ involves dragging the transmitter and receiver antennas together over the ground at a fixed rate, called fixed offset reflection mode. The transmitter emits pulses at regular intervals along a transect which are picked up by the receiver. A laptop computer controls data collection and displays the data as a two dimensional profile.
The GPR is able to detect subsurface features whose electrical properties contrast with those of the surrounding soil. The GPR can detect archaeological features in several ways. It may detect disturbed soil, breaks in the natural stratigraphy or soil profile, or reflections from buried archaeological features.
Ground Penetrating Radar Data Processing
GPR data can be plotted as vertical profiles or horizontal planviews of specific depths. It can also be used for 3D modeling, as in the animation above, showing a 19th century tomb.
The analysis of GPR data is carried out by processing the data using different gains and filtering techniques. Gain is a value by which raw data are multiplied to enhance low-amplitude reflections. Signal amplitude commonly decreases exponentially at increasing travel times (greater depth below surface). This was compensated for by designing a custom time gain that increased the signal strength to a constant average value. Filtering is the use of mathematical processing algorithms to "clean" noise from the data and/or enhance certain characteristics of the data. Among the filtering techniques which can be applied are: Bandpass frequency (noise reduction), temporal median (noise reduction), spatial lowpass (noise reduction and continuity of horizontal events), spatial highpass (background removal), and deWOW (removal of very low-frequency inductive phenomena).
GPR data is most often plotted as single transects, which appear as vertical profiles. A technique known as time slicing makes it possible to construct planview maps of an area which has been surveyed with multiple adjacent transects. This not only makes interpretation of the data in the horizontal plane much more intuitive, but also allows us to isolate specific depths (or more properly, the two-way travel times of reflected waves) for examination. Recent developments in GPR data processing and imaging software has enabled plan view presentation GPR data as average enveloped amplitude time slices and as statistical activity analysis within a time window. Average amplitude time slicing methods and case studies are presented in some detail by Conyers and Goodman (1997). Statistical activity analysis of GPR data was shown to be an effective alternative method of plan view presentation by Barker et al. (1998).
Annan A.P. and Cosway S.W. 1998 Ground Penetrating Radar Survey Design. Paper Prepared for the Annual Meeting of SAGEEP. April 26-29, Chicago, Illinois.
Barker P., Fletcher M., Bradley J. 1998. Reflections on the past: Progress in the application of GPR in Archaeology. Proceedings of the Seventh International Conference on Ground-Penetrating Radar. May 27-30, 1998, Lawrence, Kansas, USA
Clark, Anthony J. 1996 Seeing Beneath the Soil. Prospecting Methods in Archaeology. B.T. Batsford Ltd., London, United Kingdom
Conyers, Lawrence B. And Dean Goodman 1997 Ground Penetrating Radar: An Introduction for Archaeologists. Walnut Creek, CA.: Altamira Press.
Sharma, Prem V. 1997 Environmental and Engineering Geophysics. Cambridge University Press, Cambridge, Unted Kingdom.