Data Selection and Reliability Criteria

6.2.1 Selecting Data for Paleomagnetic Analysis

The laboratory and analytical techniques used by paleomagnetists have evolved and improved substantially with time. Cryogenic magnetometers with their greater sensitivity and increased speed of measurement have led to larger and more elaborate field collections. The use of computers has enabled more sophisticated data analyses (such as principal component analysis) to be carried out on large data sets. These developments have led to a general improvement in the quality and quantity of data, which in turn has led to the currently available data being of highly variable reliability. One cannot therefore merely take all paleomagnetic measurements and use them indiscriminately as if they were equally valid representations of the paleomagnetic field at the time a given rock unit was formed. Thus, it is necessary to set up generic selection criteria that can be used to eliminate those data that may be suspect. Naturally, these reliability criteria are not foolproof and additional careful selection may still be required.

6.2.2 Reliability Criteria

With the improving quality of the data several different schemes for judging the reliability of data have been suggested (see e.g., Irving, 1964; McElhinny, 1973a; Khramov, 1987; Van der Voo, 1993). Some hierarchical schemes have been proposed (Briden and Duff, 1981; Li et al., 1990) but these have now been abandoned because results that failed the first hierarchical level were automatically rejected even though they had other excellent attributes that would have survived subsequent levels. The problem that arises in setting up any selection scheme is that even if some criteria are not satisfied, the pole position may still be a valid record of the ancient magnetic field, and conversely those results that satisfy more than the minimum criteria may sometimes turn out to be seriously in error. In practice it is often easier to know when a pole position has been well determined than it is to know with any certainty that it is flawed. Van der Voo (1990a) proposed a useful set of seven reliability criteria that can be used for the selection of paleomagnetic data for tectonic studies. The selection criteria are simple to use and overcome the objections to the hierarchical schemes.

There are three basic criteria for a good paleopole determination: structural control, age of the pole, and the appropriate laboratory treatment of sufficient samples. In the scheme proposed by Van der Voo (1990a) one point is scored for each of the criteria that is judged to have been satisfied in any study. A quality factor Q in the range 0-7 is then assigned to each study and the data can be accepted or rejected according to the overall point score. A summary of the seven criteria is given in Table 6.1.

Van der Voo (1990a) suggests the following standards for the different criteria. For criterion 1 the age for Phanerozoic rock units should be constrained to be within a half-period, such as Late Jurassic or Early Silurian, or to be within a numerical age range of ±4%, whichever is larger. For Precambrian rocks the age limits should be ±4% or ±40 Myr, whichever is smaller. It is important that both the minimum number of samples and the minimum precision parameter and the minimum error be satisfied (No. 2). The demagnetization procedure must appropriately isolate the various magnetic components (No. 3). Blanket demagnetization treatment is not sufficient for this criterion to be satisfied. Only when vector subtraction is performed as illustrated by orthogonal vector diagrams, by the use of stereonets giving change in direction combined with M/Ma intensity decay plots, or by principal component analysis can one be

Was this article helpful?

0 0

Post a comment