--> Quantitative Chemostratigraphy: A New Approach for Processing and Analysis of Geochemical Data, Melo, Dean P.; Capistrano, Gabriel G.; Meloni, Raphael B., #90100 (2009)

Datapages, Inc.Print this page

Quantitative Chemostratigraphy: A New Approach for Processing and Analysis of Geochemical Data

Melo, Dean P.1
 Capistrano, Gabriel G.1
 Meloni, Raphael B.2

1LAGEMAR, Universidade Federal Fluminense, Niteroi, Brazil.
2
Informática, Pontíficia Universidade Católica,
Rio de janeiro, Brazil.

The authors present the results of hierarchical cluster analyses (HCA), principal component analysis (PCA), and support vector machine (SVM) methods for the geochemical data of the ODP drilling cores of sedimentary deposits of the Leg 199 (site 1215 to 1221) acquired from the equatorial zone of the Central Pacific Ocean, with the objective of chemostratigraphics unit definition. The first step is compare the classification by the pattern-recognition method with classification based on lithologic variation described in the ODP reports. Considering that each lithology has a specific geochemical fingerprint, the experiment consists in the identification of these fingerprints using the pattern-recognition tools. The PCA has obtained excellent results, that is, 97% and 100% of the classifications of the ODP sites 1215 and 1217 are concordat to the visual classifications of the drilling cores. The above-mentioned classification are submitted to 3D graph (PC1, depth, and PC2) and it is fundamental for the observation of stratigraphic position of the samples and the groups. For the site 1218, about 80% of the PCA classification are concordant to the visual descriptive ones. However, the nannofossil ooze and nannofossil chalk layers, which were classified as different units, are grouped as one unit by the present method. For the site 1219, also about 80% of the PCA data are concordant to the visual descriptions. But, the upper and the lower radiolarian layers classified by conventional visual descriptions show notable chemical differences by the PCA. The PCA method defines the radiolarian composition as the main component of an interbedded unit made up of radiolarian, diatom, and nannofossil layers. By the way, the sites 1220 and 1221 have about 90% concordance, that is, better than the sites 1218 and 1219. The HCA results are not as favorable as the PCA one, being about 60% of concordance for above-mentioned drilling cores. But, this method has advantage in amount of works and times, being adequate for preliminary gross-mode analyses. The SVM has been applied to the analysis of the 1216 based on the training set data of the 1215 and the 1217. In this case, 28 in 32 samples, 87%, have obtained their chemical classification concordant to the visual ones. The above-mentioned methods can accelerate the data interpretation of the samples difficult to be analyzed by manual methods for chemostratigraphic unit definition.


AAPG Search and Discover Article #90100©2009 AAPG International Conference and Exhibition 15-18 November 2009, Rio de Janeiro, Brazil