Validation of Sediment Accumulation Regions in Kansas, USA
345
The challenge in genetic and sequence stratigraphy is to detail three dimensional changes to develop models of sedimentation for use in prediction. Models
that can be applied to petroleum geology in a maturely developed province such
as Kansas necessarily need to be three-dimensional and consider large volumes
of data in order to be most useful and successful. Regionalized classification is
an inverse procedure that extracts patterns that can be used to develop a tectono-stratigraphic model tailored to the local setting.
2
Methodology
The validation and testing of SARs and their boundaries is addressed using two
approaches. First, a test is made of the significance of SARs and their boundaries
in controlling the internal stratigraphy and lithologies associated with the stratigraphic interval used in the classification. This test is accomplished through the
use of a regional, 300-km long color gamma-ray cross- section with each well
supported with analyses of accompanying neutron-density porosity logs. This
section crosses eight SARs.
Gamma ray logs were digitized from 25 wells with the interval of interest, ranging in thickness from 100 to 350 m . The natural gamma radiation is recorded by a
potassium iodide crystal in the wireline tool that responds to radiation contributed by potassium, uranium, and thorium. The magnitude of the gamma radiation
is associated with major lithologies - shale, sandstone, and carbonate rock.
Color is assigned to the gamma-ray intensities and presented as a cross-section to facilitate recognizing lithologic changes. The PfEFFER software is used
to read digital LAS (log ascii standard) files and to construct the color cross-sections. Well logs are digitized by scanning the logs into TIFF files which were importing into a software package, Neuralog, used to digitize the log image on a
screen. Digitized log ascii standard files were created that were read into the Excel-based log analysis software to generate the cross-sections.
A second approach in substantiating SARs is through their comparison with
gravity and magnetic data. The digitized boundaries of the SARs were overlain
on maps of the (1) residual Bouguer gravity map of Kansas with second-order
regional trend removed and vertical sun-shading (Xia et al. 1995; Fig. 4) and (2)
aeromagnetic map of Kansas, reduced to pole, and to a horizontal plane and
with vertical sun-shading (Yarger 1983; Fig. 5). These potential fields data were
imported into a workstation for image analysis and overlain with other data.
Regionalized classification is conducted as two fundamental phases, typification and regionalization. Typification is accomplished using cluster analysis, independent of the locations of the wells. Regionalization is conducted by estimating the probabilities that a specific well location should be assigned to each possible class, represented by an individual cluster. The sum of these probabilities
will be 1, so the well location is certain to have some assignment. Maps are then
made on the probability of membership of every location into each class. Every
point on the map is assigned to the class for which the probability of member-
345
The challenge in genetic and sequence stratigraphy is to detail three dimensional changes to develop models of sedimentation for use in prediction. Models
that can be applied to petroleum geology in a maturely developed province such
as Kansas necessarily need to be three-dimensional and consider large volumes
of data in order to be most useful and successful. Regionalized classification is
an inverse procedure that extracts patterns that can be used to develop a tectono-stratigraphic model tailored to the local setting.
2
Methodology
The validation and testing of SARs and their boundaries is addressed using two
approaches. First, a test is made of the significance of SARs and their boundaries
in controlling the internal stratigraphy and lithologies associated with the stratigraphic interval used in the classification. This test is accomplished through the
use of a regional, 300-km long color gamma-ray cross- section with each well
supported with analyses of accompanying neutron-density porosity logs. This
section crosses eight SARs.
Gamma ray logs were digitized from 25 wells with the interval of interest, ranging in thickness from 100 to 350 m . The natural gamma radiation is recorded by a
potassium iodide crystal in the wireline tool that responds to radiation contributed by potassium, uranium, and thorium. The magnitude of the gamma radiation
is associated with major lithologies - shale, sandstone, and carbonate rock.
Color is assigned to the gamma-ray intensities and presented as a cross-section to facilitate recognizing lithologic changes. The PfEFFER software is used
to read digital LAS (log ascii standard) files and to construct the color cross-sections. Well logs are digitized by scanning the logs into TIFF files which were importing into a software package, Neuralog, used to digitize the log image on a
screen. Digitized log ascii standard files were created that were read into the Excel-based log analysis software to generate the cross-sections.
A second approach in substantiating SARs is through their comparison with
gravity and magnetic data. The digitized boundaries of the SARs were overlain
on maps of the (1) residual Bouguer gravity map of Kansas with second-order
regional trend removed and vertical sun-shading (Xia et al. 1995; Fig. 4) and (2)
aeromagnetic map of Kansas, reduced to pole, and to a horizontal plane and
with vertical sun-shading (Yarger 1983; Fig. 5). These potential fields data were
imported into a workstation for image analysis and overlain with other data.
Regionalized classification is conducted as two fundamental phases, typification and regionalization. Typification is accomplished using cluster analysis, independent of the locations of the wells. Regionalization is conducted by estimating the probabilities that a specific well location should be assigned to each possible class, represented by an individual cluster. The sum of these probabilities
will be 1, so the well location is certain to have some assignment. Maps are then
made on the probability of membership of every location into each class. Every
point on the map is assigned to the class for which the probability of member-