Transcript Psuedo-Spatial Analysis of Ternary Diagrams for the

Spatial Analysis of Geochemical
Diagrams for the Classification of
Igneous Rocks Using ArcGIS: A
Prototype for the Central Colorado
Assessment Project
By John Horton - Penn State MGIS Candidate
Physical Scientist - US Geological Survey
July 8, 2008
Academic Advisor: Tim White
The Central Colorado Assessment
Project (CCAP)
 The Front Range urban corridor of Colorado has
experienced rapid growth in recent years.

Raises concerns of resources and land management
issues faced by governing agencies in the region.
 Project being conducted by the USGS:
“to provide comprehensive
geoscience data and interpretations
that will allow federal, state, and
local land management entities to
make informed land-use decisions
in central Colorado.”
Updating Geochemical Data
 Creation of updated geochemical databases
(including rock samples) is essential in interpretations
of geology in the study area.
 Enhances understanding of landslide hazards,
mineral resources potential, and surface groundwater pollution due to acid mine drainage
related to historic mining.


Requires petrologist assigned to project to classify
thousands of rock samples in a standardized way to
help interpret the geologic setting throughout the
area.
Database of classifications will be spatially included
with updated digital geology, structural, and other
geochemical data (soil/stream sediments) to aid in
spatial analysis of the area.
Rock Types
(quick background)
 Three basic types of rocks on Earth:

Igneous: rocks derived from melt.
granite, basalt, gabbro
Two major subtypes:
Intrusive – plutonic (deep cooling within upper crust)
Extrusive – volcanic (rapid cooling on surface)


Sedimentary: rocks derived from physical and chemical
processes of the earth.


sandstone, limestone, shale
Metamorphic: transformation of existing rock types by heat
and/or pressure.

marble (limestone), slate (shale), gneiss (granite)
Igneous Rocks - Significance
 Geologically important





95% of upper crust is igneous in origin
Provides information about composition of the Earth’s
mantle and processes occurring deep within the Earth.
Radiometric dating provides absolute ages of rock
formation or alteration – allows determination of
approximate ages of other rock types (sedimentary) for
correlating ages of geologic units worldwide.
Characteristics determine setting in which rock formed;
important for understanding tectonic processes.
Majority of the World’s and Colorado’s mineral deposits
are located within igneous rocks.
Rock Classification
 Field criteria - location and structural
characteristics present (geologic setting).
 Texture – different types of crystallization.
 Mineralogy – minerals present (thin section)
 Geochemical characteristics – chemical
analysis of samples.

This is the type of classification that will be
used in this project.
Geochemistry of Igneous Rocks
 Rock samples are chemically analyzed using different methods
to identify various elements: inductively coupled plasma (ICP),
x-ray fluorescence (XRF), and instrumental neutron activation
analysis (INAA) are the primary ones.
 Whole rock concentrations of various elements, compounds,
and isotopes can be used to classify the rock by name and also
aid in petrogenesis classifications.
 Multiple variation diagrams (graphs) have been published over
the past 50+ years to classify igneous rocks.
 The International Union of Geological Sciences (IUGS)
formed a commission in 1970 in order to standardize the
various classification schemes.
 There are many other published diagrams from the literature
that provide more specific classifications (missing from IUGS
diagrams) and also petrogenesis classifications based on
geochemistry.
A Current Day in the Life of an
Igneous Petrologist
 Representative rock samples are prepared for analysis, analysis
is returned from lab, get table of concentrations for various
elements/compounds.
 Plot values on multiple diagrams for classification – visually
identify classification and record.

Very tedious process. Requires using multiple graphing
programs (1 program doesn’t have all the diagrams).
 Learning multiple interfaces.
 Merging of multiple output files.
 Hard to automate (different programs have different
scripting/schemes).

Compare rock concentrations to various geochemical rock
databases (GEOROC, PETDB).

Also tedious. Results not always useful.
 Requires data entry of multiple specific chemical parameters.
 Multiple matches to sample; classification uncertain.
Question: Is it possible for ArcGIS to automatically return
classifications based on point positions in graph space?

Need to classify 5,000+ rock samples within CCAP study area from various
sources. The current National Geochemical Database only has 1445 igneous
rock samples classified in the study area.

Focusing on igneous rocks for now since they are the most dominant lithology.

Usual classification process is very tedious and infeasible for this large of a
sample population.

An automated method is needed to facilitate the classification process. The
petrologist can then focus efforts on interpretation and collaboration with other
scientists on project.
Answer: YES!
ArcGIS can certainly be used to help automate the process.
A spatial join will return classifications (point in polygon spatial analysis
using geoprocessing tools) and a Python script can be written to repeat
the process through all the diagrams.
Overview: Using ArcGIS to
‘automate’ the classification process.
 Remove geographic portion of GIS by not defining a projection. Space will be




entirely numeric.
Treat diagrams as ‘maps’ and create polygons of classifications from them
(each diagram is a separate shapefile).
 Scan complex diagrams (curved lines) and spatially reference image to
numeric space using diagram’s chemical parameters.
 OR provided coordinates of polygons (boundaries) by petrologist.
Plot each rock sample’s chemical concentrations for the appropriate diagram
as XY points.
Perform a spatial join of points in polygons to return classification attributes
from polygon and assign it to rock sample.
Output a table with all classification results from all the diagrams (shapefiles)
to petrologist for further analysis.
 Creation of an igneous rock classification geodatabase for CCAP project.
Will be able to query on rock name/tectonic setting.

Example: show me all the samples classified as gabbro within the study area;
instead of show me all samples with 48-52% silica oxide, 5-9% magnesium
oxide and greater than 8% calcium oxide.
Step 1 - Digitizing the Diagrams
 Scan complex classification diagrams (curved
lines) and spatially register image to numeric
graph space in Arc.
Step 1 - Digitizing the Diagrams
 Scan complex classification diagrams (curved
lines) and spatially register image to numeric
graph space in Arc.
 Create polygon shapefile and name it
according to name of classification diagram.
 Add field for classifications (name of field
same as name of diagram; since final table
output will have attributes from multiple
diagrams).
 Digitize polygons from diagram and attribute
the classifications.
Digitizing – Alternate Method
 Diagrams with straight lines or very complex
structure.

Given coordinates of polygons (boundaries) by
petrologist, so a point file can be generated
and then polygons are digitized by snapping to
these points.
Partial List of Diagrams
(50+ when finished)
Plot
Type
Reference
SiO2 vs K2O
binary
Peccerillo and Taylor 1976
Sr vs Rb
binary
DeWitt, adapt of Ewart 1982
SiO2 vs K2O
binary
Ewart 1982, modif. Pec. & Taylor 1976
Ti/100 vs V
binary
Shervais 1982
SiO2 vs Na2O+K2O
binary
Cox at al. 1979, revised by Wilson 1989
SiO2 vs Na2O+K2O
binary
Cox at al. 1979
Fe2O3+MgO vs Al2O3/SiO2
binary
Bhatia 1983
Fe2O3+MgO vs TiO2
binary
Bhatia 1983
Zr/10 vs Sc vs Th
ternary
Bhatia 1986
Sc vs Th vs La
ternary
Bhatia 1986
Y vs Zr/4 vs 2Nb
ternary
Meschede 1986
P2O5*10 vs MnO*10 vs TiO2
ternary
Mullen 1983
SiO2/Al2O3 vs Fe2O3/K2O
binary
Herron 1988
K2O/Na2O vs SiO2/Al2O3
binary
Rosert 1986
SiO2 vs K2O/Na2O
binary
Rosert 1986
4Fe-2Mg-3Ca vs 12CCO2-4Al-8(Na+K)
binary
DeWitt 1994 (IF1 vs IF2)
Ta/Yb vs Th/Yb
binary
Pearce 1982
Ta vs Th vs Hf/3
ternary
Wood 1980
Nb/16 vs Th vs Hf/3
ternary
Wood 1980
Ta*3 vs Hf vs Rb/30
ternary
Harris 1986
Nb*3/16 vs Zr/40 vs Rb/30
ternary
Harris 1986
Ta vs Th vs Hf/3
ternary
Pearce 1996 mod of Wood
Nb/16 vs Th vs Hf/3
ternary
Pearce 1996 mod of Wood
Nb/Y vs Zr/Ti
binary
Winchester and Floyd 1977 + Pearce 1996
Y*3 vs Zr vs Ti/100
ternary
Pearce and Cann 1973 - mod 1996
Nb/8 vs La/10 vs Y/15
ternary
Cabanis? and Lecolle 1989
Ta*2 vs Th vs Tb*3
ternary
Cabanis? and Thieblenont? 1988
MgO vs Al2O3 vs FeO + TiO2
ternary
Jensen 1976
Y*3 vs Zr vs Ti/100
ternary
Pearce and Cann 1973
Zr vs Ti
binary
Pharoah?
Nb/Y vs Zr/TiO2
binary
Winchester and Floyd 1977
4Si-11(Na+K)-2(Fe+Ti) vs 6Ca+2Mg+1Al
binary
R1 vs R2 ????
4Si-11(Na+K)-2(Fe+Ti) vs 6Ca+2Mg+1Al
binary
R1 vs R2 ????
4Si-11(Na+K)-2(Fe+Ti) vs 6Ca+2Mg+1Al
binary
R1 vs R2 ????
4Si-11(Na+K)-2(Fe+Ti) vs 6Ca+2Mg+1Al
binary
R1 vs R2 ???? Alkalinity Numeric
Yb+Ta vs Rb
binary
Pearce et al. 1984
Yb vs Ta
binary
Pearce et al. 1984
Y+Nb vs Rb
binary
Pearce et al. 1984
Y vs Nb
binary
Pearce et al. 1984
SiO2 vs A/CNK
binary
?????? Peraluminous
SiO2 vs A/CNK
binary
?????? Peraluminous
(FeO+.89Fe2O3)/(FeO+.89Fe2O3+MgO) vs SiO2
binary
?????? Iron Oxides
K2O/K2O+Na2O vs SiO2
binary
?????? Potassium Oxides
Step 2 - Scripting
 A Petrologist will have a table of multiple rock samples each
with chemical concentrations from the results of analyses.
 Need to classify all the samples using multiple diagrams.


I will create a Python script to batch process all the samples
through all the diagrams.
Use of ArcGIS by non-GIS scientists.



USGS has an enterprise license for ArcGIS - installation
and use is free; hence no new software expense.
The Python script utilizes the geoprocessing tools of Arc
without prompting the user for commands.
Possibly create a custom igneous classification tool using
ArcObjects.
Initial Problems/Issues
 Some classification diagrams have log values (must




be re-plotted into normal arithmetic scales).
Some diagrams are ternary (use 3 variables). These
must be repositioned using only 2 variables.
Some diagrams have uneven X and Y axis ranges.
One set of values must be multiplied/divided by a
constant in order to spatially register the image in
numeric space.
Limited scripting experience.
Many new samples are not in digital format. Relying
on others to receive digital tables of the data for our
use.
Expected results
 Creation of an igneous rock classification feature class within a
geodatabase for the CCAP study area. Addition of thousands of
previously unclassified samples. New spatial queries on rock
samples by different classifications can now be analyzed
(instead of relying on chemical values only).
 Creation of maps and spatial analyses of new information
combined with other GIS layers as directed by CCAP project
chiefs.
 Publication of the ArcGIS igneous rock classification process as
a USGS Open File Report (public) so anyone in the world can
use it for other projects.
 Illustrate that this GIS-based classification process can be
extended to other rock types or to other classification problems
in other disciplines.
Project Timeline
 May: Digitized a few diagrams and developed a prototype script.




Tested on small sample population. A few bugs, but worked
correctly.
June/July: Continue to digitize diagrams as they are received.
Refine script to ensure correct output for petrologist.
August: Finalize diagram list and all diagram polygons.
Continue testing script on more sample suites from petrologist;
edit as necessary. Finalize script and give to petrologist for
testing. Verify results with previously classified samples.
Sept/Oct: Receive and incorporate feedback from petrologist
into finalized versions of script and diagram shapefiles. Get
approval on results and begin work with petrologist to classify all
samples within the CCAP study area. Incorporate or provide
output for use with other spatial databases in project (geology,
structure, geochemistry). Analyses as directed by project chiefs.
Nov/Dec: Present results at a USGS talk and a presently
unknown conference. Publish results with USGS publications.
Thank You!
Questions?
[email protected]
Central Colorado Assessment Project website:
http://minerals.cr.usgs.gov/projects/colorado_assessment/index.html
USGS - Central Mineral Resources Team website:
http://minerals.cr.usgs.gov/