Everyone knows that lava is made of melted rocks. It
makes sense that different types of rocks are produced from different types of lava;
for example, the lava in Hawaii is different than the lava common in a
Cascadian volcano like Mt. Saint Helens. The rocks and minerals that lava, or
magma, become also depends on temperature and pressure conditions during
cooling.
Geochemists are interested in understanding how
minerals are formed as molten rock cools. This knowledge helps them better
understand Earth’s geologic processes from the creation of critical minerals and
elements like lithium, to the way plate tectonics can build mountains and cause
earthquakes.
Experimental geochemists around the world melt rocks
and minerals inside special laboratory furnaces that recreate the environment
deep inside the earth. They then look at the minerals that form when various
mixtures of material are cooled at different temperatures and pressures.
The large amount of data from such experiments are a
challenge to compile for analysis. A team at lead by Gokce K. Ustunisik, Ph.D.,
associate professor of geology and geological engineering at South Dakota Mines
helped build a system to
compare results from thousands of experiments conducted around the world. This new study, thanks to a five-year National
Science Foundation (NSF) grant totaling nearly $470,000 will help align the data
from various sources to build the big picture understanding.
This award is one of four of Ustunisik NSF funded
research projects totaling nearly $750,000. The work also ties into the
university’s new Ph.D. program
in data science and engineering.
A challenge when compiling data from various experiments
is that data produced by different labs is not always collected and presented
in the professional literature in the same way.
“The way data was collected in multiple experiments
can differ greatly. It’s possible to develop bias in predictive models if you
don’t consider the boundary conditions of experimental data,” says Ustunisik, the
principal investigator on this research.
Roger Nielsen, Ph.D., a co-principal investigator on
this research, a research scientist at Mines, and an emeritus professor at
College of Earth Ocean and Atmospheric Sciences at Oregon State University uses
this analogy to describe this work.
“If your experiment is driving on Interstate-90 east
across South Dakota, the data you collect along the first part of the journey
tells you you’re going straight and flat and there are no big corners. If you
do a second experiment on I-94 across North Dakota, the data shows you the same
thing, straight, flat, no corners,” says Nielsen. “A model you might produce based on this data from
these two experiments would predict a straight and flat road, and this model
would work great, until you hit the Missouri River, and you end up over a cliff
in the water. In the two different experiments, you’d go into the water in two different
places. The two models going the same direction in two different places would line
up for a time but then have different results,” says Nielsen.
Nielsen says the work needed on these experimental
geologic datasets includes improved understanding of limitations, gaps, and
anomalies such as the Missouri River in the analogy above. With this new
understanding, researchers can then examine when different experiments have
data that correlates and diverges. They can use this improved understanding to
build better models.
“Whatever we do with experiments, our goal is always
to simulate what was observed in nature, what we are doing with this is to try
and understand the boundary conditions in different experiments so we can warn
the modelers about this bias,” Ustunisik adds.
With this new broader understanding, geologists hope
to build a more unifying theory or model for the innerworkings of the entire Earth.
Nielsen says another analogy is to consider the system that makes up the Earth
like the system that makes an automobile.
“Twenty-five years ago, when I began this work, we
were trying to determine what all the parts in the car were, the tires, the fenders,
the nuts and bolts and bearings that hold the pistons inside the motor. Today
we are trying to better understand how these parts fit together and make the
car run.”
A model of the entire Earth’s system would be
valuable in helping geoscientists predict natural disasters, like earthquakes
and volcanic eruptions. This new data analysis research at Mines, brings
geologists one step closer to this goal.