Antarctica is buried under a dense blanket of ice, spanning 14 million square kilometers making it the largest mass of ice on Earth.
Less than 1% of this continent is ice-free. And in that fraction lies almost all of Antarctica's biodiversity.
And in those exposed regions, buried deep beneath the ice, lies the substance that most people
never think about.
The soil
These arid, gravelly soils are shaped by
extreme cold and dry conditions.
This climate have created an incredibly
unique geochemistry that serves as the
baseline for everything that can survive here.
Despite low biological activity relative to other ecosystems, these soils support a diverse and unique set of organisms.
Without these unique soils and
conditions as a baseline,
none of this life could survive
Thousands of expeditions have been made over the
past century in service of Antarctic geochemical
and environmental research
And yet the geochemistry of these
critical soils remains largely unmapped
These soils are barometers of environmental change
if we can't gain a good understanding of them,
we can't track what's changing, and what's being lost
As global temperatures rise, Antarctica's
ice shelfs are melting and exposing soils
that have been buried for thousands of years.
Having geochemical baselines is critical for understanding the impacts of these changes.
One approach to gaining this understanding of these soils is to take a diverse array of thousands of samples from all 28 distinct
ice-free regions and analyzing their geochemical properties in a lab
Antarctic Conservation Biogeographic Regions (ACBRs)
But sampling everywhere just isn't realistic
Because of these constraints, for the
entire continent, our lab has analyzed
just 171 samples in 4 regions
Ideal coverage would require thousands
of samples from all 28 ACBR regions.
What we have is a mere fraction of that.
171 samples · 4 regions · 28 locations
Open-access satellite data can give us detailed observations of the physical properties of every square meter of Antarctica's surface.
For our purposes, we sourced 9 layers of
spatially-derived environmental data.
Elevation
Precipitation
Slope & Aspect
Lithology
However, there is another way to look at Antarctica,
one that doesn't require a single expedition
By combining our satellite data with our lab analysis
we can make detailed predictions about soil geochemistry
Machine learning algorithms can be used to learn the relationships between our different
data sources to make guesses about the conditions in places we've never been before
Each layer reveals one of our best predicted geochemical properties in unseen regions.
Why does it look like this?
Penguin and seal colonies enrich coastal soils with marine nitrogen via guano.
Inland areas that are far from any animals show much lower values.
Why does it look like this?
Coastal zones with high animal activity show elevated phosphorus from guano.
However, inland weathering of parent rock is the main source.
Why does it look like this?
Sea salt spray carried inland by coastal winds. The Transantarctic Mountains act as a barrier, creating a sharp drop-off into the interior.
Why does it look like this?
Nickel follows mafic and ultramafic rock formations almost exactly. The sharp prediction color changes indicate real geological boundaries.
Why does it look like this?
Titanium predictions follow the Ferrar Dolerite volcanic formations. Weathering in Antarctica is so slow that parent rock chemistry is preserved almost unchanged in the soil.
Climate Science
Geochemical predictions are a baseline for tracking how soil conditions evolve, allowing us to quantify the change of elements as global temperatures rise.
Smarter Field Work
Predictions can help future researchers plan more informed field work expeditions, saving time, resources, and logistical costs,
ultimately maximizing the scientific yield of remote studies
Life Beyond Earth
Antarctic soils are our natural best analogue for extraterrestrial life. Understanding what survives here helps us in our search for life beyond our planet.
Predicting Antarctic Soil Geochemistry
Created by
Lily Eliason
Interested in learning more? Visit my Github :)
Sampled Geochemical Data
Soil geochemical data provided by the LeMonte Lab of
Environmental Geochemistry at Brigham Young University.
Open-Access Satellite Data
Environmental spatial layers sourced from REMA, WorldClim, and
SCAR open-access repositories.
This project reflects the very preliminary results of this ML project.
DSPX315 Final Project · Spring 2026 · Brigham Young University