CLIENT
Paleobiology Database
GOAL
Develop a predictive model that forecasts the prevalence of marine
mammal fossils across the globe.
PRIMARY
TOOLS
R & RStudio, GIS, Adobe Creative Suite
BACKGROUND
The client is a non-governmental non-profit public resource for
paleontological data. Their mission is to encourage and enable
data-driven collaborative efforts that address large-scale
paleobiological questions. The client requests a way to predict
potential fossil localities to discover new fossils.
PRODUCT
Our team developed a predictive model using marine mammals as study
organisms. The model uses the geospatial coordinates of all known marine
mammal fossils and identifies key geographic, environmental, and
socioeconomic predictors of fossil localities. The model successfully
forecasts locations with a high probability of yielding new fossil
discoveries.
The Challenge: different groups of organisms have their
distributions influenced by completely different factors.
The Solution: develop independent models based on organisms with
similar ecologies and geographic distributions to avoid homogeneity of
the model.
The Challenge: fossil distribution is heavily biased by
socioeconomic factors, such as GDP, which do not reflect a true
biological or geologic signal.
The Solution:
standardize the dataset for GDP and other socioeconomic factors to avoid
implicit Northern Hemisphere bias.
Now that we’ve delivered the model to the client, what’s next?
Marine mammals were an ideal starting point because their distribution is easily modeled to modern coastlines. Now, we can expand to other fossil groups, including terrestrial organisms.
The dstribution of fossils is just the first step. A major factor is being able to excavate fossils is the surface exposure of the bedrock. By combinind our model with geologic maps from the USGS, we’ll be able to forecast high value localities for paleontologists to prospect.