Earth data science & questions impacting society

With the widespread availability of large-scale computational resources, and the unprecedented quality and quantity of scientific data being collected today, we have opportunities to progress on problems relevant to society (climate change, water availability, management of resources, and many more) by asking in-depth questions and performing analyses that would have been impossible only years ago. In this talk, I discuss three problems where the methods of data science are opening new avenues of research in the Earth Sciences. The first example considers the detection and classification of unexploded ordnance from geophysical data. I demonstrate how machine learning can be used to augment, and in some cases supplant, technologies that are currently used for making decisions on which objects are ordnance and need to be dug up and disposed of and what items can be left in the ground. The second example focuses on advancing geophysical inversion for mineral exploration and incorporating machine learning to extract a geological model from geophysical data, petrophysical data, and a priori geologic knowledge of the area. In the final example, I discuss groundwater applications, including a humanitarian geoscience application: exploration for groundwater resources in stressed villages in Myanmar. Together, these examples demonstrate components of data science including numerical simulations, machine learning, inversions, and their combination, for solving applied problems. Additionally, they motivate the need for the integration of information across disciplinary lines. Open-source projects provide a means of facilitating multidisciplinary connections; I will discuss progress on open-source projects in research and education that I am involved in, including Jupyter meets the Earth, SimPEG, and GeoSci.xyz, and provide a vision for how these can be used to advance education and research so that problems can be tackled in a manner which is reproducible, facilitates collaboration, and enables a diverse set of stakeholders to engage with geoscience research.