Subterranean Visualization Through Multi-Sensor Fusion*

We depend upon the subsurface for much of the energy and raw material that enables
civilization. Groundwater sources often supply water for agricultural, drinking, and industrial use.
However, we remain relatively blind in the subsurface environment. Geological formations can
be strongly heterogeneous at all length scales, making it difficult to locate resources.
Furthermore, the performance of subsurface engineering is difficult to monitor due to our
blindness.
I will describe challenges associated with engineered geothermal systems (EGS) and hydraulic
fracturing in shale formations. In both instances, our optimization of an engineered system is
impeded by gaps in understanding of the fundamental processes and our inability to monitor
and visualize system performance.
I will present sensors that help us image the subsurface. Of note is the recent growth of fiber
optic measurements. Leveraging interferometry applied to light pulses sent down a fiber, we
infer the state of the fiber along its entire length. However, each sensor technology has
strengths and weaknesses. By combining multiple sensors, we can best avoid blind spots.
I will discuss examples of multi-sensor fusion in petroleum and EGS field projects where
concurrent measurements of velocity, acoustic, chemical, thermal, and resistive properties
enable determination of system performance. Today, such projects rely on large teams of
scientists and engineers to collect and interpret data and make engineering decisions.
Increasing automation through machine learning is showing promise for decreasing the lag
between measurement and decision making. I will close by describing a new initiative that seeks
to develop a comprehensive platform for integrating data and providing a virtual environment for
exploring the subsurface.