19-33. Machine learning and the next generation of geothermal energy assessments
Closing Date: January 4, 2021
This Research Opportunity will be filled depending on the availability of funds. All application materials must be submitted through USAJobs by 11:59 pm, US Eastern Standard Time, on the closing date.
How to Apply
Apply Here
Background: The understanding of what constitutes a geothermal energy resource has evolved over time. Conventional hydrothermal systems were the first systems identified and developed to produce electric power, taking advantage of natural groundwater circulation through rock at elevated temperatures, resulting in significant advective transport of heat to near land-surface. Geothermal heat is plentiful, and if deep permeability could be located or engineered (EGS) within the identified hot rock underlying the western U.S., then geothermal energy could provide electric power equivalent to approximately 50% (~500 GWe) of the current U.S. electric power production capacity. This estimate is based primarily on the volume of sufficiently hot rock above 6 km depth, so more energy could conceivably be available if drilling and fracturing technology were sufficient to allow the creation of connected permeability at greater depths.
Broadly, EGS systems are characterized by favorable stress conditions such that the mode of failure is anticipated to create open fracture sets that create connected flowpaths between injection and extraction wells. While EGS is a potentially large resource, it is not yet a widespread commercially-viable resource. As such, a great deal of effort over the past decade (e.g., DOE FY20 funding is ~$46M) has been and is being expended to characterize EGS resources and to advance technologies necessary to establish viability. Since the last EGS assessment (provisional assessment; Williams and DeAngelo, 2011), the USGS Geothermal Resources Investigations Project has made a concerted effort to better characterize these resources, focusing efforts on knowledge-gaps identified during the previous assessment and fundamental science questions about creation of permeability for the purposes of circulating fluids to extract heat. Significant questions remain regarding the constraints on EGS development, including the roles of temperature, lithology, stress, depth, fluid composition, structure, and induced seismicity. International research, DOE-funded initiatives (e.g., FORGE, Collab, etc.), and other important research by USGS and others has greatly improved our understanding of EGS resource accessibility, and this understanding can be applied to focus on parts of the map that are accessible using current technologies, thereby allowing for quantitative estimation of potentially accessible resources. This will allow identification of regions where the resource should be easier and more economical to develop.
Description of the Research Opportunity: Potential research topics are myriad, but the ideal might be conceived as creating a computational environment (e.g., machine learning; hereafter ML) that takes raw data from a range of instruments, with a range of qualities and resolutions, and from that raw data, create maps of EGS resources. Data fusion will be predicated on the last decade of research, ongoing research, and use of massive new geophysical datasets. The focus area for research is anticipated to be the Great Basin of the southwestern U.S.
New Data: Recently, USGS Energy and Minerals Programs have combined resources with the U.S. Department of Energy Geothermal Technologies Office to fund massive geophysical surveys across vast tracts of the western United States (Earth MRI), to aid in mapping subsurface structure and geology. These surveys will cover areas that are known to be rich in both geothermal and mineral resources. The quantity of data and potential to inform natural resource assessments are unprecedented.
Significant research contributions can be made at many steps along this research process. Many topics have been formulated by the ML community, and a few of these are translated into example USGS assessment needs here:
- Cleaning and processing data often consist of steps that are both art and science. Can these be automated? Can human verification and interaction steps fit into this process to handle complex variants that are not properly accounted for in automated processes (e.g., expert skills to distinguish data artifacts)?
- Can a range of “soft” models be used to facilitate the incorporation of human understanding into machine learning paradigms? For example, supposing we identify several types of EGS systems, then can we use conceptual models of these systems to improve understanding of “big data” models?
- Can mathematical models of process inform selection of both input data and desired predictions?
- Can favorability maps be constructed to better characterize EGS resources, estimating favorable areas for resource development?
- Can we evaluate the effect of data granularity on data-worth and predictions?
- For interpreted inputs (e.g., regional heat flow maps, filtered geophysical data), can we assess sensitivity of predictions to interpretation-assumptions?
The preceding list of questions is intended to encourage creative thinking, but this list is far from comprehensive. Applicants are encouraged to think creatively about research needs and novel approaches. Interested applicants are strongly encouraged to contact the Research Advisor(s) early in the application process to discuss project ideas.
Proposed Duty Station: Portland, OR or Moffett Field, CA
Areas of PhD: Geology, geophysics, mathematics, engineering, statistics, computer science, physical science, or related fields (candidates holding a Ph.D. in other disciplines, but with extensive knowledge and skills relevant to the Research Opportunity may be considered).
Qualifications: Applicants must meet one of the following qualifications: Research Computer Scientist, Research Engineer, Research Geologist, Research Geophysicist, Research Mathematician, Research Physical Scientist
(This type of research is performed by those who have backgrounds for the occupations stated above. However, other titles may be applicable depending on the applicant's background, education, and research proposal. The final classification of the position will be made by the Human Resources specialist.)
Human Resources Office Contact: Beverly Ledbetter, 916-278-9396, bledbetter@usgs.gov