GSA Annual Meeting in Indianapolis, Indiana, USA - 2018

Paper No. 208-3
Presentation Time: 2:05 PM


BROOKS, Edward M.1, STEIN, Seth1, SPENCER, Bruce D.2, SALDITCH, Leah1, NEELY, James S.1, PETERSEN, Mark David3 and MCNAMARA, Daniel E.4, (1)Earth and Planetary Sciences, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, (2)Department of Statistics and Institute for Policy Research, Northwestern University, Evanston, IL 60208, (3)Geologic Hazards Team, U.S. Geological Survey, Denver Federal Center, MS 966, Box 25046, Denver, CO 80225, (4)US Geological Survey, Geologic Hazards Science Center, 1711 Illinois Street, Golden, CO 80401

Seismicity rates in the central United States have increased dramatically in the past decade due to the underground injection of wastewater produced by oil and gas extraction. In response to the increased seismic activity, the USGS began creating one-year probabilistic hazard models and maps to describe the increased hazard posed to the central and eastern United States. Using the intensity of shaking reported to the “Did You Feel It?” system, we assessed the performance of the 2016 model using metrics to quantify the performance of hazard models and describe different characteristics of map behavior. By both the fractional exceedance metric, which describes how well a probabilistic model compares to a shaking record, and the squared misfit metric, which measures how similar a shaking record is to the predictions of a model, we found the map to work well. The model was remade the following year with an updated earthquake catalogue. Here, we assess the 2017 model’s performance using the same methodology. The 2017 model overpredicted shaking in the area surrounding Oklahoma, where most of the injection activity occurred. Unexpected events in Montana and Delaware resulted in shaking exceeding expectations elsewhere. Despite these discrepancies, overall the 2017 model performed well, with only slightly worse metrics than the prior year. Analysis of different sub-regions— including Oklahoma and surrounding states, Montana, and the east coast— categorizes performance by different metrics in different locations. The aggregate, with all the sub-regions combined, yields a map that works well as a probabilistic model, indicating that the model for generating these one-year maps is robust and useful. These results imply that one-year induced seismicity hazard maps have the potential to be valuable tools for policy makers and regulators in managing the seismic risks associated with unconventional oil and gas production and informing decisions pertaining to induced seismic hazard.