We have a great program lined up for this SEG with oral and poster presentations as well as one-on-one presentations at our booth #2214
Noise Attenuation in Sparse Surface Microseismic Datasets
Presenting author: Ben Witten
Day: Oct 15, 2018
Time: 1:50 PM - 2:15 PM
Variations on a noise attenuation workflow for sparse surface networks are described, implemented and tested. The flow makes use of the data redundancy afforded by super-stations, groups of closely spaced 3C sensors, which permits the application of a Multi-Channel Weiner Filter (MCWF) as well as non-linear stacking operators such as semblance weighted stacks to the records. Using a dataset recorded on 3 super stations, we examine the Signal to Noise Ratio (SNR) of arrivals from 71 events observed during 6 hours of hydraulic fracturing operations in order to test the efficacy of different variations in the workflow. In particular, we test the impact of the use of multi-component sensors on the noise cancelling properties of the MCWF, finding that including the the multi-component nature of data in the MCWF formulation provides 0.5-1 dB of additional SNR enhancement. We also examine the effect of using linear and non-linear stacking operators such as semblance weighted stacks both with and without prior coherent noise attenuation. The optimal variation on the workflow utilizes both a multi-component MCWF and non-linear stacking operators, it improves arrivals SNR by 7.2 dB relative to the original records. This in turn translates into a significant decrease in the detectable magnitude of events, and a corresponding increase in the number of events a given survey can expect to detect, thus allowing more detailed structure to be revealed by the microseismic cloud.
A journey from high quality data sets to real time risk management
Presenting author: Sepideh Karimi
Day: Oct 18, 2018
Time: 9:20 AM - 9:45 AM
Recent increases in the number and size of induced seismic events has led to demand for effective management of induced seismicity risk. This requires an accurate forecast of the largest magnitude event in near real-time, in order to allow adjustment of operational parameters for mitigation of the probability of inducing such large events. Many models have been proposed to estimate the magnitude of strongest event. Some of these models are rely solely on statistics of recorded seismicity while others account for the relation of event size with operational parameters. There are also models relating the maximum magnitude with existing geological and tectonic conditions.
When it comes to seismic risk management and mitigation, accurate analysis in real-time is a crucial factor. It is also essential to realize that enrichment of the input data in real time can improve the maximum magnitude estimations.
In this study we provide examples from seismic monitoring of hydraulic fracturing operations, in which the observed seismicity was played back to simulate real time monitoring conditions. Using a case study, we show how a statistics-based model (Van Der Elst et al., 2016) was able to forecast the maximum magnitude in real time. We also show an example that illustrates the limitations of the adopted model and discuss potential scenarios where it would fail.
Automatic event detection and location using Feature Weighted Beamforming
Presenting Author: Sepideh Karimi
Day: Oct 16, 2018
Exhibit Hall C, Poster Station 2 A
Seismic catalogs yielding the location and origin time of events serve as the foundation to risk management and public safety strategies, where the accuracy and response time of the catalog creation system is important. In a scenario where low magnitude events contribute heavily into real-time decision making, conventional autopicking and association techniques can yield large origin time and location differences due to the inclusion of noise picks. Here we present a new technique, Feature Weighted Beamforming (FWB), which can be applied in near real-time to increase the accuracy of automatically generated catalogs, while maintaining the systems sensitivity.
This method has been tested on multiple surface arrays ranging in radius from 5 to 250 km, and number of stations between 6 and 90. All tests have shown similar results for location accuracy improvements and noise event removal. The results from the private Duvernay Subscriber Array are given here. As compared to standard STA/LTA picking with subsequent associations, FWB reduced the number of false positives by 75% as well as reducing the average difference in the event location between automatic and manually picked solutions by 63%.