Processing of Magnetotelluric Data for Monitoring Changes in Electric Resistivity During Drilling Operation

Abstract

An increasing interest in magnetotelluric monitoring of hydraulic stimulation experiments results from soft stimulation techniques that reduce induced seismicity to a minimum. Long-term magnetotelluric monitoring of different injection and production experiments at the Rittershoffen geothermal site in Alsace (France) provided a first continuous data set over several month, covering the end of drilling phase of GRT2 (1), mostly, production from, but also injection into this well, injection into GRT1 and a circulation experiment. Transfer functions showed particular variation pattern for different operations, i.e. an increase in uncertainty, conductivity and phase during test operation with a preferential direction sub-parallel to Shmin, i.e. perpendicular to the expected extension of the fractures controlling the reservoir. In particular fluid injection, either into GRT2 or GRT1 causes a strong decrease in resistivity by up to one order of magnitude in the YX component between about 8–25 s of period. Here, we present particularities of processing of continuous magnetotelluric monitoring for the example of deepening of the RN-15 well on the Reykjanes peninsular (Iceland) to 4665m (IDDP-2 well) with a magnetotelluric dataset from November 2016 to January 2017. The drilling progress during this period was accompanied by partial and up to total circulation loss (2). Two continuous running MT stations, GUN and RAH, were installed on the Reykjanes peninsula. RAH and GUN are located about 6 and 1 km away from IDDP-2. Both MT stations are equipped with two electric dipoles in N-S and E-W direction, as well as three magnetic sensors oriented in N, E and vertical direction. MT monitoring was carried out with a sampling frequency of 512 Hz. Processing revealed the bad data quality of RAH. Consequently, MT data were processed using single site method with the code Bounded Influence Remote Reference Processing (3). Due to a temporally noise signal in the time series, they were down filtered to get the lower frequency bands and hence to clean the time series. Additionally, a remote station from Germany has been tested to improve the data quality in the dead band. First results might reveal a correlation between fluid loss, induced seismicity and low resistivity in the magnetotelluric data. References: (1)Abdelfettah, Y., Sailhac, P., Larnier, H., Matthey, P. D., Schill, E., 2018. Continuous and time-lapse magnetotelluric monitoring of low volume injection at Rittershoffen geothermal project, northern Alsace – France. Geothermics, 71, 1-11. https://doi.org/10.1016/j.geothermics.2017.08.004. (2)Chave, A.D., and D.J. Thomson, 2003. A bounded influence regression estimator based on the statistics of the hat matrix, J. Roy. Stat. Soc., Series C, (Appl. Statist.), 52, 307-322. (3)Friðleifsson, G. Ó., Elders, W. A., Zierenberg, R. A., Stefánsson, A., Fowler, A. P., Weisenberger, T. B., Mesfin, K. G., 2017. Scientific Drilling, 23, 1-12.