Barents Sea Simulation Study

A large amount of magnetic data from both observatories and variometers is publicly available for a variety of scientific and technical applications. Unfortunately, many of these data have significant issues, such as unreliable baselines and baseline jumps.

Significant effort was therefore spent to derive a corrected and validated data set of all available global 1-minute data from 1995 through 2012. These were used here in a large-scale simulation study to characterize and compare the performance of the nearest observatory, IIFR and disturbance function methods. The locations of all observatories and variometers in the data set are shown in Figure 4.9.

Recently, some observatories and variometers have also started reporting 1-second data. However, this leads to very large data sets which are beyond the scope of this study. Furthermore, depending on the data transmission method used, downhole surveys can usually only be timed accurately to within about 1 minute, so that minute- averages are considered an appropriate choice of temporal sampling.

Because the study was carried out on observatory data from 1995 through 2012, some of these stations are no longer available for use, and some new stations have been added. Figure 4.10 below shows all Barents Sea region observatories that should be available for use going forward.¹

1 Data are available from http://flux.phys.uit.no/map/, http://geomag.gcras.ru/obs.html, and

Figure 4.9: Locations of magnetic observatories with publicly available data

Figure 4.10: Magnetic observatories applicable to the Barents Sea

In order to characterize the extent by which the disturbance function and IIFR methods reduce the disturbance error, all station triplets above 50˚ latitude among whom no two were farther apart than 600 km were identified. This subset of stations is shown in Figure 4.11. The list of selected stations is given in the Acknowledgements section, with acknowledgement of the organizations running these stations.

The Nearest Observatory, IIFR and DF correction methods were then applied to each triplet of stations. 600 km was chosen for a couple of reasons. First, as can be seen in Figure 4.10, that is the approximate distance at which it is no longer beneficial to use uninterpolated single-observatory disturbance data, which is the simplest default method. Second, this is approximately the distance from mainland Norway to the Svalbard archipelago, as well as the radius of the Barents Sea, so it is a good representative distance for this region. The triplets can be seen in Figure 4.11 as connected triangles.

Figure 4.11: Triplets of stations (depicted as triangles) within 600km of each other upon which a disturbance simulation was performed.

For any selected "target" observatory, mimicking the drill site, the other two observatories were used to synthesize disturbance values at that target location using the nearest observatory, IIFR and DF methods.

After synthesis, residuals were taken by subtracting the synthesized value off the actual value, and then these residuals were split into systematic and random parts. The industry is concerned with both long-period (“systematic”) and short-period (“random”)

disturbance field variations. Here, we defined the systematic disturbance field contribution as the 3-day average residual, which may be considered representative for the time taken for a single downhole BHA run. The random variation is then defined as the per-minute deviation from that running mean.

Statistics were performed on the absolute values of these residuals in order to determine the confidence intervals expectable through each method. It is important to note that while a 3-sigma value is generally used to calculate 99.7 percent confidence intervals, the distribution of actual geomagnetic disturbances does not follow a Gaussian normal distribution. In fact, the 99.7 percentile on geomagnetic disturbances is generally closer to the 6-sigma value. In order to more accurately predict these 99.7^th percentile values in this scenario, the residuals were binned to determine the actual value of the 99.7 percentile.

The goal of the simulation study was to assess the level by which downhole data would be affected by the disturbance field, before and after applying corrections. Specifically, we were interested to compare how much of the disturbance field could be reduced with the different available mitigation methods. Without applying a disturbance field mitigation method, the remaining disturbance is simply the actual disturbance. Thus, the same 99.7^th percentile value finding method described above was also used on the raw, uncorrected disturbance values measured at the observatory representing the drill site.

The plots in Figure 4.12 through Figure 4.17 below show the systematic and random disturbance field contributions as a function of geomagnetic latitude, all of which have been fitted with a smoothed Bezier spline for more clarity. As expected, the uncorrected 99.7^th percentile value (shown in red) is always higher than the IIFR value (shown in yellow), which is in turn higher than the disturbance function value (shown in green).

This confirms that the disturbance function method achieves the most significant reduction in the disturbance field, followed by the less-accurate, though relatively simple, IIFR method, and that using no disturbance mitigation method at all can lead to large uncorrected errors.

Figure 4.12: 99.7 % error percentiles for systematic declination disturbance remaining after corrections

Figure 4.13: 99.7 % error percentiles for random declination disturbance remaining without (red) and after corrections

Figure 4.14: 99.7% error percentiles for systematic dip disturbance remaining after corrections

Figure 4.15: 99.7% error percentiles for random dip disturbance remaining after corrections

Figure 4.16: 99.7% error percentiles for systematic Btotal disturbance remaining after corrections

Figure 4.17: 99.7% error percentiles for random Btotal disturbance remaining after corrections

It is of note that there are auroral electrojet effects visible in the above plots. As the electrojet current traverses east-west, it has different effects on the declination, dip, and total field. Declination disturbance is mostly caused by other effects, thus the bump visible in the uncorrected plot (red) at 65° in Figure 4.13 is fairly small. The largest effect on dip is at the crest of the current, with smaller effects to each side, thus the uncorrected disturbances are quite large around 65 ° in Figure 4.15. For total field, the largest effect is north of the current’s crest, with medium effect south and small effect at the crest. This is illustrated in Figure 4.17 by the larger bumps to either side of the crest.