Geological Survey of Canada, 1 Observatory Crescent, Ottawa, Ontario, K1A 0Y3

Gaia Geoscience, P.O. Box 4175, Station "E", Ottawa, Ontario, K1S 5B2

I. Introduction

While recent magnetotelluric (MT) investigations of the Trans-Hudson Orogen (THOT) have focussed on subsets of the THOT92 dataset (see Jones et al., 1993a, Ferguson et al., 1996), the analyses presented here represent the first stage of a study which has as its central aim the derivation of a 2D regional model for the entire craton-to-craton transect. The MT distortion decompositions discussed in this report were undertaken, therefore, with a view towards establishing a regional strike direction for the whole transect. In the event that a suitable regional strike could not be found for all of the sites, the aim was to obtain regional strikes for subsets of the THOT92 site set.

In the following, we describe the procedures followed to determine the best-fitting strike directions for the data. "Best fitting" is defined as the strike direction that is consistent with the most number of sites and the broadest frequency range and leads to the lowest misfit between the distortion decomposed response estimates and the observations. As part of the procedure for determining the best strike direction, distortions caused by local, near-surface features are also removed. The resulting impedances should represent the best possible estimates of the underlying two-dimensional (2D) regional structure.

The THOT92 transect is an east-west profile which is comprised of 108 sites (Fig. 1a, Fig. 1b, Fig. 1c, Fig. 1d, and Fig. 1e). The profile extends from the Hearne craton at approximately 55^O N, 109^O W in northeastern Saskatchewan to the Superior craton at approximately 55^O N, 97^O W in northwestern Manitoba. At almost half of the sites (52 in total), soundings were made in the audio-MT (AMT) frequency band only (10 kHz -- 10 Hz). These sites are numbered "amt92-s__" in Figs. 1b-e and Tables 1-3. At the remainder of the sites (56 in total), data was collected in both the AMT and MT bands (10 Hz -- 0.0005 Hz), and these sites are numbered "mt92-s__". The data were acquired by a commercial contractor (Phoenix Geophysics), and acquisition was supervised by a number of bird-doggers working in rotation (Alan Jones, Ian Ferguson, Trevor Boyce, Jim Craven, Rob Ellis Colin Farquharson and Gary McNeice).

The basis of the Groom-Bailey (GB) distortion decomposition method is that the Earth can be modelled as a regional two-dimensional (2D) structure overlain by small distorting features. This regional structure has a strike direction, and an electromagnetic response which can be described by two complex numbers at each location and at each particular frequency. One of the numbers describes the response for electrical currents flowing along the structure (the transverse-electric, or TE, mode of induction), whereas the other describes the response for electrical currents flowing across the structure (the transverse-magnetic, or TM, mode of induction). Unfortunately, it is rarely possible to record these responses from the structure on the surface of the Earth without the effects of other structures distorting them. Particularly prevalent are the effects from very local, small-scale, surficial inhomogeneities that distort the electric field without perturbing the magnetic field. The effect of these galvanic distorters must be recognised in the data, and must be removed prior to interpretation. In their classic paper, Groom and Bailey (1989) describe a physically-based matrix decomposition approach that attempts to identify and correct the data for such artifacts. The responses from each site and at each frequency are fit to the seven parameter model according to the methodology outlined in Groom et al. (1993). Applications of the method can be found in Jones et al. (1993b) and Jones and Dumas (1993); a brief summary of the method is included here for completeness. This approach has been extended by McNeice and Jones (1996) to be applicable for a range of sites and a range of frequencies.

At each site, eight data exist for each frequency -- these correspond to the complex elements of the 2x2 MT impedance tensor. The eight data are fit to the seven parameter model which describes the local distortion of 2D regional electric fields. The parameters are the regional 2D strike direction, two measures of telluric distortion (twist and shear), and the two complex regional impedances for the two modes of induction. The regional impedances are scaled by indeterminable static shift factors, which can be resolved into a site gain, and the action of an anisotropy tensor which splits the observed apparent resistivity curves apart. This anisotropy can be removed (to first order) by shifting the high frequency asymptotes for the two apparent resistivity curves (one curve for the TE and the other for TM) to their high frequency geometric mean asymptotes. The site gain is a locally active static shift factor which raises or lowers the apparent resistivities at a site, and can only be established after resistivity curves from many sites are superimposed on the same set of axes. Removal of the anisotropy, and the correction of the site gain, is the last stage in the process.

In general, the site by site decompositions revealed two, and perhaps three, trends in the geoelectric strike direction across the THOT92 transect (see Table 1) (Note: a full set of decomposition plots and handwritten log entries of every site by site decomposition is available on request). Map summaries of the GB strike directions are included as Fig. 2 and Fig. 3 for the two datasets (amt-mt in Fig. 2, and amt-only in Fig. 3).

Sites located at the western end of the transect yielded strikes in the 28^O-34^O range with little difficulty when the method outlined above was applied. These sites included mt92-s39, s38, s37, s36, s35, s34 and s32 (Fig. 1b). The decomposition of the data from site mt92-s39 is typical for this group of sites, and is used as an example.

The unconstrained GB plots for mt92-s39 reveal that the twist and shear are well defined over most frequency bands-- except for the highest AMT frequencies where the data show a great deal of scatter (Fig. 4a). The shear was constrained first to a value of 0^O, as it appears to be the most stable, i.e., frequency-independent, of the three parameters (Fig. 4b). Next, the twist was constrained to its mean value in the band 900-1 Hz (Fig. 4c). This figure shows how the strike has stabilized in the 25^O-35^O range in the 100 Hz-0.01 Hz range. Finally, the strike is constrained to its estimated mean value in that range of 34^O (Fig. 4d). The model fits show a very good agreement with the observed data over almost the entire range of frequencies.

Moving from the west towards the east, the transect angles toward the Needle Falls Shear Zone at site mt92-s31 (Fig. 1b). Strikes obtained from the decompositions of sites mt92-s31, s30, s29, s42, and s28 fell into the 12^O-25^O range. It should be noted that these site by site results are in accord with the 22_^O strike used by Jones et al. (1993) in the construction of a 2D model for a subset of sites in this region sensitive to the presence of the North American Central Plains (NACP) conductivity anomaly. The decomposition of site mt92-s30 is representative of the sites in this group (Figs 5a-d).

The unconstrained plots for site mt92-s30 reveal more frequency dependence of the GB parameters (Fig. 5a). Initially stable at low values in the 900-1 Hz frequency band, the twist and shear increase dramatically at 5-7 Hz, and the misfit increases to unacceptable values. The shear approaches its maximum permitted value of 45^O and asymptotes to around 43^O, at the longer periods, indicating very strong distortion resulting in an almost singular impedance matrix. We concentrate, therefore, on the higher frequency values and attempt to fit a model to the data at those frequencies. The twist is constrained at its mean value of 2_^O over the 900-1 Hz band (Fig. 5b). The shear is constrained next to a value of 7^O (Fig. 5c). The third plot of the GB parameters for this site shows that the strike has now stabilized over the 100-0.001 Hz band (Fig. 5c). It is constrained to a value of 12^O (Fig. 5d). The Z_yx and Z_yy tensor elements are particularly well modelled in this decomposition; the fits to the Z_xx and Z_xy components are good in the range 1000-1 Hz and poorer at frequencies below 1 Hz.

As the transect angles away from the La Ronge Belt at site mt92-s28 (Fig. 1c), the strikes obtained from the GB decompositions fall back into the 25^O-35_^O range; this set includes sites s27, s41, s26, s44, s25, s24, s23, s22, s21, s20 and s19.

At the Manitoba/Saskatchewan border (Fig. 1d), the transect encounters a rapidly changing surficial geology from west to east, and the individual site decompositions become much more difficult. The unconstrained parameter plot for site mt92-s13 (Fig. 6a) shows a marked change in the behaviour of the GB parameters at 10 Hz. At frequencies higher than this, the twist and shear are almost frequency independent near 0_^O, indicating very little distortion At frequencies less than this they increase and decrease to their maximum and minimum physically-permitted values respectively. The strike shows almost no stability at all. A strike of 32^O is imposed on these data (Fig. 6b), as it was the strike obtained from site s23 to the west and was well fit. The GB parameters show better stability, but the shear is near its maximum value of 45^O over the entire frequency band which means that any strike can be fit to the data. Constraining the twist to -9^O yields the decomposition solutions in (Fig. 6c). At many neighbouring sites these high twists and shears predominate and make it nearly impossible to assign a strike without establishing a more general trend in the strike direction first. The sites affected in this way included s18, s16, s15, s13, s12, s11, s45, s46, s17, s10, s09, s14, s08, and s07 (Fig. 1d). These high decomposition parameters and high errors of misfit indicate that the distortion model is inadequate in this region. There are clearly 3D inductive effects at all frequencies which cannot be removed in the decomposition procedure.

Further to the east from this region, the GB decompositions become slightly easier with more reasonable twists and shears fit to most sites. For sites mt92-s01, s02, s03, s04, s05, s06, s48, s49, s50, s51, s52, s53, s54, s55, and s56 (Fig. 1e), the strikes obtained fall into one of two ranges. Sites nearer to the eastern end of this line have strikes which fall in the 15^O-35^O range, while sites closer to the extreme western end of the transect have higher strikes which fall in the 45^O-60^O range.

For example, site mt92-s06 shows considerable variability in an unconstrained plot of its GB parameters (Fig. 7a). The strike seems to fall in the 30^O-60^O range in the midband with a strike in the 10^O-30^O range suggested at lower frequencies, but it is very unstable. The twist is the most stable of the unconstrained parameters, and accordingly is constrained to a value of 2^O (Fig. 7b). The shear stabilizes in the 9-0.01 Hz band, and is constrained to a value of 3_^O (Fig. 7c). Finally, the strike stabilizes in the 80-0.01 Hz band, and it is constrained to its mean value over that band of 51^O (Fig. 7d).

For site mt92-s48 however, the unconstrained GB plots show a much better behaved strike which ranges between 15^O and 45^O in the 9-0.001 Hz band (Fig. 8a). The twist and shear increase steadily from a minimum value near zero in the 10,000-100 Hz band, to near 40^O at 0.01 s. The twist is constrained first to its mean value of 18^O (Fig. 8b). This stabilizes both shear and strike, and the shear is next constrained to 25^O (Fig. 8c). The strike has stabilized considerably from the first plot, and it ranges from 30^O-35^O over the 1-0.001 Hz band. It is constrained to a value of 32^O (Fig. 8d).

Strike directions from the decomposition of the AMT-only sites are given in Table 2. In general, the strikes obtained from these decompositions supported the trends established from the AMT-MT decompositions. For sites located at the eastern end of the transect there was great consistency between sites as the strikes initially fell in the 25^O-35^O range and then swung to the north at site amt-s32 to a value of 17^O. This trend continued for sites amt-s31 to amt-s26 (moving to the east) as the strikes obtained fell in the 5^O-20^O range. At site amt-s41, the strike rotated eastward again, and the strikes for sites amt-s41 - amt-s17 fell in the 30^O-45^O range. Over the central part of the transect, there is great variability in the strikes obtained from the AMT-only sites-- which was to be expected from the locally varied geological terrain. No consistent trends exist in the strike direction obtained from sites amt92-s15 - amt-s09. Towards the eastern half of the transect, AMT-only sites yielded strikes that fell in the 10^O-35^O range; sites at the extreme eastern end of the transect (amt-s51 and amt-s52) were fit with strikes of 69^O and 62^O respectively.

IVc. Preliminary Observations/Conclusions

After completion of the site by site decompositions, a strike that falls in the range 25^O-35^O is indicated for most of the sites in this transect. There are some suggestions of locally different strikes near the Needle Falls Shear Zone, and towards the western end of the transect, but it is possible that data at the shorter periods are dominating the fit in these areas. The data from the central part of the transect were almost impossible to fit with any degree of certainty due to the high twists and shears present in most of the decomposition models. Information gleaned from the multi-frequency, multi-site decompositions will have to be used to fix the strike at these sites.

To begin the multi-site, multi-frequency decompositions, the THOT92 dataset was broken into small groups of sites using the site by site strike information as a guide. The aim was to fit strikes in several frequency bands to each site in the group simultaneously. As more sites were added to each group, the resulting twists and shears obtained for each site were monitored closely to watch for extreme or maximal values-- this would indicate that a particular site did not contribute to the overall model and strike obtained. Multi-site, multi-frequency models which revealed high or extreme twists and/or shears at the majority of sites in the group required more information to fix the strike value.

The results for the multi-site, multi-frequency McNeice-Jones decompositions are summarized in Table 3. The filenames used as input to the "strike.old" program (McNeice and Jones, 1996) are listed in the first column of the table, with the sites comprising each group listed in the second.

In general, the first round of modelling produced strikes in accord with the regions established in the site by site decomposition. The western end of the transect was expanded from a small group of sites (thot92.lis1: s39, s38, s37, s36, s35) to a large group (thot92.lis2: s39, s38, s37, s36, s35, s40, s34, s33, s43, s32, s31). The strikes fit to the period bands 1-100 s, and 10 -1000 s are consistent between both groups; a strike of 30^O was obtained for both groups in the 1-100 s interval and a strike of 34_^O was obtained for both groups in the 10-1000 s band. Since we are looking for regional strike information over local information, we favour the fit at the longer periods and therefore 34^O is the indicated regional strike from the group of sites at the western end of the THOT92 transect.

The next series of fits reveals a region where the strike is locally rotated farther to the west-- near 20^O. Groups thot92.lis3 and thot92.lis5 show this pattern, with the strikes obtained from the latter group pointing at -15^O and -7^O for the 1-100 s and 10-1000 s band, respectively. The twists and shears fit to each site in the thot92.lis5 group were unacceptably high in both bands, however, so thot92.lis3 and thot92.lis5 were combined in an effort to produce a better fit to those sites. The union of these two groups is listed in the table as thot92.lis16; the strikes fit to this group are reasonably consistent between the 1-100 s band (26^O) and 10-1000 s (22^O). This latter strike value is consistent with the value used by Jones et al. (1993).

Moving farther to the east, the transect moves away from the exposed orogen and a small group of sites (thot92.lis7: s24, s23, s22, s21, s20, s19) was fit to a strike of 34.1^O in the 10-1000 s band. This provides further evidence that the regional strike for the western half of the transect is close to 34_^O, with the exception of those sites within inductive scale length of the NACP anomaly and which sense its strike direction (22_^O, Jones et al., 1993).

The central region of the transect posed problems for the multi-site fits as well. The thot92.lis8 group (s18, s16, s15, s13, s12, s11, s45, s46) was fit to a 33.6^O strike in the 10-1000 s band-- but most sites showed very high shears. Only sites s18 and s46 were fit to this strike with acceptably low twists and shears. These two sites therefore controlled the fit strike in this group. Still, the strike indicated by the fits to these two sites is in agreement with the information obtained from multi-site fits over the western half of the transect.

The multi-site, multi-frequency fits in the eastern half of the THOT92 transect reveal slightly different strike information than those groups in the western half of the transect. The group designated thot92.lis9 overlaps with thot92.lis8, and its strike fits were plagued by very high twists and shears in all bands. Further to the east, strikes of 19^O and 0^O were fit to the group thot92.lis10 (s09, s14, s08, s07, s01, s02) in the bands 1-100 s and 10-1000 s respectively; the majority of sites from this group contributed to the fit in these bands. These two groups were combined to form thot92.lis17, and the strikes obtained in these same bands were 15^O and 25^O respectively. Three to four sites out of the 13 sites in the group show high twists and/or shears in each band, but otherwise the strikes are well fit. The short period (100 Hz - 1 s) strike of 54^O indicates that many of the site by site decompositions produced the local, short period strike rather than the regional strike.

Group thot92.lis10 was also combined with group thot92.lis11 and strikes of 50^O, 19.3^O and 24^O were fit in the 100 Hz- 1 s, 1-100 s and 10-1000 s bands respectively.

Finally, at the transect's eastern extreme, thot92.lis12 (s49, s51, s52, s53, s54, s55, s56) was fit with strikes of 45^O, 25^O, and 29^O in the three period bands respectively. The twists and shears produced for the short period fit were all acceptably low; the result is further confirmation that the individual Groom-Bailey decompositions of these sites produced short period strikes. The twists and shears increased for many of the sites in the mid and long period bands, with the sites s53, s54 and s55 controlling the fit.

The Groom-Bailey analysis performed at each individual site showed that the regional strike for the western part of the THOT 92 transect fell in the 25^O-35^O range. Considerable uncertainty associated with the sites in the middle of the transect, where extreme twist and shears made it impossible to extract a strike using conventional site-by-site decomposition methods. Strikes in the 45^O-60_^O range near the eastern end of the transect fell outside the range established in the western part of the transect.

The multi-site, multi-frequency McNeice-Jones analysis clarifies the above results. A 34^O strike was obtained from a simultaneous fit of a large group of sites at the east end of the transect (thot92.lis2) in the mid and long period bands. A 34^O strike was also fit to a smaller group of six sites (thot92.lis7) in these same bands. While a localized group of sites was well fit to a strike of 22^O (thot92.lis) in the 10-1000 s band, the regional trend in the western half of the THOT92 transect is a strike of 34^O in the 1-1000 s period band. The decomposition of AMT-only sites in this region confirm the local trend.

In the eastern part of the transect, the fits to thot92.lis17 and thot92.lis15 suggest a strike between 19^O and 25^O -- with a slightly higher strike indicated at the eastern end of the site (thot92.lis12). A strike of 22^O is chosen for these sites, again in the 1-1000 s period band. AMT-only sites yield higher strikes in this region, suggesting that the higher strikes (45^O-60^O) obtained from site by site decomposition the eastern sites are due to the biasing of short period fits to the data.

The sites in the middle of the transect (from west to east, mt92-s18 to s09) were grouped with sites to the east and sites to the west, but very few sites from this central area appeared to be contributing to the overall fit in any of these groups. A 34^O strike was obtained for the thot92.lis8 grouping in the 10-1000 s band, but only two of eight sites contributed to the fit.

In order to deal with the uncertainty present in the assignment of a strike to the central sites, two models need to be created. One model should include, from west to east, all sites between mt92-s39 and mt92-s09. The data from these sites best fit to a distortion model with a constrained strike of 34^O. A second model should include all sites between site mt92-s18 and mt92-s56, which best fit a distortion model with a 22^O strike. Thus, there will be complete overlap in the central region of the transect and the two models should yield better resolution of this problem area.

Groom, R.W. and Bailey, R.C., 1989. Decomposition of magnetotelluric impedance tensors in the presence of local three-dimensional galvanic distortion. Journal of Geophysical Research, 94, 1913-1925. Groom, R.W., R.D. Kurtz, A.G. Jones and D.E. Boerner (1993). A quantitative methodology to extract regional magnetotelluric impedances and determine the dimension of the conductivity structure. Geophysics Journal International. 115, 1095-1118. Jones, A.G., J. A. Craven, G. W. McNeice, I. J. Ferguson, T. Boyce, C. Farquharson, R. G. Ellis (1993a). North American Central Plains conductivity anomaly within the Trans-Hudson orogen in northern Saskatchewan, Canada. Geology. 21, 1027-1030. Jones, A. G. and I. Dumas (1993). Electromagnetic Images of a Volcanic Zone. Physics of the Earth and Planetary Interiors. 81, 289-314. Jones, A.G. and Groom, R.W. (1993). Strike angle determination from the magnetotelluric tensor in the presence of noise and local distortion: rotate at your peril! Geophysical Journal International, 113, 524-534. Jones A.G., R.W. Groom, and R. D. Kurtz (1993b). Decomposition and Modelling of the BC87 Dataset. Journal of Geomagnetism and Geoelectricity. 45, 1127-1150. McNeice, G. and A.G. Jones, 1996. Multisite, multifrequency tensor decomposition of magnetotelluric data. Contributed paper at the 66th Annual Society of Exploration Geophysics meeting, held in Denver, Colorado, on 10-15 November. Expanded Abstracts, 66, 281-284.

Table 1: Site by site Groom Bailey strike directions for the amt-mt sites.

Table 2: Site-by-site Groom-Bailey strike directions for amt-only sites

Table 3: McNeice-Jones strike directions for multi-site, multi-frequency fits.

a) All sites

b) Westernmost sites over the Wollaston domain and Needle Falls Shear Zone

c) Sites across the Rottenstone and La Ronge domains

d) Sites by the Saskatchewan/Manitoba border

e) Easternmost sites over the Thompson Belt

Figure 3 Individual strike directions for the amt-only sites

Figures 4 Decomposition of the data for site mt-s39. For each sequence of 6 plots:

a) Free decomposition. No parameters constrained

b) Shear constrained to a value of 0_^O

c) Twist next constrained to a value of -2_^O

d) Strike finally constrained to a value of 34_^O

Figures 5 Decomposition of the data for site mt-s30

a) Free decomposition. No parameters constrained

b) Twist constrained to a value of 2_^O

c) Shear next constrained to a value of 7_^O

d) Strike finally constrained to a value of 12_^O

Figures 6 Decomposition of the data for site mt-s13

a) Free decomposition. No parameters constrained

b) Strike constrained to a value of 32_^O

c) Twist next constrained to a value of -9_^O

d) Shear finally constrained to a value of 30_^O

Figures 7 Decomposition of the data for site mt-s06

a) Free decomposition. No parameters constrained

b) Twist constrained to a value of 2_^O

c) Shear next constrained to a value of 3_^O

d) Strike finally constrained to a value of 51_^O

Figures 8 Decomposition of the data for site mt-s48

a) Free decomposition. No parameters constrained

b) Twist constrained to a value of 18_^O

c) Shear next constrained to a value of 25_^O

d) Strike finally constrained to a value of 32_^O