You are on page 1of 17

DIURNAL CORRECTIONS TO MAGNETIC SURVEYS-AN

ASSESSMENT OF ERRORS *

BY

R. P. RIDDIHOUGH **

ABSTRACT
RIDDIHOUGH, R. P., 1971, Diurnal Corrections to Magnetic Surveys-An Assessment of
Errors, Geophysical Prospecting 19, 551-567.
The comparison of a group of continuous total field stations in Ireland with the record
of the total field from Valentia Observatory enables the errors which would be involved in
using the Observatory as a diurnal correction base to be examined. For the most part,
for a single correction, these errors lie between * z and f 6 gammas. However, they have
three notable characteristics : (i) they are not closely dependent on the magnetic ‘noisiness’
of the day, (ii) they are only marginally reduced by the application of station-base time
differences and (iii), they are strongly related to differences between the amplitude of the
daily magnetic variation at the station and at the base. A permanent geographical basis
for the latter differences enables a system for the estimation of the errors to be proposed.

I. INTRODUCTION
The use of diurnal corrections in magnetic surveys is dictated by the need
to separate the temporal variations of the earth’s magnetic field from its
geographical variations. This is done by two main methods, either a self-
checking system as in the rapid ‘looping’ of aeromagnetic surveying, or the
use of a static base station which continuously records variations of the mag-
netic field against time. This base station is very often located as near as
possible to the area of the survey. However, equally often, it may be up to
IOO or zoo km away and in this case it is essential to the reduction of the surveys
that the errors involved in applying the magnetic variations at the base station
to the survey area are realistically assessed. This is seldom done. The present
paper attempts such an assessment for the reduction of ground total field
magnetic surveys in Ireland in the hope that this may provide a useful example
for surveys in other areas.

2. BACKGROUND
The main guiding lines in the application of diurnal corrections have been
provided by world wide studies of the earth’s magnetic field (see Chapman
* Received October 1970.
** Dublin Institute for Advanced Studies, Dublin, Ireland, presently at: Geomagnetic
Div., Earth Physics Branch, Dept. Energy, Mines and Resources, Ottawa, Canada.
Geophysical Prospecting, Vol. XIX 3‘5
552 R. P. RIDDIHOUGH

and Bartels 1940). These established that the major part of the daily variation
of the earth’s magnetic field is related to the effect of the sun on the earth’s
ionosphere exposed to it and is therefore strongly dependent upon local
time. Disturbance or storm variations however, have a separate origin and are
more directly related to activity on the sun. In a time sense, they are indepen-
dent of the orientation of the earth relative to the sun and can be regarded as oc-
curring simultaneously over the whole earth’s surface. With regard to surveys,
the practice has been to avoid disturbed days as much as possible, and for
the rest to consider that the variation at the nearest observatory can be applied
to the survey area with a correction for local time.
Detailed discussion of the effect of this method on the reduction of surveys
has been limited, although in the British Isles the problem was recognised by
Walker as early as IgIg (Walker IgIg). An important paper on the subject
is that of Whitham and Niblett (1961) which investigated the differences
between two continuously recorded stations in Canada and attempted their
estimation in terms of the level of magnetic activity and the distance apart
of the stations. Sucksdorff (1968) considered in some detail the errors produced
by various methods of correcting for daily magnetic variation in Finland.
He showed that in Finland, an interpolation method between two observatories
either side of the station improved reduction errors over those from a single
observatory when the latter was more than IOO km north or south of the
station. A recent paper, Le Borgne and Le Moue1 (Ig6g), outlined the reduction
procedures involved in the production of the Carte Aeromagnetique de la
France and discussed the errors resulting from the projection of diurnal
corrections over distances of up to 300 km.
The reduction problems of marine magnetic surveys have received more
attention because cross-overs cannot be made rapidly enough to provide suit-
able checks and because of the technical difficulties of maintaining buoyed
local base stations. The characteristics of a daily magnetic variation at sea
were investigated by Hill and Mason (1962) in the English Channel area, sub-
sequently Roden and Mason (1964) discussed the prediction of the variation
at a point between two widely separated observatories using data from the
Indian Ocean. They concluded that a suitably weighted average of the two
stations provided the most valuable estimate of the variations at the inter-
mediate point. A more recent analysis of similar problems in the reduction
of surveys in the Gulf of Aden is that of Whitmore and Jones (1969).
By far the greatest amount of relevant literature however is concerned not
directly with survey correction, but with the differing characteristics of the
magnetic field variations recorded at one place as compared with those at
another, and their interpretation in terms of induction and conduction in the
earth’s crust and mantle. On a wide scale this has led to considerable knowledge
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 553

of the earth’s structure in depth (e.g. Banks 1969). In more local areas and
often in terms of periods of magnetic variation of a few hours and less, consid-
erable anomalies have been discovered. A useful review of these is given by
Rikitake (1966, Chaps 17, 19).
Within the European area, the North German anomaly has received much
attention and shows dramatic changes in the recorded vertical field variations
over distances of the order of IOO km. Over the British Isles, a comprehensive
investigation of field variations for periods of 3 hours and less has been reported
by Edwards and Law (1970). In some areas they found similar large changes
over small distances and they have interpreted their results in terms of induced
electric currents flowing either within or below the adjacent shallow seas and
ocean.
The present author, in two previous studies of the geographical variability
of the magnetic daily variation over the British Isles and NW Europe (Riddi-
hough 1967, 1969) has tried to show that not only are there departures from
a local solar time relationship but that there are also systematic and significant
changes in the amplitude of the variation as it ‘moves’ westwards.

3. METHODS
The arithmetic calculation, which forms the basis of the elimination of daily
magnetic variation in surveys with a base monitoring station, is the subtraction
of the field value at the base station from that at the survey station. With
‘local time’ assumptions, the result of this is normally taken to be as near a
‘true’ static value for the survey station relative to the base as is possible.
However, if the daily variation does in fact alter in amplitude and timing over
short distances, then any single subtraction will be subject to an error which
will be related to any such alterations in the variation between the survey
area and the base station. The characteristics of this type of error can be seen
by reference to a model.

3.1. Theoretical Model


From the author’s previous studies of the daily magnetic variation (Riddi-
hough 1967, 1969) it appears that in Ireland time differences in occurrence
of the order of 20 minutes and amplitude differences of up to 20% could be
expected across the country. In fig. I therefore, a daily variation curve of the
total field (All Days, Equinoctial Months, Valentia 1966) is taken and compared
with a hypothetical series of ‘stations’ at which the same curve occurs at
different times and amplitudes. At each hour on the primary ‘base’ curve, the
arithmetic difference between this curve and the ‘station’ model is taken.
This results in 24 values of the difference between each ‘station’ and the ‘base’.
The error between any one of these 24 differences and the true difference
554 R. P. RIDDIHOUGH

(represented for practical purposes by the mean of the 24 and in this case
virtually equal to zero) can be expressed by the standard deviation of these
values around their mean. As this is the limit within which 68% of the readings
lie from the mean, so it is also the limit within which the true mean value will
lie from any single one in 68 out of IOO cases. It is in fact the standard error
of an individual difference, called SE in the subsequent text.

amplitude
%
Fig. I. Model study of time and amplitude
A-Daily variation

reductions
(thick line) with 20%
(thin line), vertical intervalsiro gammas.
(%). C-Correlation
as a function
C
changes in daily total field magnetic variation.
amplitude
B-SE
reduction and 20 minute time shift
(gamma) for time-shifts (minutes) and
coefficients for time-shifts. D-SE
of RD (gamma).
(gamma)

Time differences of o, IO and 20 minutes before and after the ‘base’ for o, IO
and 20% amplitude reductions are considered, the results also being shown
in fig. I. The graphs show that SE is a simple function of both the time and
amplitude differences. For time differences, they show that there is a regular
and identifiable minimum indicating the position of closest fit of the two curves.
However, the graphs also show that the postulated maximum amplitude
difference (20%) produces greater values of SE than the postulated maximum
time difference (20 minutes) alone, and that with an increasing amplitude
difference, the time differences have a lessening effect on the range of SE.
The predominance of the amplitude controls is also shown in fig. 1 D where
SE is plotted against the actual difference in gammas between the mean
daily ranges of the ‘base’ and ‘station’ curves (mean daily range is taken both
here and subsequently to be the difference between the hourly mean value
centred at 1230 GMT and that at 1830 GMT and as such is an arbitrary but
representative measure of the actual range of the daily variation).
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 555

This model calculation thus shows that a variation in the broad charac-
teristics of the daily variation as suggested could produce errors in the present
practical context of up to 5 3 gammas. These errors would be a function of
the relative time of occurrence and amplitude of the daily magnetic variation
at the station and at the base concerned.

3.2. Data and Treatment


For the present study, 43 separate days of total field proton magnetometer
measurements at a series of stations in Ireland are compared with the total
field as deduced from the H and Z magnetograms recorded at Valentia Obser-
vatory. The availability of continuous recording equipment meant that for
the most part only one station and Valentia were recorded on any one day.

DUBLIN 3.10

- IOOKM

Fig. 2. Location Map. Number of days recorded followed by month of year.

The days were spread over the years 1967-70 and are concentrated within the
summer months; however, their distribution of K, indices was similar to
that of the whole year at Valentia with a regular peak at approximately
C K, = IO. The location of the stations is shown in fig. 2 together with the
number of days recorded and the months in which they were recorded. Fuller
details of the days and the results of their analysis are being published in a
Geophysical Bulletin of the Dublin Institute for Advanced Studies.
For each day concerned, the values of the total field at Valentia were cal-
culated at 0005, 0015, 0025 . . . 2355 GMT giving 144 values of the total field.
For the station to be compared, values were also taken at ten-minute intervals
but from 2235 of the previous day to 0125 of the following day, giving a total
of 162 values. These were then compared in a series of calculations which
correlated the 144 values at Valentia with the 144 station values beginning
at 2235, those beginning at 2245, at 2255 . . . etc. up to those beginning at
556 R. P. RIDDIHOUGH

0135; that is, rg separate calculations. This process was essentially one of
‘sliding’ the curve for Valentia alongside that for the station so that the varia-
tion in correlation parameters could be seen. The parameters calculated for
each of the rg positions were:

(i) the standard deviation of the arithmetic differences between the 144
pairs,
(ii) the linear correlation coefficient between the two data series.
(iii) the linear regression coefficients between the two dataiseries.
A

Fig. 3. Comparison of Rosslare and Valentia total field. A--Plot of ten-minute values
with 20 minute time difference (scale intervals IO gamma). B-correlation coefficients and
C-SE (gamma) for varying time differences (minutes).

The process was in fact the basis of previous work on the geographical
pattern of the daily magnetic variation (Riddihough x969), parameter (ii)
producing an indicator for estimating the time difference between the varia-
tions at two stations and (iii) producing the amplitude ratio of two variations.
In the previous work, a smoothed version of the daily variation curves was used
for the calculations but omission of the smoothing seems to produce no signif-
icant difference in the results.
Typical results for one day are detailed in fig. 3, a comparison between
Rosslare and Valentia on 8.7.69. Fig. 3 A is a graph of the data points for the
comparison of the Valentia data set beginning 0005 with the Rosslare set begin-
ning 2345, that is, with a 20 minute time difference. It shows the strong linear-
ity expressed by the correlation coefficient of 0.9814 for this group. In fact,
all the days considered showed a highly significant 1inear”correlation between
the station and Valentia, the lowest value of maximum correlation coefficient
for any one day being 0.9081 for the 144 points.
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 557

Figs. 3 B and 3 C are parts of the graphs of correlation coefficient and


standard deviation against the rg time positions and show regular and iden-
tifiable maxima and minima exactly analogous to those of fig. I, as do the
results for the rest of the 43 days considered. The positions of the correlation
coefficient and standard deviation turning points are again coincident. The
similarity of the model and real analysis emphasized in the comparisons of
figs. I B and I C and figs. 3 B and 3 C indicates that the type of variation
envisaged in the model of 3.1 is probably valid for the real case.
The values which are deduced and used from the calculated parameters
for each day are:

(a) the standard error (SE) of a single reading of the difference between the
value at the station and at Valentia for a range of time differences.
(b) the time difference between the occurrence of the daily magnetic varia-
tion at Valentia and at the station (identified from the position of the maxi-
mum correlation coefficient)
(c) the amplitude ratio (mean of the two linear regression coefficients)
between the daily variation at the station and at Valentia.

The calculation of (a) assumes that the true difference between the station
and Valentia can be represented by the mean of the 144 differences of the
24 hours considered. As the day is the basic unit of the magnetic variation
being considered and 144 values can be regarded statistically as a large sample,
this was a valid theoretical approach. However, in detail there is some varia-
bility in the mean differences between stations from one day to the next,
this is commented upon further in 5.

4. RESULTS
4.1. Time Relations
The time differences between Valentia and any one station, which it must
be stressed are merely the time differences in the above calculations which
are characterised by the closest fit between the two daily variation curves,
show in some cases considerable day to day variability (up to 20 minutes
between one day and the next) and in others reasonable consistency. There
is no apparent relationship between this and the ‘noisiness’ of the day as
measured by the K, index or with the correlation coefficients between Valen-
tia and the station.
The mean values of the time differences are shown in fig. 4 A and can be
seen to demonstrate a general westward movement pattern, although the
contours as drawn contrast with the maps of Riddihough (1969) for single
558 R. P. RIDDIHOUGH

days. It is notable that the field measured at St. Finan’s Bay, only 15 km
from Valentia, has a significant time difference of 8 minutes for the period of
study.
A 6

I 100 KM

Fig. 4. Daily magnetic variations in Ireland. A-mean values of time relationships with
Valentia (negative, minutes before Val., positive, minutes after). B-mean values of
amplitude relationships with Valentia (Val. IOO units).

4.2. Anqblitude Relations


Values of relative amplitude were taken at the maximum correlation position
between Valentia and the station; however, positions up to 30 minutes either
side of this were less than 1% different from this value. The values themselves
at any one station show a day to day variability with again no apparent depen-
dence upon either the noisiness of the day or the actual amplitude of the varia-
tion. However, the map drawn on the basis of the average values (fig. 4B)
gives a picture of a steady increase towards the west coast such as suggested
by the .larger scale studies. It does therefore appear likely that this could
be a reliably permanent feature of the daily variation in Ireland. It is again
notable that the amplitude measured at St. Finan’s Bay is consistently larger
than that at Valentia.

4.3. Standard Errors


4.3.1. Influence of ‘Noisiness’ of the Day.
The minimum values of SE between the stations and Valentia vary between
1.4 and 19.8 gammas and are shown in fig. 5 plotted against the K, index
sums CK.p and ZCKi for the days concerned. These sums can be generally
taken to Indicate the ‘noisiness’ of the day in a magnetic sense over and above
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 5.59

the main solar daily variation. Mayaud (1969) has discussed the significance
of indices in this respect at some length and concludes that an essentially
disturbed day, in a real rather than a formally classified manner, has A, > 20.
The equivalence of this in terms of K, is by no means fixed but may be con-
sidered as approximately CKp > 24 and C Kg > gz.
The present group of days indicate that for values of XK, < 20 and of
XKi < 60 there is a large group for which there is no clear relationship between
‘noise’ level and SE. This at first sight conflicts with the conclusions of Whit-
ham and Niblett who deduced a quantifiable relationship between magnetic
activity and the differences between stations. However in their work they

50- .+b
200’
~.HP CK$
.* .+

Fig. 5. Graphs of daily SE (gamma) against K, indices.

used not K,, but an activity level derived from the deviation within an hour
from the hourly mean of that hour and which would thus appear to include
both ‘noise’ and part of the main solar daily variation. The extent to which
the K, index succeeds in separating out the effect of the solar daily variation
from the superimposed ‘noise’ is debatable, however to the limits that it does,
the graphs of fig. 5 do suggest that below XK, = 20 and X;Ki = 60, there
is little correlation between ‘noise’ and SE.
The few points remaining above this level suggest that as K, increases
considerably, so also will the minimum SE, However, one point with SE =
19.8 gammas conflicts with any relationship indicated by the remainder of
the points. Although this point corresponds with neither the largest values
of XK, nor XK;, it does correspond with a day having an extreme single
K, value of 7 and two subsequent values of 6 + and 6. This applied in Ireland
to a group of deep embayments between oooo and ogoo GMT on 27.7.69 which
produced negative departures in the total field of around 160 gammas.
560 R. P. RIDDIHOUGH

4.3.2. Influence of Amplitude of Daily Variation


With these considerations, the further analysis of SE is now considered
with the points starred in fig. 5 omitted. Fig. 6 shows the remainder of the
points in a graph relating the mean daily Range Difference (RD) between
the station and Valentia to the values of minimum SE for the day. RD is
derived by multiplying the difference in the amplitude relationship between
the station and Valentia for the day by the actual mean daily range (defined
in 3.1) of Valentia on that day. It thus represents the difference in mean
daily range between Valentia and the station.

I 234567 SE MIN

Fig. 6. Graph of RD (gamma) against minimum SE (gamma) for non-storm days. Dash-
ed line is from the model regression of fig. I D, solid and dotted lines are regression of SE on
RD with 95% confidence limits.

On the same graph is plotted the theoretical line derived from the model
analysis of fig. I with the addition to the SE values of the latter of a funda-
mental error of 1.5 gammas. This, as indicated by the analysis of results from
a total field magnetometer run at a distance of IOO m from the Valentia Obser-
vatory variometers for three days, can probably be taken as the minimum
error resulting from magnetogram copying and reduction methods.
Clearly from fig. 6, the relationship between RD and minimum SE is signif-
icantly linear, the correlation coefficient for this graph being 0.7751. It thus
appears to be analogous to the model of fig. I and to confirm that errors are
primarily related to differences in daily variation occurrence, that to a large
extent the controlling factor in the development of errors between a station
and Valentia is the varying amplitude relationship between the daily variation
as measured at the station and that measured at Valentia.

4.3.3. Influence of Time Differences


The influence of the time relationship between a station and Valentia can
be seen in fig. 7. This diagram shows the regression lines of SE as a function of
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 561

RD when the curves of daily magnetic variation are considered at a fixed time
difference for each station. This will not necessarily be the time difference
corresponding to the minimum value of SE which-as mentioned in 4.1--is
different from day to day. However, in practical terms, a fixed time difference
for each station can provide the only viable basis to a reduction procedure.
The lines are shown together with the boundaries of the standard error
of estimate of SE and it is seen that although there are changes in the graphs

Fig. 7. Regressions of SE (gamma) on RD (gamma) with standard errors of estimate of


SE for fixed time differences on non-storm days. A-mean time difference, R-local solar
time difference, C-zero time difference.

(the correlation coefficients for mean time diff., solar time diff., and zero
time diff. are 0.7439, 0.7420 and 0.7507 respectively) there is clearly only
a marginal difference between taking account of estimated mean (given from
the data of fig. 4 A) or local solar time differences and considering Valentia
as simultaneous to the station. The mean SE of the present groups changes
from 4.057 f 0.136 t0 4.114 f 0.134 t0 4.197 f 0.140 respectively. There iS
a weakening of the correlation from the data of fig. 6. However, in terms of a
situation in which the SE of survey work on a particular day needs to be esti-
mated to the nearest gamma, the differences between the graphs are negligible.
In practice this is very important because it means that applying a local,
562 R. P. RIDDIHOUGH

or indeed any, time correction for the position of the station is unnecessary
in the sense that the effect of such a correction on the reduction of errors is
only slight. The only effective reduction could come from a full day analysis
as done here, and even then the error from this source would be less important
than that arising from the differences in the amplitudes of the daily magnetic
variation between stations.

4.3.4. Fixed Time and Amplitude Relationships.


Finally, to bring these regressions into their most practical form, RDm-
a mean value of RD-is plotted in fig. 8 against SE at zero time difference.

Fig. 8. Graph of RDwz (gamma) against SE (gamma) for zero time difference on non- storm
days. Regression of SE on RD with standard error of estimate of SE shown. Dashed line
is regression from fig. 6.

I 234567
STANDARD ERRORS
OAMMAS

Fig. 9. System of estimation of errors for Ireland on non-storm days. Map shows amplitude
factors (contours) leading to RD. Graph shows SE values with 68% prediction intervals
for values of RD. For use see text.
ERRORS IN DIURNALMAGNETIC CORRECTIONS 563

RD, is a modification of RD which involves not the unique amplitude rela-


tionship between a station and Valentia on the particular day concerned, but
the mean amplitude relationship for the station derived from the group of
days considered. In this case it is the mean derived with the starred days of
fig. 5 omitted (shown against each station in the map of fig. 9). This mean
amplitude relationship for the station is combined with the actual mean daily
range at Valentia on the day in question to give RD,. Its plot against SE
at zero time difference in fig. 8 can be seen to be still strongly linear (correla-
tion coefficient 0.7003) and very little different from the most accurate ana-
lysis of fig. 6 represented by the dashed line.

5. PRACTICAL ESTIMATION OF ERRORS


The results can now be collated to provide a practical system for the esti-
mation of the errors of ground surveys using Valentia as the diurnal correction
base. Any such system is of course limited by the 43 unique days which have
been studied in the present case. However, as merrtioned in 3.2. their ‘noisiness’
(considered as a function of K, index) is not unrepresentative of the distri-
bution of a whole year in Ireland and although the days are concentrated in
the summer months, the SE and RD values for two mid-winter days (25,
26.1.67) lie neatly within the distributions that are demonstrated.

The information needed for the estimation of errors on this basis is thus:
(a) An evaluation of the magnetic ‘noisiness’ of the day concerned, either
by K, index or experience. In fact all that is necessary is to decide whether
the day is really disturbed or merely average and below.
(b) The position of the station.
(c) The mean daily magnetic range of Valentia in the total field (the differ-
ence between the hourly means centred at 1230 and 1830 GMT) for the day.
This will very often have been calculated in the reduction of magnetograms
carried out for the diurnal correction.

If the ‘noisiness’ of the day is such as to place it in the categories with


IZK, > 20 and ZKi > 60 then from the data of fig. 5 it seems that the stand-
ard error of differences between the station and Valentia can be f6 gammas
and greater. At this level of disturbance it will not be possible to predict the
error with any certainty and extreme errors such as that noted here for 27.7.69
at *rg.8 gammas may undoubtedly occur. For most practical purposes, given
a choice, these days should be avoided.
For the majority of days, for which activity levels will be less than CK, = 20
and CK; = 60, fig. g can be used to give an estimate of the errors to be
564 R. P. RIDDIHOUGH

expected. Taking firstly the position of the station, the contours of the map
of fig. g will give an amplitude factor which is essentially the proportionate
difference between the daily magnetic variation at the station and that at
Valentia. The map is constructed from the mean amplitude relationships
for the 3g days (shown against each station, Valentia IOO units) which formed
the basis of the calculation of RD, for fig. 8. The amplitude factor given by
the contours is then multiplied by the mean daily range in total field for that
day at Valentia to give a figure in gammas, the Range Difference RD.
The graph of fig. g is now entered at this value of Range Difference to give
the Standard Error to be expected for readings taken on that day at that
station and reduced with the Valentia magnetograms. The prediction or
confidence range around the solid line of this graph is calculated from the
68% confidence limits of the regression used (fig. 8) and the standard errors
of estimate of SE. Because of the linear spread of the points considered, it is
virtually identical to the latter.
Thus a hypothetical survey on the east coast of Ireland on a day on which
the mean daily total field range at Valentia was 50 gammas, would have a
Range Difference value of IO gammas and thus a Standard Error due to diurnal
correction of * (5 f I) gammas.
As referred to in 32, the foregoing calculations are based on the assumption
that the mean value of the 144 Valentia-station differences for the day can
be regarded as the true value of the difference. This is not entirely correct
because the main shape of the daily variation is essentially a negative embay-
ment of the total field and the calculated mean value of the day will be affected
by its amplitude. Differences between the amplitudes of the variation at
Valentia and at the station will thus effect a slight change in the relative
daily mean values. From the stations and days considered, the standard
deviation of the daily mean differences is approximately I gamma. The final
values of Standard Errors deduced from fig. g could therefore be marginally
increased by between 0.01 and 0.2 gammas to give a more rigorous answer.
While this is, in some sense, an unduly sophisticated estimate of the reduction
errors, which could in most cases be dismissed as being less than f5 gammas,
the fact that on the present basis this can now be said with some confidence
and indeed with measurable confidence, is of extreme value.

6. CONCLUSIONS

Considering those days which are not essentially disturbed, earlier wide
scale studies (Riddihough, Ig67,rg6g) had attempted to establish the concept of
systematic patterns in occurrence of the daily magnetic variation as it was
recorded across Europe and the British Isles. The influence of this on the reduc-
tion of magnetic surveys by the use of diurnal monitoring from a distant station
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 565

would of course be a function of the relationship between the survey area and
the base station both in respect of the time of occurrence and amplitude of the
daily magnetic variation. For the series of stations and days in Ireland which
have been compared with Valentia, this seems to be true and to confirm that
the concept of systematic geographical patterns in the daily magnetic variation
is applicable.
In detail, while both the time of occurrence and amplitude affect the errors
involved in reduction, the effect of the time variations which are observed
in Ireland is much less than that of the amplitude variations. This is true to
the extent that in the absence of any method of removing the errors entirely,
applying a time correction, i.e. applying the daily variation at the base station
to the survey stations with a lag or an advance of a fixed number of minutes,
produces no significant improvement in the final results. Therefore to reduce
the amount of work necessary, the base station daily variation observations
may be applied as if they occurred simultaneously in the survey area.
The errors introduced by differences in amplitude of the daily magnetic
variation are subject to considerable day to day change but are seen to be
strongly related to the actual difference between the range of the variation at the
station and the range at Valentia on the day in question. The consistent geo-
graphical pattern of amplitude indicated by both the present and previous studies
enables an estimate of this range difference to be made for each day on the
basis of the range at Valentia alone. This in turn leads to a system whereby
the standard errors of subtractions of base station values from survey station
values can be estimated.
It was initially hoped that these studies might lead to a system of removal
of the errors due to amplitude differences; indeed it is conceivable that the
amplitude of the daily magnetic variation at Valentia could be mathematically
reduced or expanded to more closely resemble the diurnal which is assumed to
occur at the survey area. However, this process would involve the reduction
of complete 24 hour periods of magnetograms and the calculation of the daily
mean value about which the reduction could be made. While this would be an
interesting academic exercise, it would be unlikely to reduce errors by more
than 3 gammas and then only to the point where the error due to magneto-
gram reading, copying etc. appears in practice to be of the order of 2-3 gammas
(see zero RD values in fig. 6). For normal survey procedures it is clearly not
a practicable proposition.
While the present study has been confined to Ireland, it would, in conjunc-
tion with Riddihough (1967, 1969) seem to indicate three important pointers
for magnetic survey diurnal reduction in comparable areas:

(a) The use of a time correction for local time may not significantly reduce
the errors of diurnal corrections and may therefore be uneconomic of effort.
566 R. P. RIDDIHOUGH

(b) The predominant source of errors for days which are not magnetically
disturbed is likely to be the difference in the amplitude of the daily magnetic
variation between the survey area and the base station. The errors from this
source are not easily susceptible to removal but may be estimated on the basis
of a permanent pattern of differences of daily magnetic range and amplitude
between the two points.
(c) The amplitude and range differences of daily magnetic variation, though
geographically distributed, are not purely related to distance. Their pattern
is likely to be permanent and determinable and is probably related to geolog-
ical and structural features on a large scale, e.g. conductivity anomalies in
the crust, oceans, deep sedimentary basins etc.

7. ACKNOWLEDGEMENTS
I would like to thank the Director of the Meteorological Service, Dept. of
Transport and Power, Dublin for permission to use Meteorological Stations
as recording sites and to express my appreciation of the assistance and kindness
shown by the staff of the stations at Belmullet, Clones, Rosslare and Valentia
Observatory during the recording of the observations. My thanks are also due
to Prof. T. Murphy and the staff of the Geophysical Section of D.I.A.S. for
assistance on innumerable occasions with the supervision and maintenance
of the magnetometers and recorders.

8. BIBLIOGRAPHY
BANKS, R.. J., 1969, Geomagnetic Variations and the Electrical Conductivity of the Upper
Mantle. Geophys. J. 17, 457-487.
CHAPMAN, S. and J. BARTELS 1940, Geomagnetism. Clarendon Press, Oxford.
EDWARDS, R. N. and L. K. LAW 1970, Magnetic Variations in the British Isles, Abstract
of paper presented at IAGA Symp. Madrid. 1969. J. Geomagn. Geoelectr. zz, p. 71.
HILL, M. N. and C. S. MASON 1962, Diurnal variation of the Earth’s Magnetic Field at
Sea, Nature 195, 4839, 365-366.
LE BORGNE, E. and J. LE MOUBL 1969, La Reduction des Observations et la Precision des
Leves Aeromagnetiques de la France Continentale et de la Mediterranee Occidentale,
Ann. Geophys. 25, 371-380.
MAYAUD, P. N. 1969, A Propos de la Classification des Jours Magnetiques Internationaux,
Ann. Geophys. 25, gor-922.
RIDDIHOUGH, R. P. 1967, Daily Variation of Total Magnetic Field over the British Isles,
Nature 215, 5102, 720-722..
RIDDIHOUGH, R. P. 1969, A Geographical Pattern of Daily Magnetic Variation over North-
West Europe, Ann. Geophys. 25, 739-745.
RIKITAKE, T. 1966, Electromagnetism and the Earth’s Interior. Elsevier, Amsterdam.
RODEN, R. B. and C. S. MASON 1964, The Correction of Shipboard Magnetic Observations,
Geophys. J. g, 9-13.
SUCKSDORFF, C. 1968, Measurements at Geomagnetic Secular Stations in Finland, rg45-
1967. Finnish Met. Institute-Studies in Earth Magnetism, No. 20.
WALKER, G. W. rgrg, A Magnetic Re-Survey of the British Isles for the Epoch, Jan. I,
1915, Phil. Trans. R. Sot. A 219, 1-72.
ERRORS IN DIURNAL MAGNETIC CORRECTIONS 567

WITHAM, K. and E. R. NIBLETT, 1961, The Diurnal Problem in Aeromagnetic Surveying


in Canada, Geophysics 26, 211-228.
WHITMARSH, R. B. and M. T. JONES 1969, Daily Variation and Secular Variation of the ”
Geomagnetic Field from Shipboard Observations in the Gulf of Aden, Geoph. J.
18, 477-488.

Geophysical Prospecting, Vol. XIX 37

You might also like