experience with a digitally recording magnetometer system at wingst geomagnetic observatory...

Experience with a Digitally Recording Magnetometer System at Wingst Geomagnetic Observatory (Erdmagnetisches Observatorium Wingst)

Gfinter S c h u i z

UDC 550.380.87

Summary

At the Erdmagnetisches Observatorium Wingst of the Deutsches Hydrographi- sches Institut, a computer controlled magnetometer system has been recording the Earth's magnetic field since 1980. Primary elements are three fluxgates, a proton magnetometer (PRM), and two thermometer probes. In our case, the PRM does not just act as a part of the recording system, but also as the indicator of the absolute proton vector magnetometer (PVM) installed on the observatory's main pier. Thus, the calculation of the base-line measurements can be carried out on-line to a large extent by the computer. Moreover, this configuration permits the scale values and the temperature coefficients to be easily controlled, and any residual crosstalking to be detected by operating the PVM as a reference variometer of a higher class.

In this paper, the aforementioned system is described, the fluxgates' main parameters - the base-lines' short-term and long-term stability, the certainty of the scale values, the quality of the probes' orthogonal adjustment - are demonstrated and their limiting values are discussed.

Erfahrungen mit einem digital registrierenden Magnetometersystem am Erdmagnetischen Observatorium Wingst (Zusammenfassung)

Am Erdmagnetischen Observatorium Wingst des Deutschen Hydrographischen Instituts wird seit 1980 ein rechnergestfitztes Magnetometersystem zur Registrierung des Erdmagnetfeldes betrieben. Primfire Megwertaufnehmer sind drei Fluxgates, ein Protonenresonanz-Magnetometer (PRM) und zwei Temperatursonden. Das PRM dient hier nicht nur als Teil des Registriersystems, sondern auch als Indikator des absolut messenden Protonenkomponenten-Magnetometers (PVM). Dieses Ger/~t steht auf dem Hauptpfeiler des Observatoriums; daher k6nnen die Basismessungen zum grogen Teil fiber den Rechner in Echtzeit ausgewertet werden. Darfiber hinaus lassen sich mit dieser Konfiguration die Skalenwerte und Temperaturkoeffizienten bequem bestimmen. Setzt man das PVM als Variometer h6herer Klasse ein, so k6nnen Orientierungsfehler der Fluxgates ermittelt werden.

In diesem Aufsatz wird das System beschrieben, die Kenngr6Ben des Fluxgate- magnetometers, nfimlich die kurz- und langfristige Stabilitfit der Basen, die Sicher- heit der Skalenwerte und die Gfite der orthogonalen Ausrichtung der Sonden werden dargelegt und ihre Grenzen diskutiert.

174 Dt.hydrogr.Z.36, 1983. H.5. S c h u l z , Experience with a Magnetometer System

Exp6rimentation d'un magn4tom~tre enregistreur digital ~ l'observatoire g4omagn6tique de Wingst (Erdmagnetisches Observatorium Wingst) (R6sum6)

Au Erdmagnetisches Observatorium Wingst du Deutsches Hydrographisches Institut, un syst6me magn6tom6tre s 'appuyant sur un ordinateur pour l'enregistre- ment du champ magn6tique de la terre est en service depuis 1980. Les principaux capteurs de mesure en sont: trois inducteurs ~t noyau saturable, un magn4tom6tre r6sonance protonique (PRM), et deux sondes thermom4triques. Dans le cas pr4sent, le PRM n'intervient pas uniquement en tant qu'616ment de l'enregistreur, mais aussi en tant qu'indicateur du magn6tom6tre vectoriel fi proton (PVM), instrument de mesure absolue. Cet appareil est plac6 sur le pilier principal de l'observatoire. Aussi, le traitement des mesures de la base peut-il atre effectu4 en temps rdel, dans une large mesure, par l'ordinateur. De plus, cette configuration permet de contr61er facilement les 6chelles de valeurs et les coefficients de temp4rature, et de ddterminer toute erreur d'orientation des inducteurs fi noyau saturable, en utilisant le PVM comme variom6tre de rdfdrence de classe sup6rieure.

Dans le pr6sent article, on ddcrit le syst6me ~ savoir, les principaux param6tres des inducteurs 5 noyau saturable, la stabilit6 g court terme et fi long terme des bases de mesure, la fiabilit6 des 6chelles de valeurs, la qualit4 des ajustements orthogonaux des sondes sont d4montrds et leurs valeurs limitatives discut4es.

1 Introduction

As a rule, modern stationary digitally recording magnetometer systems do not just include the registration of three orthogonal components of the Earth's magnetic field, but also its total intensity by means of a proton magnetometer (PRM). The advantage is quite obvious: the magnetic vector recorded is over-determined, the time behaviour of the closing error permits - at least partially - the variometer's base-lines to be permanently controlled. Moreover, the system's redundancy permits outliers of the readings to be detected and eliminated ( D e l a u r i e r , L o o m e r , J a n s e n v a n B e e k , etal . [1974]).

At the Erdmagnetisches Observatorium Wingst, a fluxgate magnetometer of the type FM 100 C (EDA), purchased in 1978, forms the main part of a system of which the following instruments form further parts (Fig. 1): a. A proton vector magnetometer (PVM) for the purpose of measuring the base-line

values of H and Z according to the compensation method, and D (relative) according to the addition method (S c h u 1 z [1981]). This instrument (V o p p e 1 [1972]) consists of - two Braunbek coils mounted on a theodolite (Askania), - a highly stable DC current source (Heinzinger), and - a frequency measuring proto_u magnetometer (PRM) of the type V75 (Varian);

b. a digital clock (Patek Philippe) tracked by the time code transmitter DCF77, supplying UTC;

c. two thermometer probes PT100 for the purpose of recording temperature variations, the first of which at the fluxgates, the second inside the variometer's electronics;

d. a computer for the purpose of - controlling the data acquisition, - calculation of secondary variations of any component desired from the primary

variations, - calculating mean values, and - making the output data available to the tape, the printer, and the plotter. In addition, the magnetic tape forms the physical data interface to the Deutsches

Hydrographisches Institut's central computer.

Dt.hydrogr .Z .36 , 1983. H.5. S c h u l z , Experience with a Magnetometer System 175

E

'6"

~ o . . . . . . . . . . . . . . . . . . . . . .

~'~ s:-~

J.___ c p

~ . l . l - O

_ 13)

kO q) (13

Z - 6 0 > t r

- ~ d c O

g

z c % N ~.2 m o ~

"6 co r

o -.~ ._=_

= ~ - . >

>

E-~ ~

r f i l e

- - ~ ~._~

~3::" "d z - r

o co

co ~-

:.E ~

z

@ ~ "~" 09 ~ E z ~

~ o o ~ ~:~ ,_o ,X"-"

d . ~ o~ -6 ..

co> ~ '~ ~ J ~ o .___ o ~ , ~ ~ j o r r E

0

"13 ~E c ~

g-g

~ ~.-s ._~.~ --~ tm ~ c.)

o3 e-

O

E :8 E ~oo co C O ~ . O O

C CO C C ' - - 13) . e E

E c ~ m o~_ "~ E O . - ~ ~_/ k9

t - f - "~

~ - 1 3 0 C O) O C t 3 . - -

t-- r m

4-~ 23 O

~ O 0 ~)

. E o - ~ ~ 0 ~ OO -co o co

~ N o ~ 0 ~ -~-"

0 ~ ----~ C)3 ..~-"

~ o = ; S g 0 " ~.~ . . ---

0 ~ ' ~

,.z e,i e " J ~

"~ ca 2;

e-

co -~

O . . ~ ,~ cor

~. ,OE

0

1


The fluxgates, orientated to geographic co-ordinates, supply variations of the primary components X, Y, and Z. The system's stability is monitored by Z and the secondary components D, H, and F. The following orders of magnitude are valid for Wingst in 1983: D = - 1 ~ 40' (West),

4

H = 1.81-10 nT, 4 , .

Z = 4.53.10 nT, downward posltave.

2 Calibration system

The declination's absolute base-line values are read from the station theodoli te (Schulze 75). For F, and thereby for H and Z, the old period measuring PRM V4931 (Varian) from 1961 serves as a standard and not the V75. A comparison has shown that deviations between both the PRMs have been proved to be due to the V75, and that these deviations are subject to long-term fluctuations of some 0.1 nT (S c h u 1 z and C a r s t e n s [1979]).

The Helmholtz coil's factors had been checked by means of an 82 cm Helmholtz coil (B/(#I~o) standard) before their installation. That makes it possible to continuously monitor the scale values galvanicly by means of the Voltage/Resistance standard (Solartron/Guild- line) with an uncertainty of +0 .5 .10 .3 .

3 Preparations and installation

From 1975 to 1977, the tilt 's progress of the pier for the sensors to detect the horizontal components had been checked (Fig. 2). A clockwise loop, starting from NE, is superimposed by small seasonal motions. The maximum amplitude during a year is some 0.3'. That means some 4 nT, expressed in terms of the horizontal components. As any short-term motions have not been detected, it can be assumed that the long-term motions would be eliminated by the base-line measurements.

In order to achieve an opt imum orientation of the fluxgates referred to geographic co-ordinates, the triple of the fluxgates has been disassembled and the single sensors have been mounted separately on levelled QHM* bases equipped with suitable adjustmelat facilities (Fig. 3). In order to adjust the horizontal fluxgates it is necessary to know their zero field readouts. They have been determined by means of an auxiliary theodoli te which permits observations in four positions. The azimuthal adjustment demands the knowledge of the declination; it has been derived from the analogous system. By means of this procedure, the fluxgates' orientation has been attained to within + 3', if one assumes that the local difference between the absolute and the variometer house does not exceed this value. This assumption seems to be justified, because the difference of H is only 2 nT. 3' corresponds to a crosstalk of 10 -3.

4 Variations of temperature

Due to the double function intended for the PRM, the recording of F will become incomplete. Therefore, one cannot take advantage of the method of eliminating the component-proport ional temperature infuence at the fluxgates by referring the components to F (D e 1 a u r i e r , et al. [1974]). In Our case, the temperature 's variations are recorded simul- taneously and eliminated by the computer. The same procedur is applied to the temperature 's variations at the electronics (resistor within the compensation path), where component-proportional temperature coefficients would also be expected.

* Quartz Horizontal Magnetometer

Dt.hydrogr.Z.36, 1983. H.5. S c h u l z , Experience with a Magnetometer System 177

Fig. 4 shows the temperature coefficients. Note that the coefficients of the relative outputs (except X) are considerably greater than those of the quasi-absolute outputs connec- ted before them. Therefore, not the relative output voltages but the quasi-absolute ones (as by AMOS) are fed to a high resolution digital voltmeter (DVM) (Solartron).

Fur thermore , note that the coefficient of Y is of the same order of magnitude as that of Z. That means that the values are not proport ional to the ambient field components as has been assumed. It is important to know this fact, because the exchange of any board or single IC can lead to changes in the coefficients.

As a rule, the temperature 's variations do not exceed +0.5 ~ at the electronics. The corresponding magnetic corrections lie at _+0.1 nT. Variations of _+1.2 ~ which can be seen in the lowermost trace (Era) of Fig. 8, remain the exception. They have been caused by an anomalously large hysteresis of the air-conditioning. A statistic analysis of this day's period from 6.00 to 10.00 UTC resulted in a coefficient of +0.23 nT/~ for F, whereas + 1.82 nT/~ were to be expected according to Fig. 4. The corrections of the temperature 's variations at the fluxgates are of the same order of magnitude. The lower trace (Sin) in Fig. 8 shows a gradual temperature increase of about 0.2 ~ during that day. From Fig. 4, one can take a value of 1.7 nT/~ for Z, which corresponds to a correction of - 0 . 3 4 nT.

Beside the temperature coefficients in Fig. 4, one can also see the scattering of the (sign-inverted) base-line values (scaled by the right-hand ordinate) as a function of those temperature ranges upon Which the regression has been based (abscissa).

0.5 ~

0

o c

0./-,

0.3

. 77 /

i �9 ~ "*Feb. 7 5

Fe b.77~i~

011' ' 0 2 ' ' 013' tilt e o s t w a r d , r e t a t i v e ,~

Fig. 2. Progress of the tilt of the northern base carrying the fluxgates for the horizontal components.

Monthly means are shown from February 1975 to November 1977


Fig. 3. View of the fluxgates mounted on QHM bases.

The components to be recorded can be seen by the orientations of the Helmholtz coils: X and Y are positioned in the background (North), Z in the foreground.

The thermometer probe is installed between them


8

2

1-

f l ux g ates e lec t ron ics

quas i - abso lu te r e l a t i v e

O. �9 z

T 0" �9 �9 . Y ( Z - o r i e n t a t e d ) sca le o f

+ �9 sca t te r i ng x

. + X ( Z - o r i e n t a t e d )

§

x �9

. , X + x

x ++ + Y + + + Y

, Y § ~

+ �9 § 7 f

" -X " . Z x

X

3o 1'o "c 3'0 t e m p e r a ture ~--

x

X

Fig. 4. Temperature coefficients (left-hand scale) of the fluxgates and the magnetometer system's electronics, respectively.

The coefficients are represented by regression lines, which have been transformed into their horizontal positions. These lines result from a regression of the base-line values against temperature,

the ranges of which are given by the abscissa.

The right-hand scale measures the base-line values' scattering (dots and crosses) against the regression lines. Base-line values of different fluxgates (in different positions) are denoted

as follows:

Crosses: Y; diagonal crosses: X; dots: Z; circles: Y in vertical position;

enlarged dots: X in vertical position

180 Dt.hydrogr.Z.36, 1983. H.5. S c h u t z , Experience with a Magnetometer System

Contrary to theory, the fluxgate for Y has a value of 0.3 nT/~ However , from the strong scattering, one can see that this value is not significant. If one brings the sensor into a vertical position, a value results which is comparable with that of Z, that means that the coefficient for Y can be neglected in the calculation.

For X (in the vertical position), the value differs significantly from that of Z - a further reason for not making use of a reduction via F.

With the resolution of the DVM's within the temperature channels (Fig. 1) of ___0.1 ~ the sensitivity of the registration of Z is limited to + 0.17 nT.

5 Scale values and crosstalk

Fig. 5 shows the result of the scale values' galvanic measurements from the commence- ment of the recording to date. Whereas Y seems to be long-term stable, X shows an increase and Z a decrease.

The variometer 's basic function requires that these trends are also to be found (with the same sign) in the long-term base-line behaviour. A glance on the Z base line .in Fig. 10 confirms this assumption: Z decreases by a rate of the order of magnitude expeCted. Therefore, the Z component 's long-term drift seems to be due to an increase of the sensitivity, presumably caused by an ageing of the compensation coil and/or an ageing of the resistor within the feedback path. An ageing of the Voltage standard can be excluded.

A remaining residue can be interpreted as a tilt 's progress of the sensor and/or a drift of its zero field readout. A t present, the two effects cannot be separated. The behaviour of X in Fig. 5 does not reflect that of H in Fig. 10. This fact can be interpreted in the same way.

In addition, the scale value of Y was checked by means of simultaneous readings of the station theodolite (circles in Fig. 5) in August 1981. The result suggests that the decrease common to all scale values at that time is not real but has been caused by a defective Voltage/Resistance Standard. However, it must be kept in mind that the values checked this way can be falsified by the other components ' crosstalk on D (or Y).

For H and Z, the effects of crosstalk and uncertainties of the scale values can be separated by means of recording with the PVM at the same time. In doing so, the PVM is used as a variometer of a higher class - not only with regard to the scale values but also to the orientation of the component in question.

For that purpose, the compensating field and, in the case of Z, also the azimuth of the compensating coil have to be adjusted precisely. This condition is undoubtedly fulfilled at the moment of the adjustment. After that, the influence of variations perpendicular to the component in question have to be eliminated by the computer. This procedure is performed by means of the recorded variations themselves. Their influence increases quadratically with increasing amplitudes.

In addition, it must be taken into account that the variations are scanned serially via a multiplexer within the analogous circuit, i. e. with time delays between the components. The error attached thereto is proport ional to the variations' gradients. In this case, they have been eliminated by spline interpolation including adjacent values. A comparison of the uppermost traces in Fig. 8 and Fig. 9 show to what sum this error can amount to. In F!g. 8 (format: R O H D A T ) , the course of the F base-line is derived from uncorrected values. The delay between the triggerings of the PRM and that of the fluxgate magne tomete r s Z channel is 5 seconds.

Fig. 6 shows the magnetogram of the storm of 25 and 26 July, 1981, to which the procedure has been applied. The intervals of simultaneous recordings are denoted by widened traces of the corresponding components. The uppermost trace shows the corresponding scattering of the H and Z base-line, respectively.


A multiple regression leads to the following partial regression coefficients which can be interpreted as corrections of the measured scale values (Fig. 5) or corrections regarding the crosstalk:

6 H o = 10 -4 (5.8 X + 1.6 Y - 4.4 Z)

6Z o = 10 -4 (2.3 X - 2.3 Y - 7.0 Z)

Accordingly, the measured scale values of X and Z are to be corrected by +5.8-10 -4 and - 7 . 0 . 1 0 -4 nT/pars, respectively. The corrected value of X (cross in Fig. 5) lies very close to the value expected (regression line).

The crosstalk of the horizontal components on Z can easily be checked experimentally: a turning of the Z base's upper part around its vertical axis confirmed the value for X. How- ever, the influence of Y proved to be somewhat positive.

Fig. 7 shows the values of the H base-line as a function of Z, under the secondary condition of constant horizontal components: dots (5 minute samples) and circles (30 minute samples). The same samples are denoted in Fig. 6 by circles and short lines, respectively�9

�9 . . - 4 .

The course of their regression hne (dashed) supports the value of - 4 . 4 - 10 (continuous line).

Now one can combine both results of the analysis into one equation:

6F o = 10 -4 (4.3 X - 1.5 Y - 8.1 Z)

Y also contributes, because it is a matter of the influence of terms of second order. This equation can be checked by an analysis of the storm of 13 and 14 July, 1982, when F was recorded at the same time (Fig. 9). The result is:

a F o = 10 -4 ( - 4 . 4 X - 1.6 Y - 6.6 Z)

I 1OOO1 nTlmV 1

~ | 9"961

I (3.00 1 �9 �9 o - - 0 �9 �9 �9 e O �9 �9 �9 - - �9 - -

998 1 y " . �9

10.02- "":-" " " ~ "~o

lc~oo-I z �9 I I l

1980 1981 1982 I

19 83 year �9

Fig. 5. Scale values as a function of time from January 1980 to June 1983. Different methods of their determination are indicated as follows:

Dots: galvanic; circles: comparison with the station theodolite;

crosses: regression analysis


- ' - I

P - ! , 0 o

eA.~la= o~ ~ o E

H ~ r--

i , i ,

I | I

~- % I

I '

i

% I

i i

~ %

r~ ~ t t ~

"= eq

~

~ �9

N =

~ . ~

&

b. 0


cl

-r

18 072-

nT

18 071

0

_ _ - 4 , 4 . 1 0 - ~ �9

�9 0

+ # §

x x

e O 8

I

+ § ** ++

x x x xx

x x

I I I t I I I

-150 -100 -50 0 50 nT 150

var ia t ion A Z >

T -0 %"

-.---

-nT ~ X

- -100

Fig. 7. H base-line as a function of + 150 nT variations of Z (crosstalk). The strong line shows the result of a regression analysis of the storm of 25 and 26 July, 1981.

Dots and lines indicate samples taken under the secondary condition of fairly constant horizontal components (right-hand scale).

The dashed line smoothes the samples linearly. See the corresponding symbols in Figure 6:

Variations are denoted as follows: Crosses: Y; diagonal crosses: X


z o

:i: ~: i

:,,

:i!7, <

t '

} _ !

1

:r L- I

r - - i |

N

o

o o

v-- E4 c o - -

t,,-- - -

' ' I r O I >

c~ s +

| i i '

E-~ o o o

t " - " ~ kD

O I > O

c~ ~ o

m ~

o ~ ~ ~ ~ c~ ,,o ~-

,r

I

o

d c~

1 I

o ~ o ~

, r

0

m

= "-? .o

.~ N

~ [ ~ = ~

o > .~ 2 .

~ o ~ o

~ ~

o E x~

o6 &


t en

( '4 .~

g

~ -...

.~.~

u %

i I ' o o

c 0 co

0--

o o

0~

-#'

o o p.l

I '

i I o o ~-~

5 /

I

f ~

o

o

o

i

o o

t ~

I

O

CO

I

I O

O

M3

I

o

% I

I o

H w~ %_ ~ 3

d

c-

o

O

O

O

�9

=.


The influence of Y (crosstalk on X and Z) and that of Z (crosstalk on X and correction of the Z scale value) correspond to each other in their combined effects. A particular result is that the galvanicly determined scale value of Z seems to be falsified by a systematic error (perhaps an erroneous coil factor).

The discrepancy of the X terms can only be understood, if the fluxgate of Z tilted within the magnetic meridian during the period between both the storms, and/or the scale value of X measured in July 1982 was wrong. However, the difference can also be interpreted as an expression of statistic uncertainty of the analysis itself: the partial regression coefficients' statistic significance still remains to be checked, their confidence limits have not yet been determined.

6 Stability

In Fig. 10, one can see the measured base-line values of D, H, Z, and F for a two year period. The closing error C of the components within the magnetic meridian is displayed with higher sensitivity (left-hand scale) in the uppermost trace.

The trend of the Z base-line has already been discussed in connection with the scale values. This trend is superimposed by an annual variation of an amplitude of _+ 10 nT. A similar course leading slightly in phase can be found in H. A comparison of the amplitudes leads to the conclusion that the variations are proportional to the components themselves; therefore, they cannot be due to (joint) motions of the piers (Fig. 2). Accordingly, seasonal

_~_ 4 fluctuations of the scale values (of _2 .10 of Z) are expected, but cannot be resolved because of the scale values' scattering (Fig. 5).

The oscillations seem to be correlated with the annual variations of the humidity inside the variometer house (lowermost trace in Fig. 10), the minima of which are ahead of the base-lines' maxima by some 30 days.

There is every reason to believe that leakage currents emerged at uninsulated points within the compensation loop and then flowed across. A great humidity in the late summer led to small transfer resistance (perhaps organic material), which were passed by component proportional currents, resulting in corresponding drops of the base-lines.

The D base-line also shows an annual fluctuation which could be interpreted as a fluxgate's motion perpendicular to the magnetic meridian.

A short-term variation can be seen in Fig. 12. This figure demonstrates the base-lines' course during April 1983. Beside the base-line values of F at the times of measurement (enlarged circles), the uppermost trace contains the course of the F baseline's hourly values. The maximum value of some 1 nT occurring between 20 and 26 April is of interest. It has just been resolved by the sequence of base-line measurements of H and Z. A drift of the V75's quartz can be excluded as being a cause, because the quartz's frequency has been monitored continuously by the frequency standard.

As the values of the base-lines' closing error - which is also shown in the uppermost trace (dots, right-hand scale) - denotes, the H base-line's deviations on 25 and 28 April in all probability, are due to the PVM.

The lowermost trace shows the D base-line. The relative (dots, PVM) and the absolute (circles, station theodolite) values are well in line, pointing to a good stability of the coil theodolite. That can also be seen from the long-term representation in Fig. 10, where the absolute measurement's results are denoted by small vertical lines.

The hourly F base-line values in Fig. 12 permit the hourly instability to be estimated. The average scattering does not exceed +0.2 nT.

The F base-line in Fig: 9 permits the minute instability to be estimated. The mean of the scattering does not exceed some +0.4 nT. The corresponding trace in Fig. 6 (top) gives somewhat greater values for Z and H, respectively.


On August 12, 1982, a heavy thunderstorm initiated a series of jumps and strong gradients of the H base-line, lasting until the repair in January 1983 (Fig. 10). During this period, the H base-line was traced by the Z base-line and hourly values of the F base-line. The Z base-line has been approximated by linearly smoothing adjacent values. Fig. 11 shows a detail. In Fig. 10, the corresponding base-line values are indicated by dots (H) and crosses (F), respectively. Note that the base-lines in both Figures refer to different nominal values, i.e. different readouts of the DVM. It has been ascertained that this anomalous behaviour was neither due to the electronics nor to the sensor cable and their connections to each other and to the fluxgate. The base-line became quiet (Fig. 10 and Fig, 12) only when the fluxgate had been exchanged by a suitable spare sensor. The examination of the fluxgate revealed that where the spare part showed a cast feeder channel, the removed fluxgate had an unsealed hole which permitted a view of the interior (fabric of the wire). Probably moisture which had penetrated had played the major r61e.

1981 1982 1983

c ~ T "~-~ *'* '** .............. ' ....... i i ' " '"'; ....... " " ' '~'"'*iti~ " ~ ' ~ ' i '~ .....

~ 1 7 6 1 7 6 " """""" ..... . . i ~''''''''*~ ' ..... . . . . . . . . . :'::'~" " "'"' ....... ' ........ """ ........ " -

"~ 18065tnT1 '"""""" """' ........... '""""| *""""""J /' '"'

-"""~ -]" '-" " ............. ~---t,,t,.,,,,""-"""-"-"""_"",,,,-r- , , , 9 o t .... ,,,-_ ""'",,,,,,, Zo oT1 " " " " " ' ~ " " " " ~ ~516o i '"""',,, ,, I , , 'Y' ........ ""'"""'"'",, """""" ......... "'"' ,

T -,7o. __ ,,,,,,,.,,,,,,,,~,,,,,y, """"","""'",,~,,S I,,,,,'"'"" ................ '"'",,,,,

~_ .,. ~ E 7 5 ' , ,

Fig. 10. The variation system's base-line values and the humidity inside the variometer house from July 1981 to June 1983.

The dashed line indicates the long-term drift of the Z base-line.


|

->- 10-

ClJ

~- nT- "T 01

r

m- 5

0

1982 D e c e m b e r

Z(~= &5170nT . A B e ;

x - - - x "---~......"-'"""":'"'-" +.

==,

"-k

"Jo'.; " " " F o = 48 665 nT * ~,B

I=.= e . . . -..."

. "* '~ =o" . . " eo �9

". r i o �9 i o "" Wf".;""'""""'"""'"" :

�9 " . QI

,,",,..: �9 " . ~ 1 7 6 .." Ho__.18102nT§

17 18 19 20 i i 1

21 22 23 t i m e >

Fig. 11. Anomalous short-term drift of the H base-line from 17 to 23 December, 1982, represented by hourly values (enlarged dots).

The values have been derived from hourly F base-line values (dots); the Z base-line has been approximated by a regression line,

Base-line measurements are represented by circles and crosses


/.8778-i I *03

4877 18116]

H, nT~ " =

18110

452844

D, -1"28' 1 .

-1*29'J

4" 4- "f" 4"

1983April 1 5 10 15 20 25 30

Fig. 12. The variometer system's base-line of April 1983, after exchange of the X fluxgate on 25 January, 1983.

Top: F base-line, tracked by hourly values. The enlarged circles indicate the time of the base-line measurements;

the right-hand scale measures the values

C = F o - V ~ o + Z~, denoted by dots;

bottom (declination): circles indicate measurements carried out by the station theodolite; dots indicate (relative) measurements carried out by the proton vector magnetometer


7 C o n c l u s i o n a n d o u t l o o k

It has been shown that, at the Erdmagnetisches Observatorium Wingst, fluxgates could be orientated, and, furthermore, these adjustments can be monitored long-term with an

�9 -4- - 3 absolute uncertainty of _0 .5 -10 . For the scale values, the same value of uncertainty applies�9

Moreover, the temperature variations' influence can be eliminated sufficiently, referred to six carefully determined temperature coefficients, if one presumes that the temperature variations do not exceed certain limits.

It follows that 1000 nT variations can be recorded with a remaining uncertainty of +_ 1 nT. Hereby, a limit seems to be set for a system which includes fluxgates.

On the contrary, improvements of the long-term stability of the system at Wingst Observatory are possible: - In order to eliminate the humidity 's annual influence upon the base-lines, the fluxgates and feed lines should be sealed against moisture; - in order to include Y in the continuous stability check and the occasional checks of the crosstalk via F, the co-ordinate system of the primary components shall be rotated by 45 ~ by means of an azimuthal turning of the horizontal fluxgates in the same direction.

Furthermore, we intend to install two further DVM's (Solartron) in order to achieve a strictly synchronous sampling of the components and to increase the sampling rate (at present 10 s) limited by the analogous multiplexer.

The mechanical multiplexer is the weak link of the whole system. Many disturbances have been due to this component . Therefore, in conclusion, a word about the system's failure rate. In 1982, 517801 minutes were recorded, i .e . 1.5 % of the information had been lost by technical disturbances, maintainance, and breakdowns of the operating system, because the computer is also used for off-line data evaluation and editing tasks in the background (Fig. 1). Therefore, the tried and tested good old fibre suspended magnet system (Schulze, La Cour) will continue to be kept in condition in order to provide a complete stand-by system.

References

D e l a u r i e r , J . M . , E . I . L o o m e r , G . J a n - s e n van B e e k , et al., 1974: Editing and evaluating digitally recorded geomagnetic c0mponents'at Canadian observatories. Publ. Earth Phys. Br. Dep. Energy, Ottawa. 44, No 9, S. 235-242.

S c h u l z , G., andU. C a r s t e n s , 1979: Ape - riod measuring proton magnetometer with a direct readout. Dt. hydrogr. Z. 32, 119-125.

S c h u 1 z, G., 1981: Base-line measurements of the declination, by means of a proton vector magnetometer, at the Wingst Geomagnetic Observatory (Erdmagnetisches Observa- torium Wingst). Dt. hydrogr. Z. 34, 26-37.

V o p p e 1, D., 1972: The proton vector magnetometer at Wingst Observatory, Deutsches Hydrogra'phisches Institut. Erdmagn. Jahrb. 17, 133-149.

Eingegangen am 19. Dezember 1983

Angenommen am 17. Januar 1984

Anschrift des Verfassers: Dipl.-Geophys. Gfinter Schulz, Deutsches Hydrographisches Institut, Erdmagnetisches Observatorium, Am Olymp 13, 2177 Wingst

experience with a digitally recording magnetometer system at wingst geomagnetic observatory...

Documents