Newly Digitized Data From Scandinavian Magnetometer Array Network Shows Large Regional Differences in Magnetic Environment

The International Magnetospheric Study (IMS) took place from 1977 to 1979. An objective was to study the magnetosphere at different heights on the Earth and in space simultaneously. The Scandinavian Magnetometer Array (SMA), where a dense magnetometer array was placed in the northern part of Scandinavia, was part of the IMS. This array extended 1,570 km in north‐south and 1,290 km in east‐west directions. The SMA‐magnetometers contained a camera with 35 mm film and three wire‐suspended magnets. These instruments recorded the movement of the magnets optically on the film. The usability of the SMA data has been limited by time‐consuming digitization by hand. Thus, most of the recordings has been left nondigitized and unstudied. We have developed a method, named DigiMAG, to digitize the SMA recordings by using a custom‐built device. This article presents the high‐latitude dynamics of the strongest magnetic storm in 1977 between October 26 and 30. We analyze newly digitized data from Rostadalen (ROS), Evenes (EVE), Ritsemjokk (RIJ), and Kiruna (KIR) stations for the three storm‐time substorms on October 26–29, 1977. The results show that in the vicinity of the 65° latitude, the storm‐time differences of horizontal magnetic components can exceed 500 nT. During the recovery phase of a substorm, 2.8 nT km−1 difference arises over a distance of 167 km between the H‐component time derivatives of ROS and RIJ stations in 3 minutes.


10.1029/2022JA030311
2 of 23 the newly digitized data from the high-latitude SMA stations for the largest geomagnetic storm in 1977.According to the lowest Dst peak value, the digitized storm interval is the fourth largest during the IMS and the sixth largest during the entire SMA operation interval.The SMA stations used here are selected such that the digitized data can be used as AE data, and thus filling the missing data for AE indices.The results of this article are compared with data from a permanent magnetometer station Abisko (ABK).We have developed a new method named "a positive-negative derivative (PdNd)" which enables comparison of rapid magnetic field changes during substorm expansion phase.

SMA Network
Coordinated simultaneous multimethod IMS measurements at different altitudes were seen as an advantage of the SMA over other magnetometer array field operations (Küppers et al., 1979;Pellinen et al., 1982;P. Tanskanen et al., 1982).The assumption was that electrojets flowing in parallel to the magnetic oval would predominate.Thus, the SMA station's locations formed direct chains approximately toward the magnetic north.These locations are shown in Figure 1, and the coordinates are given in Table 1.The SMA instruments were buried vertically on the ground, primarily to locations where the static magnetic field gradient was low.This reduced the errors in declination measurements with the compass theodolite (Küppers et al., 1979).
At the high latitude, the dense array extended from western Norway to the Soviet Union.At lower latitudes, the distance between the stations increased, especially in the magnetic north-south direction.During the IMS, there were temporary and permanent magnetometer stations in the Scandinavian region.Temporary stations came from the Institute of Geophysics, the University of Münster, the Technical University of Braunschweig, the Polar Geophysical Institute at Apatity, and the Geomagnetism Unit of the UK Institute of the Geological Sciences.Permanent stations were Ny-Ålesund (NAL), Bear Island (BJN), Tromsø (TRO), Abisko (ABK), Loparskaya (LPY), Lovozero (LOZ), Sodankylä (SOD), Dombås (DOB), Nurmijärvi (NUR), Lovö (LOV), and Leningrad (LNN) (Green, 1981;Küppers et al., 1979;Maurer & Theile, 1978;Stuart, 1982).
After the IMS ended in 1979, the EISCAT magnetometer cross began operating in northern Scandinavia in 1982.This magnetometer cross was a joint operation of the Technical University of Braunschweig, Sodankylä Geophysical Observatory (SGO), Finnish Meteorological Institute (FMI), and the University of Tromsø.Initially, the cross consisted of the following five stations: Sørøya (SOR), Alta (ALT), Kautoketo (KAU), Muonio (MUO), and Pello (PEL).Kilpisjärvi (KIL) and Kevo (KEV) joined to the cross in 1983 (Lühr et al., 1984).ALT and KAU stations were removed from the cross in 1990 before the current IMAGE network began operating in 1991.Figure 2 shows the stations of different magnetometer arrays that were located in the vicinity of the AE station at Abisko.The magnetic longitude of ABK has changed nearly 4° since the SMA measurements.The SMA network had the densest station spacing around the AE station.Subsequent arrays have no stations within a radius of 200 km in the magnetic west, magnetic north, or the area between these directions.Only the SMA array has recorded existing data on the proximity of these directions.The data provides information on changes in the magnetic field over a short distance in the vicinity of the 65° latitude.

Type Münster Gough-Reitzel Magnetometer
The standard Gough-Reitzel magnetometers were designed for field use.Except for a battery, the components were installed inside an airtight tube.The unit price of a portable magnetometer was under $1,500 in 1967, which is equivalent to $11,600 in 2020 due to inflation.When the tube is buried in the ground, temperature-sensitive magnets and torsion wires are at a depth of 1.5 m or more.At this depth, the temperature gradient is no more  (Gough & Reitzel, 1967).Field operation experience has shown that temperature compensation also works in north Scandinavia if ground contact is sufficient.The SMA instruments were based on modified Gough-Reitzel magnetometers with a 35 mm film camera and three wire-suspended magnets (Baumjohann, 1982;Küppers et al., 1979) Note.The symbol, name, and geographic coordinates (GCS) are adopted from (Küppers et al., 1979).CGM coordinates are calculated according to (Gustafsson, 1970;Tsyganenko et al., 1987).A start date indicates the beginning of the first measurement interval and end date the closing of the last interval.

Table 1 SMA Station List
Gough-Reitzel magnetometer.Compared with the original, it was cheaper to manufacture and consumed less energy.The maximum running period increased from 24 days to over 70 days.A normal car battery was replaced with a Dryfit battery, and the location was moved into the tube.The Bulova-Accutron clock was replaced with an electronic quartz-clock, and a timing accuracy increased from ±15 to ±1 s per week on average.Before the SMA field measurements, all instruments were tested and calibrated at the site of the observatory at Wingst of the Deutsches Hydrographisches Institut.The scaling values were changed to fit the conditions in Scandinavia.When the scaling values in Germany were between 10 and 20 nT mm −1 , suitable values in Scandinavia were generally 40 nT mm −1 .The recording magnets of all instruments were aligned in relation to the magnetic north and vertical direction.After testing and calibration, the movement of the magnets was prevented during transportation to Scandinavia (Küppers & Post, 1981).
The theoretical instrument and component-specific measuring range were mostly determined by the scaling values and the maximum shift of the trace.The measuring range was increased for all the components by creating auxiliary traces recorded on the film when the main trace decreases below or increases above the specified threshold.The auxiliary traces were illuminated with other light sources and mirror splitting (Küppers et al., 1979).
The measuring range ends when the auxiliary trace crosses the edge of the film.This occurs during the major  Gustafsson (1970Gustafsson ( , 1984) ) and Tsyganenko et al. (1987). 10.1029/2022JA030311 5 of 23 magnetic storms approximately 20 mm after the main trace has crossed the same edge.The accuracy and capability of the original Gough-Reitzel magnetometers and the SMA instruments have been demonstrated in numerous field operations (Bannister & Gough, 1977;Beer & Gough, 1980;Küppers et al., 1979).The SMA instrument's magnetic resolutions were approximately 2 nT and the magnetic field variations of phenomena lasting more than a minute are well-recorded (Gough & Reitzel, 1967;Küppers et al., 1979).

SMA Data and DigiMAG Method
The SMA station was buried in the ground to measure unattended for approximately 73 days.After this, the station was prepared for a new measuring interval.The preparation included, among others, the replacement of the battery and camera (Küppers et al., 1979).The last measuring intervals were in 1980, and the total amount of recorded data is approximately 44,000 days and 29 km of Kodak RAR 2498 film.Araki et al. (1992) have sketched a graph that shows the intervals for available data.However, it does not show that there can be a long period of time between the end of the magnetogram and the beginning of a new one.The time intervals with these gaps for available data are shown in Figure 3.The resolution of the graph is 1 day.Thus, a continuous long period indicates that a new measurement interval has started during the same day as the previous one has ended.
Possible instrument failure could not be detected during unattended operation.The total loss of the recordings is approximately 10%, and minor losses are higher (Baumjohann, 1982;Küppers & Post, 1981).The end of the Z-component registration is a common failure in the recordings.Its recording magnet was located near the bottom of the tube and its wire was the longest.Data recorded by the SMA array comprises more than 620 film reels.Sodankylä Geophysical Observatory possess the original recordings.Copies of the recordings were provided to the WDC for Geomagnetism in Kyoto.Each reel has a package with station and time information that makes it possible to find the desired reel.The first SMA researchers noticed that its slow optical recording limits the usability of the SMA data (Araki et al., 1992;Küppers et al., 1979).Now decades later, a device, shown in Figure 4, is built for photographing the SMA data.The main motor rotates the film reel, and a camera attached to the device frame takes photos of the film.During one 360° rotation of the main motor shaft, rods attached to the shaft first move the film and trigger the camera.The film is pulled by a wheel with a rubber at the edges to improve friction.The rod attached to the motor shaft pushes the vertical steps at the bottom of the wheel and moves it 30°, or 1/12 of the wheel's circumference, on every round.When the copper rod attached to the main motor shaft goes between the sensor and the LED, an optical switch triggers the camera.
The auxiliary motor rotates the photographed film on a blank reel.
Photographing a magnetogram with 73 days of the data takes less than 2 hr.The height of the recorded optical data remains the same in pixels.After the pixel/mm ratio is known, data can be scaled using nT/pixel values.Scaling large amounts of the data with the same usually linearly increasing or decreasing scaling values is possible.This makes data processing less time-consuming.An example of a recording photographed with the digitization device is shown in Figure 5.The main phase of the magnetic storm begins, and the magnetic components H and Z decrease rapidly.The positive amplification direction for these components is downward on the film.Data can be collected from photographs by measuring the trace's distances to the B1 baseline at even intervals.If the trace crosses the film's edge, its theoretical distance can be calculated from the auxiliary trace until the instrument's measuring range ends.
A 3-min resolution of this article is achieved by measuring the values 20 times per hour.The best possible resolution would be 10 s because inside the instrument was a lamp that switched on every 10 s and illuminated the magnet's location in the film (Gough & Reitzel, 1967;Küppers & Post, 1981).The collected data needs to be scaled in order to know the actual variations in the magnetic field.Different film reels have different scaling values for all the components that determine how much one-mm change is in nanoteslas.These scaling values are usually different at the beginning and the end of the film reel.Several film reels also have a positive or negative time delay at the beginning and the end of the film reel.The assumption is that changes in scaling values are linear functions, and delays do not affect the results at 3-min resolution.The delays are less than 2 s in the film reels used in this article.The collected unscaled data does not show the real variations in the magnetic field.The magnitude of the scaling values varies between and along the recordings.Thus, only a visual inspection of SMA magnetogram is not a reliable method for comparing and examining data.During strong magnetic disturbances, the traces form a lot of intersections.There is a high risk of mixing up the traces if their color and shape match after the intersection.However, due to the different location reflected light rays with respect to the optical axis parallax effect exist which allow identification of the different traces.The recording mirror surfaces of some SMA instruments have been scratched to make the traces of recorded data different looking.The scratches on the mirror stand out as scratches on the recorded trace.
The distance of the trace to the baseline is compared with a quiet time reference (QTR).This reference value is film and component-specific, so it must have occurred during the same film record interval. .The Dst index was 0 nT and very stable for the previous and the next couple of hours.This reference value is measured in pixels from the film reel photos.Then, the value is reduced from each measurement point from the collected data.The value of quiet time is a new x-axis.Then, in the case of H and Z, each data point is multiplied by −1.This turns the image upside down and does not affect any zero points.The image is moved to the correct place in the coordinate axis.After this, each data point is scaled using the calculated nT/pixel value.The data point's scaling value depends on the scaling values at the ends of the film and the distance of the point to these ends.The result of the following function begins to increase or decrease linearly depending on whether the scaling value at the beginning of the film, α, is greater than the value at the end β.
The number of measurement points, x tot , determines the rate of scaling value change along the measuring points.
The measuring point number, x n , determines the number of the point calculated from the beginning of the film.With a fictional day-long recording and 1-hr resolution, at the measurement point 10:00, x tot = 24 and x n = 11.The first measurement point would be 00:00 and the last one 23:00.If the H-component scaling value is 40 nT at the beginning and 45 nT at the end of the recording, the H-component scaling value is 42,08 nT at 10:00.If the resolution is increased to the 3 min used in this paper, for the same fictional recording at 10:00, x tot = 480 and x n = 201.
Collecting data from the photos is a manual job.The collection benefit/speed ratio is at its best 60 min resolution due to the timestamps of the film.The next best efficiency is at a resolution of 6 min.During the strong disturbances, the traces of the different components intersect, and it takes time to identify traces if the recording mirrors of instruments have not been scratched enough.

Identification of Storm Period and Space Weather Conditions
During the main phase of the magnetic storm, the Earth's ring current intensifies rapidly and compresses the horizontal component of the surface magnetic field.The disturbance storm time (Dst) index stations are located near the magnetic equator, where the strength of the surface magnetic field is inversely proportional to the energy content of the ring current.As the magnetic field of the interplanetary magnetic field (IMF) turns to point north, the ring current and the Dst begin to recover (Gonzalez et al., 1994;Hamilton et al., 1988;Sugiura, 1964).Thus, the lowest Dst min value reached by a magnetic storm depends on its magnitude.When this value is between −50 and −100 nT, a moderate storm is ongoing.A strong storm reaches a value of (−100 nT ≥ Dst min ≥ −200 nT).
Severe as well as great super-storms reach values below −200 nT (Loewe & Prölss, 1997).Figure 6 shows the magnetic disturbances with the lowest Dst indices of 1977 that were measured when the number of charged particles in the magnetosphere was high.The lowest Dst values of these disturbances are centralized for ease of comparison.During the October event, the Dst decreases to −159 nT, and this period is digitized.
The LRO data is a comprehensively cross-compared near-Earth data source for plasma and magnetic field parameters shown in Figure 6.For 1977, the LRO plasma data sources are IMP-7, IMP-8, and ISEE-1 (Asbridge et al., 1976;Bame et al., 1978;Lazarus & Paularena, 1998).For the same year, the only LRO near-Earth IMF data source is IMP-8, and it was in the solar wind for 7-8 days during each 12.5 days orbit.Figure 7 shows the orbit during the October event.The spacecraft (SC) reaches the night side of the x-axis 25-10, 15:30 and returns on the dayside 30-10, 04:05.The timestamp of the last available parameter after moving on the night side is 25-10, 19:00, and the first one before the dayside 29-10, 02:00.(Asbridge et al., 1976;Bame et al., 1978;Lazarus & Paularena, 1998).
There is a correlation between the emergence of geomagnetic storms and the occurrence of coronal mass ejections (CMEs) (Burlaga et al., 1981;Wilson & Hildner, 1984).In order to cause a strong geomagnetic storm, the corona must erupt toward Earth and be accompanied by a southward magnetic field component (Gonzalez et al., 1994;Russell et al., 1974).
The Earth's magnetosphere internal rotation movement enhances as the IMF carries a negative Bz component to the magnetopause.The Earth's magnetic field is pointing in the opposite direction and tears open under the flow of the IMF.With a phenomenon known as magnetic reconnection, the IMF field lines connect to these open Earth field lines (Dungey, 1953(Dungey, , 1961)).The combined field lines travel within the solar wind to the night side of the Earth and accumulate magnetic flux in the tail lobes.Part of this energy is discharging within the returning field lines into the polar ionosphere (Aubry & McPherron, 1971).
Occasionally, the solar wind carries large structures with different magnetic characteristics compared with the ambient environment.These magnetic clouds (MC) are defined as the areas in the solar wind where the magnetic field has enhanced strength, the direction of the measured field changes smoothly when an SC passes through the MC and proton temperature (and beta) are low in comparison with the ambient values (Burlaga, 1988;Burlaga et al., 1981).
The solar wind magnetic field parameters are needed for comprehensive multimethod SMA data analysis.
The resolution of IMP-8 magnetic data is 320 ms, and data is recorded regardless of the orbit phase.This article uses a 1-min average data set created by NASA's SPDF in 2009.One-minute averages were calculated using a 15.36 s magnetic field data set.The data period is mainly flagged, meaning that the data is not from the solar wind, and critical interpretation is emphasized.
Figure 8 shows the Dst and the best available space-borne data for the October event.The magnetic equator is under the impact of the greatest magnetic storm of 1977.For the SMA network located at a higher latitude, the storm period is divided into three major substorms based on the digitized data.
Plasma speed and proton temperature are relatively low and rise slightly as the ring current begins to recover.Proton density increases as the Dst begins to decrease.Thus, it seems the Bz orientation has not changed during the passage to the magnetosphere, and IMF feeds the magnetosphere with the new particles when the night side Bz is negative.The magnetic field strength (B) rises sharply on the night side and reaches its peak as the SC passes through the magnetotail.The stretched magnetotail can be seen as a low Bz value when IMP-8 is around the maximum distance on the x-axis on the night side.

Results
The clearest average characteristic of a magnetic storm is the rapid decrease of the H-component and its subsequent slow recovery (Chapman & Bartels, 1940).This decrease is the consequence of an increase in the number of particles trapped in the magnetosphere.The ring current is formed by ions and electrons moving in different directions from midnight.Ions are moving toward dusk and electrons toward dawn.The H-component decrease lasts for several hours or even days when the southward interplanetary magnetic field (IMF) feeds the magnetosphere with the new particles (Gonzalez et al., 1994).It is recognized that the ring current is asymmetric during the main phase of the magnetic storm.Furthermore, this dawn-dusk asymmetry causes substantially stronger magnetic disturbances on the dusk side (Fok et al., 1996;Grafe, 1999;Walsh et al., 2014).The effects of the magnetic storm on the equator are seen as substorms at the high latitude.These substorms appear as strong fluctuations in the H-component shown in Figure 9.The right side y-axis of the Dst index indicates the similarity in the horizontal magnetic disturbances at a magnetic equator and high latitudes.

Digitized H, Z, and D Components
The three largest substorms have been identified from the storm of 26-30 October 1977.This separation is made according to the lowest resolution (LRO) 1 hr data to allow multimethod data analysis and is shown in Figures 9-15 with light gray shading.The beginning and ending of the substorms are given in Table 2.
Subtracting the data from lower latitude stations from the higher latitude data indicates the high momentary differences over a short magnetic north-south distance.The largest difference occurs during the recovery phase of substorm 2. The west-east differences are minor in comparison with the above-mentioned.Substorm number 1 occurs with a peak amplitude of −324 nT (ROS) during the initial phase of a magnetic storm.The Dst measured at the equator increases, and the H-component measured by the SMA network's stations studied decreases.The second substorm begins at 09:18 in ROS, at 09:57 in RIJ and KIR, simultaneously to the beginning of the storm The third substorm starts exactly at the same time in all three stations, but it ends at 20:15 in KIR, and 3 and 18 min earlier in RIJ and ROS, respectively.The second substorm was longest lasting 20hr 21 min, while the first substorm last 4 hr 30 min and the third 8 hr 15 min.While momentary differences of ROS station to the lower latitude stations remains mainly within 200 nT, at the interface between the end of the expansion phase and the beginning of the recovery phase, occurs −503 nT (ROS-RIJ) and −450 nT (ROS-KIR) differences between these stations.The largest difference between approximately the same latitude stations (RIJ-KIR) is −183 nT.
The subtraction images indicate the average differences between the stations in the vertical magnetic disturbances are smaller in the magnetic east-west direction than the magnetic northeast-southwest and the magnetic northsouth directions.After the recovery phase at the equator has begun, the digitized main substorm reaches the level of quiet time for the first time at 05:39 ROS and RIJ stations.KIR station, located 1.20 CGM latitude degrees below ROS station and 0.08° below RIJ station, reaches this level at the next measurement point 3 min later.Thus, all the three SMA stations have reached quiet time level approximately 3.5 hr after the expansion phase of the strongest substorm reaches its peak.This occurs at the interface of dawn just before sunrise.
The Z-component variation is smaller than in the H-component and the strength increase occurs in the opposite direction (Chapman & Bartels, 1940).While the largest values of the Z-component are −146 nT (RIJ) for substorm 1, 828 nT (EVE) for substorm 2, and 343 nT (EVE) for substorm 3, corresponding peak values of the H-component are −324 nT (ROS), −1,196 nT (KIR), and −542 nT (KIR).Figure 10 shows the Z-component's rapid growth during the expansion phases of substorms 2 and 3.The subtraction images indicate that the highest average difference between the stations in the eastern magnetic disturbances occurs between the southeast-northwest direction (KIR-EVE).This average difference is 35 nT for all three identified substorms and 47 nT for substorm 2. The average differences of the other D-component subtraction images range between 20 and 30 nT for all three substorms and 30-39 nT for substorm 2. The following differences arise between stations during the expansion Phase 758 nT (EVE-RIJ), 611 nT (KIR-RIJ), −577 nT (ROS-EVE), and 498 nT (ROS-RIJ).The largest differences in the north-south direction arise at the interface between the end of the expansion phase and the beginning of the recovery phase and they are 507 nT (ROS-KIR) and −455 nT (KIR-EVE).

Derivatives
Figure 12 shows time derivatives for the H-component.When t = 6 min, the substorm is considered having begun when SMA station's |dH/dt| > 10 nT and H-component trace increases in the direction of the derivative without  2.
Changes in the space weather cause geomagnetically induced currents (GICs) at the ground level.The Earth's conductive land structure acts as a conductor of GIC (D.J. Thomson & Weaver, 1975).Part of these currents are grounded in power networks.The greatest GICs are strongly associated with large changes in dH/dt value (Viljanen et al., 2001).A case is known where the dH/dt value reached roughly 10 nT/s at Eskdalemuir Observatory and GIC measuring instrument, located almost 100 km away, measured a peak value of 42 A (A).The "risk limit" for a GIC that damages power systems is considered to be 25 A (A. W. P. Thomson et al., 2005;Freeman et al., 2019).Figures 12a, 12b and 12c show that latitude affects horizontal storm-time absolute average time derivatives that are denoted  |∕| .Excluding during substorm 2 at ROS and RIJ stations, these average derivatives are larger at higher latitudes.When we digitize long measurements periods and consider the effect of latitude, it can During the expansion phase of the substorm 2, in 3 min, 3 nT km −1 difference arise between the vertical Z-component time derivatives of ROS and RIJ stations.Plasma speed and proton density, shown in Figure 8, have increased before substorm 3.These plasma properties and the associated magnetic fluctuations can be thought of as causing substorm 3.This substorm is more dynamic and causes large differences in the north-south direction.During this substorm, the largest derivatives of the stations for the H-component are −296 nT/3 min (ROS), −234 nT/3 min (RIJ), and −218 nT/3 min (KIR).While the corresponding values for the Z-component, shown in Figure 14 ).With the exception of KIR station, where horizontal absolute average derivatives reach 37 nT/3 min, the strongest average changes in the magnetic field occur in the east-west component during the substorm 2. These strongest average changes occur in the horizontal H-component at all the stations during the substorms 1 and 3.

Discussion
We analyze the newly digitized SMA network data, which was recovered by using a DigiMAG method.Data usability has long been limited by time-consuming optical recording and difficult digitization mentioned by Küppers et al. (1979) and Araki et al. (1992).A custom-built device and DigiMAG method developed allows data to be recovered efficiently.The largest storm of the year 1977 was selected to be digitized.
The AE index is available at WDC for Geomagnetism at Kyoto (Nose et al., 2015a).The AE index for 1976 (January-April), 1977 (January-December), 1988 (July-December), and 1989 (January-February and April-December) have not yet been derived.There is an urgent need to rescue old data.Old data have benefited space weather studies even for the recent events (Hayakawa et al., 2021;Knipp et al., 2016).Space weather activity has been measured in the Scandinavian region for a long time (Hayakawa et al., 2019;Nevalinna, 2006).
When a value of Dst min = −159 nT is measured at the equator, a magnetic storm is classified as strong (Loewe & Prölss, 1997).Storm-time recordings were digitized from four SMA stations: Rostadalen (ROS), Evenes (EVE), Ritsemjokk (RIJ), and Kiruna (KIR).As a permanent magnetometer station, Abisko (ABK) is an important data source for geophysical research.It is located on the edge of the auroral zone during moderately active times, and in this study we studied how the magnetic field changes in its vicinity.Advanced artificial intelligence (AI) could convert photographic data into numerical form.More than half of the data does not contain intersections and is thus digitizable with this method.Over 22,000 days of data could be read without solving this challenge.Together with DigiMAG, this would satisfy the need for automatic digitization proposed by Araki et al. (1992) and would allow for statistical analysis mentioned by Küppers et al. (1979).More advanced AI methods would be needed for digitizing the data with intersections.
Our results show strong and localized disturbances in the magnetic field in the vicinity of the 65° latitude during the magnetic storm on 24-31 October 1977.The changes in the horizontal component occur almost simultaneously with the changes in the Dst index.The horizontal differences reach more than 500 nT over a distance of 167 km.This indicates that a dense network of magnetometers would be needed to measure rapid localized changes and to study geomagnetic activity close to the auroral oval.The highest values are measured when the measurement capacity of SMA instruments has been at its maximum.Individual station data near the data gap should be viewed critically.A comparison of the digitized data to the permanent magnetometer data shows that the digitization of the storm period has been successful overall.Compared to the QTR threshold 11 October 1977, 07:00, the point on the magnetogram to which the change is compared, the lowest digitized H-component of the stations were ROS −1,012 nT, RIJ −1,175 nT, and KIR −1,196 nT.Most likely, the H-component decreases even lower values at all stations during the data gap on October 28, 02:00-03:00.In the IMAGE magnetometer network area, the mean IL intensity of storm-time substorms during years 1997 and 1999 was −665 nT (E.I. Tanskanen et al., 2002).In 3 min, 2.8 nT km −1 difference arose in H-component time derivatives between Rostadalen (ROS) and Ritsemjokk (RIJ) stations.The distance between ROS and RIJ stations was 167 km.Identified substorms are strong and large derivatives occur in all three components of the magnetic field.Applications such as navigation, aviation, energy supply, and telecommunications need information on the magnetic field strength.Our results show that the direction of the change in the intensity of the magnetic field can be very different over a short distance.The PdNd method enables comparison of rapid magnetic field changes during substorm phases.During substorm 2 studied, the growth and recovery phase has stronger effect to the H-component at higher latitude station Rostadalen (ROS), than the lower latitude stations Ritsemjokk (RIJ) and Kiruna (KIR).
Figure 16a shows the time interval when the measuring range of the type Münster Gough-Reitzel magnetometer was exceeded.The instrument recorded small fluctuations during the recovery phase.These appears as small spikes in the trace registration and shown in Figure 16c.Due to missing component trace registration, the H-component of EVE station could not be digitized.When the two missing component trace registrations and the exceeding of the instrument's measuring capacity are removed, the amount of valid data for the examined period is 82%.We were able to collect valid data values from SMA magnetograms using the DigiMAG method.Figure 17 shows the comparison of the digitized data and the data measured at ABK station.Consistency indicates that digitization has been successful.

Conclusions
1. DigiMAG method works.It was developed for digitizing magnetic data from the 35 mm film as used by the SMA magnetogram systems.The film reels are becoming obsolete and there is an urge to save the valuable data.The year 1977 is the only full year for which an AE index has not been produced since 1957.2. Differences in the magnetic field's horizontal component around a latitude of 65° can be more than 500 nT over a distance of 167 km.This difference is shown in Figure 9d, and it arises between the two SMA-station locations in (ROS) and Ritsemjokk (RIJ).We cannot assume that the difference between the stations is linearly distributed.If an application requires information on the intensity of the disturbance with an accuracy of less than 500 nT, the station spacing of the magnetometer array used by the application must be less than 200 km.When considering the data measured by the field magnetometer, an external, man-made interference cannot be ruled out.This highlights the importance of observatory quality magnetic measurements and proper data recording methods, for both short-and long-term magnetic environment monitoring.3.In the vicinity of 65° latitude, the magnetic environment can change rapidly within few minutes.Three-minute changes cannot be estimated in different locations by using the data of the nearest station.The magnetic field's strength can increase in any direction no matter how the field changes at a station located 167 km away.These differences appear in the difference curves of the time derivatives.The largest horizontal difference Δ|dH/dt| = 2.8 nT km −1 , see Figure 13, is measured between ROS and RIJ stations.The largest differences in other two directions, see Figures 14 and 15, are Δ|dZ/dt| = 3 nT km −1 (ROS-RIJ) and Δ|dD/ dt| = 5.1 nT km −1 (KIR-RIJ).

Figure 1 .
Figure 1.SMA station map in corrected geomagnetic coordinates (CGM) for the year 1977.The circles show the locations of the SMA stations from the University of Münster.The stations marked in blue were around the permanent AE magnetometer station Abisko (ABK) and are under study in this article.Over 60° (CGM) latitude, on average approximately, the distance between the stations in north-south direction was 125 km and in east-west direction 155 km.

Figure 2 .
Figure 2. (a) SMA network in CGM coordinates for the year 1977.The red and blue circles show the locations of the SMA stations.The blue square indicates the location of the permanent AE magnetometer station ABK.The inner orange circle shows a distance of 100 km from ABK and the outer 200 km.(b) EISCAT magnetometer cross in CGM coordinates for the year 1983.(c) IMAGE stations in CGM coordinates for the year 2020.The red and black circles show the locations of the IMAGE stations.The AE station ABK joined the IMAGE magnetometer network in 1998.The black-marked LEK station closed in 2005.The CGM coordinates are calculated according toGustafsson (1970Gustafsson ( , 1984) ) andTsyganenko et al. (1987).

Figure 3 .
Figure 3.The time intervals for available SMA data.The stations are arranged according to the 1977 CGM latitude.Orange intervals from the stations Rostadalen (ROS), Evenes (EVE), Ritsemjokk (RIJ), and Kiruna (KIR) were recorded around the permanent AE magnetometer station Abisko (ABK).

Figure 5 .
Figure 5.A recording measured at ROS station 27 October 1977.The arrows after the component marks H, D, and Z indicate the direction of the positive amplification of the component.At 13:45, the vertical Z-component decreases rapidly, and the Z a auxiliary trace rises above the B2 baseline.Usually, on the SMA recordings, the Z-component trace is visible above the B1 baseline, and Z a auxiliary trace decreases below the B2 baseline only during the strong disturbances.

Figure 6 .
Figure 6.The Dst and low-resolution OMNI (LRO) data.(a) All magnetic disturbances in 1977 with Dst min ≤ −100 nT.These Dst indices were provided by the WDC for Geomagnetism, Kyoto(Nose et al., 2015b).The index reaches its highest negative value in 1977 during the disturbance period marked in blue.The value was 7 hr continuously lower than the peak value of the disturbance period marked in yellow (b, c, d) SW plasma parameters measured by IMP-7 (Explorer 47), IMP-8 (Explorer 50), and ISEE-1 spacecrafts.(e, f, g) Average scalar field magnitude B, Bz (GSE), and Bx (GSE) measured by IMP 8.The OMNI data were obtained from the GSFC/ SPDF OMNIWeb interface at https://omniweb.gsfc.nasa.gov.

Figure 8 .
Figure 8.The Dst, LRO plasma parameters, and merged IMP-8 1-min data.The vertical columns divide the storm period into three substorms and the horizontal box shows when IMP-8 is on the night side of the Earth in GSE coordinates.(a) Dst index.(b, c, d) LRO plasma parameters.(e, f, g) SPDF 1-min average magnetic field parameters Scalar B, Bz (GSE), and Bx (GSE).

Figure 10 .
Figure 10.(a, b, c) Digitized Z-component traces and the Dst index.The Z-component of KIR station could not be digitized due to missing component trace registration.The difference curves are shown for (d) ROS and EVE, (e) ROS and RIJ, (f) EVE and RIJ.

Figure 11 .
Figure 11.(a, b, c, d) Digitized D-component for ROS, EVE, RIJ, and KIR together with the Dst.The difference curves are shown for (e) ROS and EVE, (f) KIR and RIJ, (g) ROS and RIJ, (h) KIR and EVE, (i) ROS and KIR, and (j) EVE and RIJ.

Figure 12 .
Figure 12.Time derivatives of H-component from 3 min resolution data for (a) ROS, (b) RIJ, and (c) KIR.The difference curves between these three stations are shown in panels (d), (e), and (f).The absolute average time derivatives  |∕| are shown in the lower left corner of each panel.The average absolute differences between the time derivatives of the stations are denoted  △|∕| .

Figure 13 .
Figure 13.Digitized H-components of (a) Rostadalen (ROS), (b) Ritsemjokk (RIJ).The difference of ROS and RIJ is shown in panel (c).Time derivatives of H-component from 3 min resolution data for (d) ROS, (e) RIJ.The difference in time derivatives of ROS and RIJ stations is shown in panel (f).

Figure 14 .
Figure 14.Time derivatives of Z-component from 3 min resolution data for (a) ROS, (b) EVE, and (c) RIJ.The difference curves between these three stations are shown in panels (d), (e), and (f).The absolute average time derivatives  |∕| are shown in the lower left corner of each panel.The average absolute differences between the time derivatives of the stations are denoted  △|∕| .
The film reels examined have been recorded during the following time periods: 19 August 1977 to 02 November 1977 (KIR), 20 August 1977 to 01 November 1977 (RIJ), 22 September 77 to 01 December 1977 (EVE) and 22 September 1977 to 04 December 1977 (ROS).These recording intervals overlap during 22 September 1977 to 01 November 1977 and we have chosen the QTR from this overlapping recording period.The QTR of this paper is 11 October 1977, 07:00 [UT]