0050A Large Magnetosphere Magnetic Field Data Base00040040D. H. Fairfield and N. A. Tsyganenko0078NASA  Goddard Space Flight Center, Laboratory for Extraterrestrial Physics0014Code 695, 0023Greenbelt, MD 2077100040017A. V. Usmanov0020Inst. of Physics0055University of St. Petersburg, St. Petersburg 19890400040016M. V. Malkov0031Polar Geophysical Institute0026Apatity 184200, Russia0014Abstract: 00040068A large magnetosphere magnetic field data base compiled from the0069measurements of 11 Earth-orbiting spacecraft during 1966--1986 is0069described. The 11 contributing spacecraft (Explorer 33,  35, Imps00564,5,6,7 and 8, Heos 1, 2 and ISEE 1, 2) provide data0069between 4 and 60 RE and the data have been edited to exclude data0071outside the magnetosphere. Currently the data base totals more than007379,000 records which sum to a total of more than 1600 days of data in0037the magnetosphere. Each of these 0071records contains the average total field vector in geocentric solar0049magnetospheric coordinates, disturbance field0075vector (measured field minus Earth's internal field), time information,0068spacecraft position, AE, AL, Kp and Dst geomagnetic indices, and0020associated solar0076wind and IMF parameters when available. These data have constituted, and0070will undoubtedly continue to constitute the standard data base for0066magnetospheric modeling studies. They are also appropriate for0073statistical studies of the magnetic field and its dependence on solar0074wind quantities. Plans are to update this data base in the future with0036data from additional spacecraft.00040016Introduction00040076Knowledge of the Earth's magnetic field is of critical importance in the0070study of magnetospheric physics. The magnetic field influences the0073motions of electrons and ions whose differing motions are responsible0075for the electric currents that in turn perturb the magnetic field. Over0073the past 30 years numerous spacecraft, sometimes making more than ten^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0070vector measurements/sec for year after year, have made hundreds of0071millions of vector measurements throughout the magnetosphere. These0075measurements were made at many different locations and for a variety of0074solar wind conditions, but at these high sampling rates the resolution0068both in space and time is much finer than what is needed in many0073applications. To reduce this immense quantity of data to a manageable0071size, it is necessary to create averages so that consecutive values0064correspond to significantly different locations and to solar0075wind/magnetospheric conditions that may be different. In this report we0073describe a large magnetosphere magnetic field data base that has been0072constructed from the measurements of 11 different spacecraft over an0075interval of 20 years. Parts of this data base have already been used in0073the construction of average magnetosphere models [Mead and Fairfield,00731974; Tsyganenko and Usmanov, 1982; Tsyganenko, 1987: 1989; Peredo et0067al., 1993].  The data base will continue to be updated and will0068undoubtedly serve for this and other purposes for years to come.00040028History of the Data Base00040073Construction of the large magnetospheric data base began in the early00741970's when Mead and Fairfield [1975] embarked on a project to produce0074the first quantitative magnetosphere magnetic field model derived from0071actual spacecraft measurements. Their modeling techniques were most0067appropriate in the near-Earth region so they limited their data0073compilation to that within 17 RE. They utilized data from 4 eccentric0073Earth-orbiting spacecraft and choose 0.5 RE as a reasonable averaging0069interval for their radially moving spacecraft; such intervals are0073typically 10 to 25 minutes long, depending on radial distance and the0071exact orbit. That choice of the averaging time intervals (T) /orbit0066segments (S) can be justified by noting that the corresponding0064characteristic scales T and S are commensurate with the best^^^^^^^^^^^^^^^^^^^^^^0075temporal/spatial resolution that is "marginally available" from average0071global models. More quantitatively, in the near magnetosphere (4 RE0074<R<10 RE) the smallest characteristic scale length for such persistent0075features as the ring current and the tail current sheet is on the order0073of 1 RE, while the characteristic time scale for buildup and decay of0071the principal current systems (leaving aside rapid substorm-related0074phenomena) are unlikely to be less than 10-20 min. Such times are also0073commensurate with typical travel times of the Alfven waves within the0017same region. 00040071When Tsyganenko and Usmanov [1982] started developing models in the0071early 1980's they extended the techniques of Mead and Fairfield and0076included models of the magnetotail and ring current.  The data base they0074used was that of Mead and Fairfield augmented by the HEOS 1 and 2 data0076taken during 1969--1974 which covered the high-latitude magnetosphere in0074the northern hemisphere [Hedgecock, 1975; Hedgecock and Thomas, 1975].0069In later efforts [Tsyganenko, 1987; 1989], the modeled region was0068extended tailward, and that required including data taken in the0071magnetotail by IMP spacecraft. The tail data were obtained from the0068National Space Science Data Center (NSSDC) in the form of hourly0073averages for X < --15 RE and (Y2+Z2)1/2 < 30 RE. The last requirement0076was imposed in order to crudely eliminate measurements taken outside the0074magnetosphere; a more accurate selection was carried out when creating0070the final data set, based on an empirical analytical model for the0070average magnetopause shape [Tsyganenko, 1976, 1981]. Actually, the0074magnetopause is a very dynamic boundary and therefore tail data should0072be selected for individual orbits after viewing high resolution data0072plots. However, due to lack of plotting facilities in Russia at that0072time, it was only possible to use the formal selection method with a^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0076single analytical model of the magnetopause. This method undoubtedly led0071to omission of some magnetospheric data near the tail boundary and,0074conversely, to erroneous inclusion of magnetosheath measurements taken0071during strong compressions and/or transient deflections of the tail0075during periods with significant non-radial flow of the solar wind. Also0072intervals of bad data produced by telemetry noise or other anomalies0030were undoubtedly included.00040076Subsequently Tsyganenko and Malkov [see Peredo et al., 1993] added  ISEE00751/2 data to the data base while Fairfield independently added HEOS data0072along with additional Imp 6 data to the original Mead-Fairfield data0073base. Below we describe the additional editing of these data sets and0076their merging into one large magnetosphere data base. Characteristics of0055the various contributing spacecraft are as follows.00040027Contributing Spacecraft00040075The Explorer 33 spacecraft was launched in July 1966 and July 1967 with0072the objective of entering lunar orbit. Explorer 33 failed to achieve0074lunar orbit but provided data from a highly eccentric Earth orbit with0075apogee near 80 RE. The magnetometer experiment [Ness et al., 1967a] was0074a triaxial fluxgate magnetometer, as were all instruments contributing0072to the data base described here. The spacecraft spun with an average0073spin period near 2.4 s about an axis oriented near the ecliptic plane0073and provided a vector field sample every 5.1 s. Only Explorer 33 data0034within 17 RE is included here.00040070Explorer 35 was launched into a lunar orbit in July 1967 and hence0075provided distant tail measurements for two to three days every month as0072the moon passed through the magnetotail.  The spacecraft spun with a0076period of 2.3 s about an axis oriented perpendicular to the plane of the0076ecliptic. The spacecraft automatically switched between ranges of 24 and007564 nT and magnetic field vectors were sampled every 5.1 s [Ness et al.,^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^00741967b]. The Explorer 35 spacecraft flipper, used to invert the sensors0074and help correct instrument zero levels, failed after almost two years0076of operation making the subsequent zeros somewhat less certain. Although0069the spacecraft produced data for some 5 years, only data from the0041initial two years are included here. 00040073The Imp 4 spacecraft was launched in May 1967 into an eccentric orbit0074with an inclination of 67.4 and an apogee of 31.4 Re near the ecliptic0071plane. Imp 4 was a spinning spacecraft with an initial spin rate of00722.587 sec and an orbital period of 4.32 days [Fairfield, 1969].  The0075magnetometer provided a vector measurement every 2.556 s and had 32 and0074128 nT sensitivity ranges that were controlled by ground commands. The0073low sensitivity range allowed measurements to be made as close to the0073Earth as approximately 6.5 RE. Occasions when unusually strong fields0073saturated the instrument in the low range were edited out of the data0076set. The instruments were mounted on 6 foot booms and the spacecraft was0074magnetically clean so zero levels are thought to be accurate to better0073than 0.3 nT. The magnetometer experiment provided vector measurements0072from launch until a solar sensor failure in March 1969 that occurred0066just two months prior to spacecraft reentry to the atmosphere.00040074The Imp 5 spacecraft was launched in June 1969 into an eccentric orbit0071with inclination 86.8 and an apogee of 28.7 RE [Fairfield and Ness,00741972]. Although its apogee was near the ecliptic plane, Imp 5 was near0070the noon meridian at the time of maximum dipole tilt which allowed0074measurements of the polar cusp at geomagnetic latitudes as high as 75 0073beyond 6 RE. The experiment was almost identical to Imp 4 except that0074the ranges of 40 and 200 nT allowed measurements as close to the Earth0074as 5.5 RE. The Imp 5 magnetometer provided 381 orbits of data prior to0040spacecraft reentry in December 1972.0004^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0075The Imp 6 spacecraft was launched in March 1971 into an eccentric orbit0076with inclination 28.7 and an apogee of 33.1 RE. Imp 6 underwent one spin0071every 11.1 seconds and the orbital period was 4.18 days [Fairfield,00641974]. The experiment provided data at a rate of 12.5 vector0073measurements/s. Instrument ranges of 16, 48, 144, 432 nT were changed0064automatically depending on the measured field, thus allowing0073measurements as close to the Earth as 4 RE without data losses due to0076the instrument being in an inappropriate range. The sensors were located0076on the end of a 13 foot boom and a spacecraft contamination field was no0074problem. Zero levels are thought to be accurate to 0.3 nT in the least0076sensitive range and better in more sensitive ranges. Imp 6 provided over0064300 orbits of data until spacecraft reentry in October 1974.00040076The Imp 7 and 8 spacecraft carried identical instruments [Lepping et al.0074private communication; IMP-8 solar wind magnetic field and plasma data0068in support of Ulysses -- Jupiter encounter: 13--31 January 1992,0072Appendix A, 1992]. Imp 7 was launched in September 1972 and provided0073data until an instrument failure in April 1973. Imp 8 was launched in0075October 1973 and the magnetometer continues to provide data at the time0072of this writing 20 years later. Both spacecraft had roughly circular0075orbits near 35 RE; actual apogee and perigee distances varied, but were0072generally near 40 and 30 RE respectively. The orbital period of both0073spacecraft was roughly 12 days. The spacecraft usually resided in the0073magnetotail for 2 to 3 days each orbit, but during some seasons IMP 80076passed north or south of the tail and provided little magnetotail data. 00040069The HEOS 1 and 2 spacecraft were launched on December 5, 1968 and0076January 31, 1972, respectively, into orbits with initial apogees of 35.80068and 38.7 RE respectively [Hedgecock, 1975]. The initial ecliptic0072latitudes of the two apogee were 51.5 and 73.5 and both increased by^^^^^^^^^^^^^^0068more than 10 degrees thus providing unique coverage of the polar0074magnetosphere not available from the IMP spacecraft. HEOS 1 and 2 each0064carried a single range instrument with ranges 64 and 144 nT,0066respectively, that provided measured vectors every 48 and 32 s0074respectively. HEOS 2 re-entered the Earth's atmosphere in August 1974;0056HEOS 1 continued to operate but tracking diminished.00040074The ISEE 1 and 2 spacecraft [e.g. Ogilvie et al., 1978; Russell, 1978]0073were launched in October 1977 into almost coincident eccentric orbits0071with apogee of nearly 23 RE, orbital periods of about 2.4 days, and0074orbit inclination of 29. The spacecraft were placed in a spinning mode0074with 3 s spin period and the spin axis aligned nearly perpendicular to0070the solar ecliptic plane. The spacecraft provided data until their0074reentry in September 1987. For now, we include data only for the first0074four years of the experiment, from the beginning of 1978 to the end of00671981. The magnetometer experiments onboard both ISEE spacecraft0065[Russell, 1978] were triaxial fluxgate instruments having two0075commandable ranges, 8192 and 256 nT. Our data set is constructed from 10074min average tapes from ISEE 2 (January 1978 --January 1980) and ISEE 10070(January 1980 -- December 1981) obtained from NSSDC . Due to fixed0068orientation of the ISEE orbits in inertial space, the spacecraft0073provided magnetotail data between February and May, while the dayside0059magnetosphere was swept mainly from August to November.00040020Data Description00040076The current data set that we describe in this paper is basically the R <007317 RE IMP half-Earth radii averages within 17 RE of the Earth and the0076IMP hourly averages beyond X=-15 RE which are augmented by HEOS and ISEE0072averages computed in a slightly different manner (see below). We now0047describe how the data base was constructed.00040074 The initial compilation of the large magnetosphere data utilized data^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0075from 4 eccentric Earth-orbiting spacecraft (Imps 4,5,6 and Explorer 33)0070[Mead and Fairfield, 1975]. Data that had already been averaged to0075resolutions ranging from 15 to 81 seconds was used as the initial data.0076This data was plotted for individual inbound and outbound orbital passes0074and at the same time averages of the magnetic field in were calculated0076over half-Earth radii intervals separated by half-integral values of the0069geocentric distance in units of Earth radii. Associated ephemeris0075elements were averaged in the same manner. Although the inclination and0076declination values were also averaged at this time and were available in0074an early version of this data, these quantities have been omitted from0072the present data set and only Cartesian component averages have been0072retained. The duration of the averaging periods depend on spacecraft0074orbit and position in the orbit but they generally range from 10 to 250073minutes.  Since modeling the inner magnetosphere was the goal at that0068time, this data set was restricted to distances less than 17 Re.0074Additional IMP 6 data totaling more than 6800 average points was later0074processed in the same manner and added to the original compilation. In0073the R < 17 data the average of individually measured field magnitudes0072(as opposed to the square root of the sum of the square of component0076averages) is not available as it is in most of the remainder of the data0008set.00040076IMP data beyond Xgsm=-15 RE consists of hourly average measurements from0075the magnetic field experiments of N. F. Ness on the spacecraft Explorer006733, and Imps 4,5, 6,7 and 8. Initially data within 30 RE of the0074Earth-Sun line and  X < --15 RE were considered. Then hours containing0071any magnetosheath data or data contaminated by telemetry noise were0075rejected based on visual scanning of data plots which were generally of0070points at 15 or 20 s resolution. Distinguishing the stronger quiet^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0075fields of the tail lobe from fluctuating fields of the magnetosheath is0064relatively straightforward. Distinguishing plasma sheet from0073magnetosheath is more difficult but with the use of a high resolution0072standard deviation that helps identify the presence of magnetosheath0075waves, a decision is often possible. For the Imp 6, 7 and 8 data, plots0069of the plasma data from the Los Alamos National Laboratory plasma0073experiments were usually available to help distinguish the regions in0075ambiguous cases.  Erroneous decisions are more likely for Imps 4, 5 and0076Explorer 35, but attempts were made to reject ambiguous data in order to0076insure a clean magnetotail data set. In addition it was required that at0072least 15 minutes worth of the hours data be present before an hourly0076average was retained. After the IMP 8 data had been processed and edited0075it was discovered that during occasional intervals during 1981-1986 the0076IMP 8 spin axis orientation was in error by as much as a few degrees [R.0073P. Lepping, private communication]. These intervals were removed from0073the data set reducing the available IMP 8 data by something like 15%.00040069HEOS data currently consists of spatially averaged data at 0.5 RE0074resolution that the principal investigator provided to one of us (NAT)0069in the form of listing and which were subsequently transcribed to0074digital form. Although this data has the similar resolution to the Imp0067data, it was averaged in a somewhat different manner. The solar0068magnetospheric coordinate system was divided into 1 RE cubes and0074averages were constructed within various cubes as the spacecraft moved0070through gsm space on each orbit. Successive averages generally had0076radial distances differences between a few tenths of an Earth radius and00741 RE. Times and average positions were provided with the data, but the0075times were center points of the averaging interval and durations of the0069averages were not available. For this reason it was impossible to^^^^^^^^^^^^0074calculate average AE/AL indices over the original averaging intervals.0075To overcome this difficulty it was assumed the data were continuous and0076averages were calculated for times between successive points. Since gaps0076in the data could lead to unrealistically long intervals, an upper limit0072of 35 minutes was placed on the AE/AL averaging time. As is the case0073with the IMP < 17 RE data, the average of individually measured field0071magnitudes is not available in this data set. We plan to eventually0076reconstruct the HEOS data base to make it more consistent with the other0009data.00040072ISEE data have also been processed slightly differently from the IMP0076data. First a crude selection was done by eliminating measurements taken0074well outside the magnetosphere and the magnetosheath, as well as those0072made too close to the Earth (closer than 4 RE). After that, a visual0076inspection of plots of the B components was done, in order to locate the0076magnetopause crossings and make a finer selection of the data within the0065magnetosphere. We did not use any plasma data for identifying0076transitions between the magnetosphere and magnetosheath regimes, relying0076completely on characteristic features in the B plots and using, whenever0074available, information on the solar wind plasma and the interplanetary0076magnetic field for the corresponding intervals of time. The intervals of0068bad data were also identified and excluded based upon the visual0075checking of the magnetic field plots. The next step was to subtract the0074Earth's internal field from the 1--min B averages, using the 10--order0069IGRF expansions and taking into account the secular changes by an0074appropriate updating of its coefficients to given year of measurement.0069After that, averaging of the 1--min external field components and0070spacecraft position along individual orbit segments was performed;0076lengths of the segments were restricted not to exceed 0.5 RE in distance^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0076and 30 min in time. (The latter limitation comes into effect near apogee0066where the spacecraft speed becomes quite small.) An additional0069requirement was that the total number of individual 1--min values0053contributing to each average be no less than six.00040074By comparing simultaneous IMP 8 and ISEE 1 magnetic field measurements0075in the solar wind, we realized that there are intervals when the ISEE 10074instrument can produce spin axis component values that are in error by0073about 2 nT in either the positive or negative sense. This occurs when0074the ISEE 1 instrument is in its insensitive 8192 range in a weak field0073region. Unfortunately there is no range indicator in the digital data0067files and hence these intervals are not readily identified. The0076principle investigator C. T. Russell has made listings of range commands0068available which should allow identification of potential problem0076intervals, but currently no such intervals have been eliminated from our0010files.00040074Hourly average solar wind and interplanetary magnetic field (IMF) data0073compiled by J. H. King at the NSSDC [King, 1977] have been associated0069with each magnetosphere field average when available. If the ISEE0069interval spanned consecutive hours, a linear interpolation of the0075interplanetary data was made for the two hours. If one of the two hours0076was missing the available hour was used. For other spacecraft the hourly0072average associated with the middle of the averaging interval was the0073value selected. Dst and Kp geomagnetic indices from the King data set0071were also included. Although most quantities are well known and are0076further described below, users should note that the solar wind longitude0067angles are given in a coordinate system in which the measured Y0066component of velocity introduced by the Earth's (and measuring0071spacecraft's) motion around the Sun has been removed. Magnetosphere0071researchers who want to know the solar wind direction "seen" by the^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0072magnetosphere should reintroduce this velocity-dependent component. 00040073The 2.5 min or 1 min geomagnetic AE and AL indices were also averaged0073over the same intervals as the individual field averages and inserted0073into the data set. This AE/AL data had been produced at NSSDC for the0071early years, the US National Geophysical and Solar-Terrestrial Data0074Center for subsequent years, and World Date Center C2 for Geomagnetism0076at Kyoto University in Japan since 1978. Since AE data for 1976 and 19770074has never been produced, these years were omitted from the data set so0076that every record in the data set could have an associated AE value. Imp00758 data more recent than 1986 has also not been included due to the lack0036of AE at the time of production.00040076Table 1 lists the amount of available data by spacecraft. Column 1 lists0067the spacecraft and column 2 the interval of data used. Column 30073indicates the number of magnetosphere data points with Xgsm > --15 RE0076and column 4 gives the number of magnetotail averages (Xgsm > --15 RE) .0074Total time of the data from each spacecraft is listed in column 5. The0075number of orbits available for each spacecraft (estimated approximately0070in some cases) is listed in column 6; for the low-altitude-perigee0076spacecraft this number is roughly half the number of radial passes which0060is roughly the number of independent data points in each0076half-Earth-radii interval. The most insensitive range of the instrument,0071as an indicator of the maximum field strength measured, is shown in0013column 7.00040021Data Distribution00040066The data in this data base is given in right handed orthogonal0076geocentric solar magnetospheric (GSM) coordinate system where the X axis0071is the Earth-Sun line and the Z axis is in the plane of the Earth's0076dipole and the X axis. In some case we will refer to solar magnetic (SM)0074coordinates where the Z axis is aligned with the dipole and the X axis0057is in the plane of the Earth-Sun line and the Z axis.0004^^^^^^^0073Figures 1a -- 1c illustrate the distribution of data. Figure 1a shows0068all data, except ISEE, for Xgsm > --15 RE with the distance YZ =0075(Ysm2+Zsm2)1/2 plotted versus Xsm.  The sign of the ordinate is that of0072Ysm. If all IMP data is present within a half RE radial interval the0074average position falls in the middle of this half Earth radii interval0068leading to the concentric circle effect. Missing data during the0074averaging interval can lead to a location between the circles, but the0076majority of these interlying points are from the HEOS spacecraft where a0074different averaging technique was used (see above). Points outside the0075concentric circle are also from HEOS. Note that there will be a lack of0076IMP points on the dawn and dusk flanks in a small region where R > 17 RE0023and Xgsm > --15 RE.00040066Figure 1b shows data complementary to Figure 1a only for solar0074magnetospheric coordinates and for Xgsm < --15 RE. Again ISEE data are0063not included. IMP 5 data inside an apogee near 29 RE can be0075distinguished but IMP 4 and 6 data inside apogees near 31 and 33 RE are0075not so readily distinguished. Explorer 35 contributes all the data near006960 RE and its initial pass out to the moon is clear. IMPs 7 and 80071contribute the rest of the data beyond 33 RE. A small number of IMP0067hourly averages with Xgsm < --15 RE and R < 17 RE that would be0059redundant with the < 17 RE averages, have been removed.00040071Figure 1c shows ISEE data in the same solar magnetic coordinates as0075Figure 1a. The concentric circle effect of Figure 1a is missing because0075of the different averaging method (see above). Note that the 34160 ISEE0072data points represent 42.8% of the total data set even though only 40075years of ISEE data are included. This is because (1) ISEE spends a much0071larger portion of its time inside the magnetosphere compared to the0073higher apogee IMP (Interplanetary Monitoring Platform) spacecraft and0076(2) beyond 15 RE in the tail the ISEE averaging technique often produces^^^^^0054two values per hour compared to one for the IMPs. 00040071Figure 2 illustrates data for Xgsm < --15 RE projected onto a plane0076perpendicular to the Xgsm axis. The vertical coordinate is the estimated0069distance from the tail current sheet using the model of Fairfield0070[1980].  Data from the northern and southern hemispheres have been0071separated according to Bx > 0 or Bx < 0. ISEE data are shown in the0076bottom two panels and the remainder of the data in the top two panels. A007420 RE circle is shown for reference. Coverage of the tail far from the0068equatorial current sheet is due largely to the IMP 8 spacecraft.0068Occasional occurrences of positive (negative) Bx in the southern0076(northern) hemisphere are probably due largely to solar wind flow out of0073the ecliptic plane; such effects will be more important at the larger0071distances and smaller at the closer ISEE distances as appears to be0073true. Note also that with a primarily northward field the positive or0044negative Bx component may be very small.00040072The distribution of data with time and by spacecraft is shown in the0073bottom panel of Figure 3. The solar cycle is indicated by the sunspot0069number in the top panel and the average annual solar wind kinetic0072pressure in the second panel [Fairfield, 1991]. The number of points0072contributed by each spacecraft is indicated by the number associated0073with each label. Spacecraft ID is word number 18 in the record format0072(see below). Note that the solar wind pressure varying by at least a0066factor of 2 over the solar cycle can produce different average0067magnetosphere characteristics at different phases of the cycle.00040076In Figure 4 we show the number of data points per year and the amount of0074associated solar wind and IMF data. Note that the best years for solar0076wind coverage occurred in 1966--1968 and 1972--1974 when many spacecraft0067were in orbit and in 1978--1982 when ISEE 3 was at the upstream^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^0074libration point. In 1983 ISEE 3 was moved to the magnetic tail and its0072solar wind coverage terminated. Note that in 1978-1982 the number of0070data points per year exceeded the number of hours in a year (8760)0074largely because ISEE often provided more than one point per hour. Data0045points beyond 1981 are exclusively IMP 8.00040015Data Format00040070The format of the data is shown in Table 2. The four column in the0069tables are a word number, an arbitrary abbreviated word name, the0072integer format, and a description of the word. The data are in fixed0076point ASCII format and some quantities have been multiplied by 10 or 1000073to preserve decimal accuracy within the integer values. Words 4 and 50070contain the beginning time and duration of the average in table 2,0067except that for HEOS word 4 is the center time of the averaging0069interval.  Note that the average of individual field measurements0076available as word 10 is different that the square root of the sum of the0072squares of the components except for HEOS and IMP < 17 RE data where0074only the latter was available. Words 1--18 contain information related0075to the magnetic field average while words 18--31 contain solar wind and0035geophysical index information. 00040075All data are given in geocentric solar magnetospheric coordinates. Near0075the Earth, solar magnetic coordinates are a more appropriate coordinate0074system and these coordinates are easily obtained by rotating any solar0075magnetospheric quantities about the Ygsm axis using the transformation 00040036Xsm=Xgsm cos(gl) -- Zgsm sin(gl)00040037Zsm = Xgsm sin(gl) + Zgsm cos(gl)00040076where gl is the geomagnetic latitude of the Sun or the dipole tilt angle0013(word 6).00040011Summary00040068The large data base described here is a significant resource for0073investigating the distant magnetic field of the Earth as perturbed by0069the solar wind and IMF. Measurements from 11 different spacecraft0074provide a total of 79,745 records, each with an average magnetic field^^^^^^^^^^^^0075vector measured within the magnetosphere somewhere between 4 and 60 RE.0073Each record also has available the Kp, Dst and AE geomagnetic indices0076and solar wind /IMF values when available (roughly half the time). These0075parameters will allow sorting the data to study the magnetosphere field0071configuration under various solar wind and geomagnetic conditions. 00040072The data base should be of interest to both students and experienced0067researchers. Students should be able to use the data to readily0071demonstrate basic characteristics of the Earth's field. Experienced0072researchers should be able to look at the data in innovative ways to0071answer specific questions and extract new knowledge. Magnetic field0074modelers will want to use words 14 -- 16 which have the internal field0072subtracted out. Users carrying out statistical investigations of the0075field will probably use total field components given as words 7 -- 9 in0073solar magnetospheric coordinates. The data is easily rotated to solar0076magnetic coordinates using the transformations given above. Rotations to0072many other coordinate systems can be accomplished using a package of0076Fortran programs compiled by one of the authors (NAT) which will be made0075available with the data. The data set is about 14 megabytes in size and0075is available from NSSDC. In the future it will probably be available on0010CDROM.00040022Acknowledgments:  00040076Many individuals are responsible for the contributing the vast amount of0074data contained in this data base. Primary among them are the principal0076investigators for the magnetic field experiments: N. F. Ness for the IMP0072experiments (and later R. P. Lepping for IMP 8), P. C. Hedgecock for0074HEOS and C. T. Russell for ISEE. Experimenters too numerous to mention0073contributed plasma and magnetic field data compiled in the OMNI solar0070wind data base by J. H. King and colleagues at NSSDC. The numerous0074indices included with the data have been compiled by various groups at^^^^^^^^^^^^^^^^^^^^^^^^0074various times. Programing support of J. A. Jones is also acknowledged.0072This work was done while one of us (NAT) held a NRC Associateship at0014NASA GSFC.00040019Figure Captions00040075Fig. 1a. Data point locations for non-ISEE spacecraft in sm coordinates0075for Xgsm> --15 RE. YZ is the distance from the Xsm axis. The concentric0070circle effect is due to averaging IMP data in 0.5 RE half integral0064distances. HEOS spacecraft contribute the data beyond 17 RE.00040076Fig. 1b. Data point locations for non-ISEE spacecraft in gsm coordinates0024for Xgsm < ---15RE. 00040067Fig. 1c. Data point locations for ISEE 1 and 2 spacecraft in sm0016coordinates.00040076Fig. 2. Data point location in the YZ plane for Xgsm < ---15 RE where Z`0074is the estimated distance from the equatorial current sheet. Data have0074been separated according to the polarity of the Bx component. ISEE 1,20075data are shown in the bottom two panels and the other spacecraft in the0015top panels.00040071Fig. 3. The era of various spacecraft contributions is shown with a0073horizontal line opposite the appropriate spacecraft ID number. Number0065associated with the labels indicate the number of data points0075contributed by each spacecraft. The solar cycle is indicated by the top0074two traces showing sunspot number, R, and kinetic solar wind pressure.00040073Fig. 4. Number of data points/year for the whole data set. Also shown0055are the number of associated solar wind/IMF values.00040027Table 1. Available Data00040020Table 2.  Format00040014References00040004^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^