Data Set Overview
  =================
    This dataset contains data acquired by the Galileo Magnetometer
    during the Venus flyby. The data have been averaged down to twenty
    second samples from the  7.68 kB Low Rate Science (LRS) tape record
    mode.  The flyby data were recorded and later played back during the
    early portion of the Earth 1 flyby. Limited space on the tape
    recorder required the magnetometer team to limit their high time
    resolution observations to a few short intervals in the
    pre-encounter solar wind, a main record period near closest
    approach, and then some post encounter solar wind data. Large gaps
    in the high rate data have been filled in with 16 RIM (~ 16 minute)
    averages from the optimal averager section of the instrument. These
    data have been fully processed to remove instrument response
    function characteristics and interference from magnetic sources
    aboard the spacecraft. The data are provided in Venus Solar Orbital
    (VSO) coordinates.

    Primary Dataset Reference:

      Kivelson, M.G., C.F. Kennel, R.L. McPherron, C.T. Russell,
      D.J. Southwood, R.J. Walker, C.M. Hammond, K.K. Khurana, R.J.
      Strangeway, and P.J. Coleman, 'Magnetic field studies of the
      solar wind interaction with Venus from the Galileo flyby:
      First Results', Science, 253, 5029, 1518, 1991.
      [KIVELSONETAL1991]

    Primary Instrument Reference:

      Kivelson, M.G., K.K. Khurana, J.D. Means, C.T. Russell, and
      R.C. Snare, 'The Galileo magnetic field investigation', Space
      Science Reviews, 60, 1-4, 357, 1992. [KIVELSONETAL1992]


  Data
  ====

    ------------------------------------------------------------------
    Table 1. Data record structure, VSO Coordinates
    ------------------------------------------------------------------
    Column    Description
    ------------------------------------------------------------------
    year      Year = 1990
    day       Day of year (40 , 41)
    sec       Second of day
    sc_clk    S/C clock counter in the form rim:mod91:mod10:mod8
    Bx_vso    Magnetic field X component in VSO coordinates 
    By_vso    Magnetic field Y component in VSO coordinates 
    Bz_vso    Magnetic field Z component in VSO coordinates 
    Bmag      |B| Magnitude of B 
    stBx_vso  Standard deviation of the X component 
    stBy_vso  Standard deviation of the Y component 
    stBz_vso  Standard deviation of the Z component 
    stBmag    Standard deviation of the Magnitude of B 
    npts      Number of points in the average
    dqf       Data quality flag

    Data Acquisition
    ----------------
      The data for this dataset were all acquired in by the outboard
      magnetometer sensors in the flip left mode in the high field mode
      (ULHR). This mode has a digitization step size of 0.25 nanoTesla.
      However, these data are acquired at 30 vectors/second and then
      recursively filtered in the instrument. The high rate data that
      are recorded to tape have a sample rate of 4.5 vectors/second. If
      there is sufficient variation in the 6-7 input samples that make
      up a single output sample, then the effective digitization step
      size becomes much smaller.

    Data Sampling
    -------------
      The high rate recorded data are not evenly sampled within a minor
      frame. However, these data have been averaged using an averaging
      routine to produce evenly sampled data. Each averaged value
      includes all available data from the prior and subsequent 10
      seconds in the high rate data. Averages are not overlapping. The
      standard deviations and the number of samples in the average have
      been included with the final dataset.


  Coordinate Systems
  ==================
    The Galileo magnetometer Venus flyby data are being archived in
    Venus Solar Orbital (VSO) coordinates. The VSO X direction is taken
    along the Venus-Sun line, positive towards the Sun. The Z direction
    is parallel to the normal of the Venus orbital plane (Venusian
    ecliptic), positive northward, and Y completes the right-handed set
    (towards dusk).


  Data Processing
  ===============
    These data have been processed from the PDS dataset:
    'GO-E/V/A-MAG-3-RDR-IRC-COORDS-HIRES-V1.0'

    In order to generate the IRC processed dataset, the following
    procedure was followed:

      1) Sensor zero level corrections were subtracted from the raw
         data,
      2) Data were converted to nanoTesla,
      3) A coupling matrix which orthogonalizes the data and corrects
         for gains was applied to the data (calibration applied),
      4) Magnetic sources associated with the spacecraft were subtracted
         from the data,
      5) Data were 'despun' into inertial rotor coordinates
      6) Data were transformed into VSO coordinates
      7) Data were averaged to 20 second sampling
      8) Optimal averager data (in VSO coordinates) were merged in to
         fill gaps between record intervals.

      1) Zero level determination: The zero levels of the two spin plane
         sensors were determined by taking averages over a large number
         (about 50) of integral spin cycles. The zero level of the spin
         axis aligned sensor was determined by a variety of means.
         First, since the spin axis aligned sensor can be flipped into
         the spin plane, the value of the zero level determined in the
         spin plane can be used in the other geometry. This works well
         if there are no spacecraft fields and the zero level is stable.
         If there are spacecraft fields present which remain constant
         over relatively long time periods (many hours), then another
         method of zero level determination is used. The spacecraft spin
         axis is along the Z direction, the data in the X and Y
         directions have already had zero level corrections applied.

            Bm(z)  = B(z)   + O(z)
           |Bm|^2  = B(x)^2 + B(y)^2 + Bm(z)^2
                   = B(x)^2 + B(y)^2 + B(z)^2      + O(z)^2 + 2B(z)O(z)
                   = |B|^2  + O(z)^2 + 2B(z)O(z)
                   = |B|^2  + O(z)^2 + 2O(z)[Bm(z) - O(z)]
                   = |B|^2  - O(z)^2 + 2O(z)Bm(z)

            m = measured value    -    no subscript = true value

         Now if |B| remains constant over a short interval and O(z)
         remains constant over a much longer interval, we can take
         averages and reduce this equation to:

           |Bm|^2 - <|Bm|^2> =  2O(z)[Bm(z) - ]

           <> indicates average value

         Data can be processed using short averages of |B| until many
         points are accumulated and then fit with a line in a least
         squares sense. The slope of this line is twice the required
         offset. The scatter in the data give an indication of the error
         in the assumption the |B| and O(z) have remained constant.
         Intervals with large rms errors are not retained. A file which
         contains zero levels as a function of time has been provided as
         an ancillary product with this dataset.

      2) Conversion to nanoTesla simply requires dividing the instrument
         data numbers by a constant scale factor. For the inboard high
         range (low gain) mode the scale factor is 2. For the inboard
         low range and outboard high range, the scale factor is 64. The
         outboard low range data has a scale factor of 1024.

      3) Calibration matrix applied: The determination of a calibration
         matrix is too complex to describe here. The method employed has
         been well described in [KEPKOETAL1996].

      4) After the data were initially processed (calibrated and
         despun), it was clear that there were still coherent noise
         sources remaining in the data. Dynamic spectra of the
         magnetometer data revealed coherent energy at high order (2nd,
         3rd, 4th) harmonics of the spin period as well as some
         subharmonic frequencies. High order harmonics of the spin
         period can be generated by spinning about a fixed dipole source
         such as a source on the despun platform. Subharmonic energy can
         be created by a dipole source which spins with magnetometer but
         changes orientation at a frequency which is near the spin
         frequency. The source of the high order harmonics was modeled
         using 2-D (clock and cone angle). Fourier transforms of high
         pass filtered data. This allows us to resolve the source in
         terms of the relative spin phase and look direction of the scan
         platform. Model fields associated with this source
         (approximately 0.15 nT at the inboard sensors in the lowest
         harmonic) have been subtracted from the data. A similar
         approach was taken for the isolation and removal of sources of
         subharmonic energy. Data were band pass filtered to isolate the
         source signature and then resolved into components as a
         function of the Energetic Particle Detector (EPD) motor
         position (look direction). EPD interference (at about 0.05nT)
         has been removed from the data on Dec 8, 1990. Both sets of
         interference coefficients were calculated using data from the
         inboard sensors. When the outboard sensors are in use, these
         values are extrapolated using the inverse power law appropriate
         for the source of each term.

         It should be noted here that both of these interference
         corrections are less then the quantization level for the
         inboard sensors. Data resolution coming out of the recursive
         filter can actually be better than that coming out of the A/D
         converter if there is sufficient noise at the single bit level.

      5) Despinning: Data are despun and checked in inertial rotor
         coordinates before transforming to geophysical coordinates. Any
         errors in the processing will be most readily apparent in
         inertial rotor coordinates.  The  nominal transformation to IRC
         from SRC is

            (Bx)   ( cos(theta)   -sin(theta)  0 ) (Bxs)
            (By) = ( sin(theta)    cos(theta)  0 ) (Bys)
            (Bz)   (   0               0       1 ) (Bzs)

         Where s denotes spinning coordinates and theta is the rotor spin
         angle.

         Frequency dependent phase delays associated with the analog
         anti-aliasing filter and the digital recursive filter have been
         removed during the despinning of the data. The dominant
         frequency in the spinning data is at the spacecraft spin
         frequency. The phase angle delay associated from all known
         sources is computed at the spin frequency and removed from the
         data during despinning.

         Analog Filter:            Digital Filter (Nyquist Freq Fn = 15Hz):

               1543                        1/3
         __________________        _____________________
         s^2 + 55.5s + 1543        4/3 - exp(-PI*i*f/Fn)

         s = 2*PI*i*f

         Imaginary = 55.5s         Imaginary = -sin(PI*f/Fn)
         Real      = 1543 + s^2    Real      = 4/3 - cos(PI*f/FREQ_N)


         f = frequency
         delay = tan^-1(Im/Re)

         In addition, there is an electrical delay associated with the A/D
         conversion of about 1 millisecond. This delay is converted to an
         angle using the instantaneous spin frequency. These 3 sources of
         delay are then summed in to the quantity 'delay' and then the
         despinning matrix becomes:


            (Bx)   ( cos(theta - phase)   -sin(theta - phase)  0 ) (Bxs)
            (By) = ( sin(theta - phase)    cos(theta - phase)  0 ) (Bys)
            (Bz)   (      0                          0         1 ) (Bzs)


         In order to create the processed VSO dataset the following
         procedure was used.

         Data are transformed to geophysical coordinates: Data are
         transformed from inertial rotor coordinates to the Earth Mean
         Equatorial (equinox 1950) coordinate system. This system is
         directly supported by the SPICE software provided by the
         Navigation and Ancillary Information Facility (NAIF) at JPL as
         inertial coordinate system 'FK4'. The angles required for this
         transformation come directly from the Galileo Attitude and
         Articulation Control System (AACS) data. The transformation
         matrix for this rotation is:

      --                                                             --
      |(cosTsinDcosR - sinTsinR)  (-sinDsinTcosR - cosTsinR)  cosDcosR|
      |                                                               |
      |(cosTsinDsinR + sinTcosR)  (-sinDsinTsinR + cosTcosR)  cosDsinR|
      |                                                               |
      |-cosDcosT                  sinTcosD                      sinD  |
      --                                                            --

         where

           R = Rotor-Right Ascension
           D = Rotor-Declination
           T = Rotor-Twist - Rotor-Spin-angle  (despun data)

      6) Once in an inertial coordinate system, SPICE software provides
         the subroutine which returns the transformation matrix to VSO
         (G_GSETRN) The spacecraft/planet (SPK) , leap second (TS), and
         planetary constants (PCK) kernels required for these
         transformations have been archived in the PDS by NAIF. These
         SPICE kernels are available on the CD_ROM which contains the
         magnetometer data. The SPICE toolkit (software) can be obtained
         from the NAIF node of the PDS for many different platforms and
         operating systems.

      7) Standard non-overlapping 20 second averages and standard
         deviations were computed about the central time stamp.

      8) Optimal averager data were taken continuously across during the
         cruise period near the Venus flyby. These data were corrected
         for offsets and phase delays associated with the averaging. The
         data were then transformed into VSO coordinated and merged with
         the 20 second averages, preserving only the data values during
         the record interval gaps.

Confidence Level Overview
  =========================
    The sources of error which are the most significant are those
    associated with magnetic sources aboard the spacecraft, especially
    those with temporal or spacecraft orientation variations. The next
    greatest contributor of error is the data from the AACS which
    affects our knowledge of the spacecraft orientation and hence
    rotates the magnetic field vector. Lastly, telemetry or software
    errors which produce 'spikes' or bit errors in the data are error
    sources.

    In regions where the magnetic sources associated with the spacecraft
    are fairly constant, magnetic interference is probably reduced by
    data processing to better than 0.05 nT at the inboard sensors. In
    these same regions, sensor zero levels (offsets) are known equally
    well. The data processing software does a fairly good job of
    removing all currently identified sources of magnetic interference.
    However, there are some time intervals when the zero levels of the
    spin plane sensors show large variations (1-5 nT) on short time
    scales (minutes to hours). After a while (hours usually) the offsets
    return to their nominal levels. The source of these magnetic fields
    has not yet been identified. The method of removing offsets from the
    spin plane sensors does remove these effects, but the method of
    determining the spin axis aligned sensor offsets does not.  In
    regions where large variations are detected in the spin plane
    sensors it is reasonable to assume that similar variations are
    taking place in the spin axis aligned sensor. The time period
    between Feb 10, 05:40 through 08:30 is the only interval in this
    dataset which shows rapid time varying offsets in the spin plane
    sensors.

    Our data processing software creates a data quality flag (dqf) which
    is an assessment of AACS and telemetry error source contamination of
    a given data point. This number is binary integer where each bit
    indicates the presence or absence of some error source. The number 0
    represents the absence of all error sources which are tested. The
    higher order bit (larger number) error sources are considered to be
    more significant error sources. Data are examined for gradients in
    the field which might be associated with telemetry bit errors, for
    regions of bad AACS angles, and for completely missing data. If the
    error is considered completely unrecoverable, the data values are
    replaced with a missing data flag. In the case of a flag in the
    rotor spin angle, the vector components may be flagged but the
    magnitude is still valid. Here is a list of all of the error checks
    and the bits they set in the dqf field.

    DQF_GOOD_DATA                 0  Good data
    DQF_BX_GRAD_WARNING         2^0  Component gradient warning
    DQF_BY_GRAD_WARNING         2^1  Component gradient warning
    DQF_BZ_GRAD_WARNING         2^2  Component gradient warning
    DQF_INTERP_ROTATTR          2^3  Missing rotor RA interpolated
    DQF_INTERP_ROTATTD          2^4  Missing rotor DEC
                                     interpolated
    DQF_INTERP_SPINDELT         2^5  Missing rotor Spin Delta
                                     interpolated
    DQF_INTERP_SCRELCON         2^6  Missing Relative Cone angle
                                     interpolated
    DQF_INTERP_SCRELCLK         2^7  Missing Relative Clock angle
                                     interpolated
    DQF_INTERP_ROTATTT          2^8  Missing rotor Twist interpolated
    DQF_INTERP_SPINANGL         2^9  Missing rotor Spin interpolated
    DQF_ROTATTR_FLAG            2^10 Missing rotor RA flagged
    DQF_ROTATTD_FLAG            2^11 Missing rotor DEC flagged
    DQF_SPINDELT_FLAG           2^12 Missing rotor Spin Delta flagged
    DQF_SCRELCON_FLAG           2^13 Missing Relative Cone angle flagged
    DQF_SCRELCLK_FLAG           2^14 Missing Relative Clock angle
                                     flagged
    DQF_ROTATTT_FLAG            2^15 Missing rotor Twist flagged
    DQF_AACS_TELEMETRY_HIT_FLAG 2^16 Telemetry hit in AACS record
    DQF_MAG_TELEMETRY_HIT_FLAG  2^17 Telemetry hit in mag record
    DQF_SPINANGL_FLAG           2^18 Missing rotor Spin flagged

    DQF_BX_GRAD_ERROR           2^25 Component gradient error
    DQF_BY_GRAD_ERROR           2^26 Component gradient error
    DQF_BZ_GRAD_ERROR           2^27 Component gradient error
    DQF_BX_FLAG                 2^28 Component flagged
    DQF_BY_FLAG                 2^29 Component flagged
    DQF_BZ_FLAG                 2^30 Component flagged


    Magnetic field gradient warning or error levels are set during the
    data processing according to expected variances depending on the
    region of space.  In the solar wind, gradient warnings are usually
    issued at gradients of 10 nT/sec and errors at 15 nT/sec. In the
    magnetosheath, these values may be 50 percent larger. In the inner
    magnetosphere, these dqf flags may be completely turned off.
    Similarly, AACS angles are interpolated across gaps during the
    processing if the gap length is relatively short (less than 10
    minutes typically). If the gaps in spacecraft attitude are long, the
    AACS angles are flagged and not interpolated.

    Errors associated with AACS angles have various effects on the data.
    The rotor right ascension and declination are crucial to the
    understanding of the spacecraft orientation. Fortunately, these
    angles are slowly varying and can be interpolated to better than 1
    degree of accuracy for long (many hour) time periods except near
    major spacecraft maneuvers. The relative clock and cone angles are
    used to remove scan platform interference. In their absence, no
    interference is removed (+/- 0.15 nT error possible in each
    component). The rotor motion spin delta is used to determine the
    instantaneous spin frequency for the phase delay computation. In its
    absence, the last known phase delay is applied to the current data
    point. The rotor spin angle and twist angle must be present in order
    to despin the data. These angles are generally not interpolated for
    more than ten minutes because the rotor spin period drifts over time
    periods on this order.