Field-Plasma-Merged 1-min IMP 8 Data Set 1. Introduction Note added January, 2010. In 2009, a new 15.36s IMP 8 magnetic field data set was created that supersedes the version originally used in 2005 for creating the 1-min merged data set discussed herein. Records in the new 15.36s data set have uniformly spaced time tags every 15.36 sec, with most averages built from 320ms values whose time tags fall within the 15.36s interval. The new 15.36s data set is discussed at ftp://spdf.gsfc.nasa.gov/pub/data/imp/ imp8/mag/15s_ascii_v3/00_IMP8_15s_data_docum.txt. We created this month a new 1-min merged data set to replace its 2005 predecessor. For each minute, we took the spacecraft position and plasma data straight from the 2005 data set. But we made new 1-min magnetic field averages and variances using data from whichever 15.36s records had the majority of their time spans in the minute of interest. That is, 15.36s records whose time tags were between T - 15.36s/2 and T + 60 - 15.36s/2 were used in the 1-min averages for the 1-min record whose time tag was T (expressed in seconds). With this approach, the parameters FCm and DWm, words 12 and 13, become meaningless and assume their fill values in this new 1-min merged data set. --------------------------- This data set was created at GSFC/SECAA in 2005, using as input both the "15.36 s" magnetic field data set and the "1 min" MIT plasma data set. It has one record per minute, regardless of whether there were field and/or plasma data for that minute. In fact, there are no plasma parameters available when the spacecraft is on the magnetosphere side of the magnetopause. Users of this 1 min data set are urged to consult documentation about the input data sets at Magnetic field: ftp://spdf.gsfc.nasa.gov/pub/data/imp/imp8/mag/15s_ascii/00readme.txt Plasma: ftp://spdf.gsfc.nasa.gov/pub/data/imp/imp8/plasma_mit/sw_msheath_min/00readme.txt The IMP position data used in this data set correspond to the ephemeris developed by GSFC's Flight Dynamics Facility in 2005. This occasionally differs from earlier ephemerides for the past ~20 years by up to an Earth radius or so. In addition to presenting the contents and format of the 1-min records of this data set (which has been aggregated into 1-month files), there are three aspects of the data preparation that we explain in detail below: The building of the 1-min averages; The despiking of the MIT plasma data; The flagging of data as being in the solar wind or not. 2. Building the 1-min averages Neither of these input data sets is gap-free, nor do they have uniform spacing between records. We determine and use the time coverage spans on which the parameter values given in the input records are based. For magnetic field data, this corresponds to the time span of the <=12 1.28-s values contributing to the 15.36-s averages. For the plasma data, this corresponds to the accumulation time of the distribution function from which the bulk plasma parameters are derived. For the magnetic field data, the time tags of the input records correspond to the first time point of the last 1.28 s value contributing to the 15.36 s average values. Using the time tag, the number (contained in the input records) of 1.28 s values in the 15.36 s average values, and the assumption that the 1.28 s values contributing to a given 15.36 s average are contiguous, we compute the first and last times (Tf, Tl) for the data in the record tagged with time T to be Tf = T-(N-1)*1.28s, Tl=T+1.28s. N is the number of 1.28 s values in the 15.36 s average. For the plasma data, we know the distribution accumulation times for the various modes. There is a mode flag in each record. We understand from MIT that the record time tag corresponds to the first time in the distribution function accumulation interval. The whole-minute time tags in the output records correspond to the nominal start times (not mid-point times) of the data in the records. Any input record that has any of its coverage span overlapping an output minute of interest is used in the calculation of the parameter values for that minute. Each input record is characterized by the duration of (the part of) its span that lies within the minute of interest. Call this Si (i indexes the input record contributing to a given minute's output values). Si = Tl* - Tf* where Tl* and Tf* are last and first times of the coverage of the input record that lie within or at the boundary of the output record. Each input record is also characterized by the extent to which its values were determined by measurements made within the minute of interest. This is Fi = (Tl*-Tf*)/(Tl-Tf). To help readers visualize this, consider the following schematic of two timelines: The first line represents three successive minutes (Ta, Tb, Tc) for which output records are to be generated. The second line represents a series of 15.36-s input records, with time tags (the vertical lines) at T2, T3, T4, T5, T6, T7, T8. The dashes in the second line represent data, and the dots the absence of data. The "...---..." in the input record tagged with T1 had contributing 1.28-s data only for the interior segment of the 15.36-s interval. (We ignore in this illustration that each 15.36-s record is actually tagged with a time immediately preceding its last contributing 1.28 value.) |--------------|--------------|--------------| |-----|...--...|-----|-----|------|......................|-- Three input records will contribute to each of the first two output records, while only one record (that tagged as T4) will contribute to the third output record. For the first output record, the first input record will have Tf=T2, Tl=T3, Tf*=Ta, Tl*=T3. For the second input record, Tf*=Tf=a time between T3 and T4, where the actual data coverage starts, and Tl*=Tl=a time between T3 and T4 where the actual data coverage ends. For the third input record, Tf*=Tf=T4, Tl=T5,Tl*=Tb. It is interesting that the input record tagged as T6 will totally determine the values in the output record tagges as Tc, even though those "T6 values" were largely determined by data taken during the Tb interval and not the Tc interval. For any physical parameter P, we determine its value for the 1-min output record by taking weighted averages of the values of that parameter in all input records having any time coverage overlap with the output record's time span:

= SUM (Pi*Si*Fi) / SUM (Si*Fi). In the output records we also carry along, separately for the magnetometer and plasma data, the numbers N of input records contributing to the output values, the fractional coverage over the minute (FC = SUM (Si)/60) and a measure of the extent to which the computed

values are based on data taken within the minute of interest (DW = SUM (Si*Fi)/SUM (Si). (Nm, FCm, DWm) and (Np, FCp, DWp) are (N, FC, DW) values for magnetic field and plasma data respectively. We create 1-min output records for field and plasma data separately, using the approach above. For output plasma records, we typically get contributions from 1 or 2 input records, while for magnetic field output records, we get typically contributions from 4-5 input records. We do not include variance information for the output plasma data but we do for the output magnetic field data. The magnetic field variances are defined as V = SUM [Wi * (Pi-

)**2] / SUM (Wi) Here the weighting factors Wi are the Si*Fi products introduced above. The output records actually contain standard deviations of the magnetic field parameters, i.e., SQRT(V). 3.Despiking the magnetic field and plasma data The IMP magnetic field data has a significant number of apparent data spikes. The IMP plasma data set also has a significant number of apparent data spikes, sometimes in the nonlinear-fit-based parameter values, sometimes in the moment-based values and sometimes in both. We have undertaken to eliminate the most egregious spikes, while at the same time trying to eliminate the least amount of good data. (There is no certain way to automatically eliminate all spikes and only spikes.) We independently set to our standard fill value all the fit-based parameters (or moment-based parameters) of a given input record if that record has a spike in any one or more fit-based (or moments-based) parameters (flow speed, flow direction angles, proton density and thermal speed). Likewise for the magnetic field data. For any parameter P and for any record, we determine the mean

and standard deviation [sigma(P)] of the P values in the two preceding and two following points. We require that the time interval between the first and last of the points be no more than 15 minutes (magnetic field data) or 60 minutes (plasma data). For P = magnetic field magnitude, flow speed, density or thermal speed, we declare a spike if |P-

| > 4 * sigma(P) AND |P-

| > k(P) *

where k(P) = 0.2, 0.1, 0.3, or 0.3 for P = field magnitude, flow speed, density, or thermal speed, respectively. For P = magnetic field component, flow longitude or flow latitude angle, we declare a spike if |P-

| > 4 * sigma(P) AND |P-

| > k', where k' = 1 nT for field components and k' = 4 deg for flow latitude and longitude. Note that this test is not effective in detecting spikes in the first or second data point after or before gaps in solar wind data (which gaps are not infrequent) nor in finding multi-point spikes (which are infrequent). 4. Region flags Each 1-min record has two region flags. First, we have used the 1-min resolution IMP 8 bow shock database at http://omniweb.sci.gsfc.nasa.gov/ftpbrowser/bowshock.html to flag whether the record is in the solar wind or not, using 0 = solar wind 1 = not solar wind Second, we have included the MIT region flags included in their input data set, wherein each record is flagged as 1 = solar wind 2 = solar wind or magnetosheath 3 = definitely not solar wind Where the input plasma record(s) for a given 1-min output record had a common region flag, the output record contains that flag value. For the rare occasion where this was not the case, the output record has flag = 4. 5. Miscellaneous other Finally, we merge the 1-min magnetic field and plasma data sets to create a 1-min field- plasma-merged data set. The magnetic field latitude and longitude angle values in the 1-min records were determined from 1-min magnetic field Cartesian components. However, for the plasma data, flow latitude and longitude values of (typically 1 or 2) input records were averaged and these averages were used to compute flow vector Cartesian components (except that the output records contain the negative of the Vx value thereby determined since the flow longitude angles have value zero along the -Xgse axis. Thus flow latitude and longitude angles have the same signs as the Vz and Vy components in GSE, i.e., positive for flow from south and west of the sun, respectively. Two more points about the flow angles are important. First, the input plasma records contain plasma parameters determined via both fits and moments. The fit-based flow longitude angles have the aberration caused by the Earth's orbital motion about the sun removed, while the moment-based flow longitude angles do not have aberration effects removed. Our new 1-min product includes both set of plasma parameters and preserves the asymmetry in handling aberration effects. Second, flow latitude angles of the input records have a 2-deg asymmetry about zero. That is, long term averages are ~ 2 deg. This effect is not understood by the MIT team, and has been included in all MIT data sets (at MIT or NSSDC) to date. However, for this new 1-min data set, we have opted to remove this offset by subtracting 2.0 deg from all the input data. Thus long term averages of the flow latitude and Vz values of this data set should be near zero. The plasma temperatures of this 1 min data set have been determined from the thermal speeds (W) of the input data set using the algorithm T (deg K) = 60.5 * W (km/s) **2. The IMP ephemeris information carried in the records of this 1-min data set is not as contained in the input field or plasma data sets but is based on the "new" IMP 8 orbit produced at GSFC around the year 2000 using a decade's worth of IMP shadow information. This new orbit is believed to be more accurate than its predecessor by ~1 Re or even more at times, from ~1990 onward. 6. The contents/format of records are: Wd Format Fill values Parameters Units, Comments 1 I4 9999 Year 1992,1993,... 2 I4 999 Doy 1-366 3 I3 99 Hour 0-23 4 I3 99 Minute 0-59 5 I2 9 Solar wind flag 0,1 See above 6 F8.2 9999.99 X, GSE IMP position component, Re 7 F8.2 9999.99 Y, GSE " 8 F8.2 9999.99 Z, GSE " 9 F8.2 9999.99 Y, GSM " 10 F8.2 9999.99 Z, GSM " 11 I2 9 Nm, Number of mag field points 12 I3 99 FCm, Coverage within minute See above 13 F5.2 9.99 DWm, Interior fraction See above 14 F8.2 9999.99 Field Magnitude Average <|B|>, nT 15 F8.2 9999.99 Magnitude of Avg Field Vr ||,nT sqrt(^2+^2+^2) 16 F8.2 9999.99 Lat. Angle of Avg Field Vr,GSE Deg. 17 F8.2 9999.99 Long. Angle of Avg Field Vr,GSE Deg. 18 F8.2 9999.99 Bx GSE, GSM nT 19 F8.2 9999.99 By GSE nT 20 F8.2 9999.99 Bz GSE nT 21 F8.2 9999.99 By GSM nT 22 F8.2 9999.99 Bz GSM nT 23 F8.2 9999.99 sigma|B| RMS Std Dev in avg magnitude, nT 24 F8.2 9999.99 sigma B SQRT[sigma(Bx)**2 + Sigma(By)**2 + Sigma(Bz)**2], nT 25 F8.2 9999.99 sigma Bx RMS Std Dev in Bx GSE, nT 26 F8.2 9999.99 sigma By RMS Std Dev in By GSE, nT 27 F8.2 9999.99 sigma Bz RMS Std Dev in Bz GSE, nT 28 I2 9 Plasma Region Flag 1,2,3,4 See above 29 I2 9 Np, Number of plasma points 30 I3 99 FCp, Coverage within minute See above 31 F5.2 9.99 DWp, Interior fraction See above 32 F7.1 9999.9 V_fit, Flow Speed km/s 33 F8.1 9999.9 Vx Velocity_fit, GSE km/s 34 F8.1 9999.9 Vy Velocity_fit, GSE km/s 35 F8.1 9999.9 Vz Velocity_fit, GSE km/s 36 F7.1 9999.9 Plasma flow long_fit angle deg, see above 37 F7.1 9999.9 Plasma flow lat._fit angle deg, see above 38 F7.1 9999.9 Proton Density_fit n/cc 39 F9.0 9999999. Proton Temperature_fit K 40 F7.1 9999.9 V_mom Flow Speed km/s 41 F8.1 9999.9 Vx Velocity_mom, GSE km/s 42 F8.1 9999.9 Vy Velocity_mom, GSE km/s 43 F8.1 9999.9 Vz Velocity_mom, GSE km/s 44 F7.1 9999.9 Plasma flow long_mom angle deg, see above 45 F7.1 9999.9 Plasma flow lat._mom angle deg, see above 46 F7.1 9999.9 Proton Density_mom n/cc 47 F9.0 9999999. Proton Temperature_mom K ---------------------------------------------------------- Related data and directories: SPDF Data and Orbits Services -------------- SPDF Contact: Natalia Papitashvili , Please acknowledge NASA's Space Physics Data Facility for data usage. Authorizing NASA Official: Dr. R.E. McGuire, Head, SPDF, NASA Goddard Space Flight Center 301-286-7794, e-mail: Robert.E.McGuire@nasa.gov -----------------------------------------------------------------------