From: NCF::DWILLIAMS "Dave Williams, NSSDC - (301) 441-4197" 24-OCT-1994 16:15:21.53 To: HILLS CC: Subj: PVO DSC -02C B44308-000A 78-051A-02C Pioneer Venus Orbiter Radar Mapper Peter G Ford Massachusetts Institute of Technology Composite Data Files To avoid the necessity of extracting all LFD data variables when plot- ting ORAD data, a single composite data file has been submitted to UADS. The file is designed to be handled as a single entity. It can be read directly into core memory on many large virtual storage computer systems (with 5 megabytes or more of addressable memory) and contains chain pointers to aid in on-line graphical representation of all or part of the total ORAD database. The file consists of 80 byte logical records and is arranged in eight sections; the first five sections are of fixed length and each fits exactly into an integral number of logical records. The remaining three sections are of varying length. Each begins at the start of a logical record, and the last record of each section is padded on the right with hexadecimal zeroes. The sections have the following contents: Periapsis 1000 double precision floating point fields containing the periapsis times of orbits 1-1000 expressed in units of milliseconds counting from midnight Dec 30/31st of the previous year. If no ORAD data is available for the orbit, the field will contain a zero value. The first of these fields does not contain a time value. No ORAD data was taken on orbit 1. Instead, the first 4 bytes of the field (i.e. of the file) contain the number of altimeter data values con- tained in the varying length sections of the file, expressed as a 32 bit binary integer. Semi-Major 1000 double-precision floating point fields containing the semi- major axes of the osculating orbital ellipse at periapsis for orbits 1-1000, expressed in kilometers. Eccentricities 1000 double-precision floating point fields containing the eccen- tricity of the osculating orbital ellipse at periapsis for orbits 1-1000. Data-Source 1000 16-bit binary integers indicating the source of ORAD data for orbits 1-1000. The codes are as follows: 0: no data available 1: quick-look data 2: first processing of EDR/SEDR tapes 3: re-processing of EDR/SEDR tapes Data-Editing 1000 16-bit binary integers indicating the number of times that ORAD data for orbits 1-1000 has been processed by the interactive CYTHP editing system. Altimetry seven arrays of 32-bit fields. The dimension of each array is given by the integer field located in the first 4 bytes of the first record of the file. There is no padding between the arrays. The first six arrays contain floating point variables: 1: Crust-fixed latitudes (RLAT) 2: Crust-fixed longitudes (RLON) 3: Planetary radius relative to 6051.2 km (RRAD) 4: Hagfors scattering law parameter (C) 5: Fresnel reflection coefficient (RRHO) 6: Spacecraft radial velocity (km/sec) and the seventh array contains a pair of binary integers in each 32 bit field: Bits 0-19: Orbit number Bits 20-31: Roll number relative to periapsis + 128 Latitude an array of 32 bit integers that index the data arrays in the sixth section of the file, used to reference the data in that section by increasing value of RLAT. Thus if, in a Fortran program, the data arrays are read into variables RLAT(N), RLON(N), etc, and the lati- tude pointers into ILAT(N), the sequence RLAT(ILAT(I)) will inrcease monotonically with increasing I value. Longitude an array of 32 bit integers that index the data arrays in the sixth section of the file, used to reference the data in that section by increasing value of RLON. Thus if, in a Fortran program, the data arrays are read into variables RLAT(N), RLON(N), etc, and the long- itude pointers into ILON(N), the sequence RLON(ILON(I)) will inrcease monotonically with increasing I value. 6. Error Analysis An analysis of purely statistical errors shows that values of SLOP and RRHO obtained from the profile fitting procedure are usually highly correlated, while RRAD is only slightly correlated with either. At an altitude of 200 km, the statistical error in RRAD is of the order of 30 meters, while the errors in SLOP and RRHO approach 10%. At 4000 km altitude, the error in RRAD varies between 100 and 300 meters (it varies with SLOP), while errors in SLOP and RRHO are on the order of 30%. These formal errors are very much smaller than the following known sys- tematic errors: 6.1. Estimation of on-board clock time pulses This chiefly affects RRAD according to ERR(RRAD) = RADVEL * ERR(RIP) where ERR(RIP) is an inaccuracy in the measurement of the Roll Index Pulse time, and RADVEL is the current spacecraft radial velo- city with respect to Venus. RRAD has been corrected for gross RIP timing errors but a residual error of several milliseconds remains leading to errors in RRAD of the order of 100 meters. Gross errors in RIP time also affect RRHO because they lead to a miscalculation of the direction in which the radar antenna is pointing. Although only weakly directional (3 db loss at 15 degrees off axis), a large error in RIP determination affects the computation of this loss and therefore affects the analysis of the total echo power. 6.2. Ephemeris errors These have been estimated by comparing RRAD values of overlaping footprints measured during separate orbits. Of greatest concern is the uncertainty in time of periapsis passage as this affects all RRAD values of a given orbit in the same manner as RIP timing. This error decreases with altitude, vanishing at periapsis. Errors in other orbital elements have less effect on RRAD. We estimate the residual effect of all ephemeris errors on RRAD as 50 meters at periapsis, increasing to 150 meters at 4700 km spacecraft altitude. 6.3. Anomalous surface properties When the composition of the surface footprint differs sufficiently from the 'model' surface used to compute the theoretical radar tem- plates, we expect erroneous results. The three most probable causes of error in our model would come from footprints that con- tain either A) a large smooth area tilted in the direction of the spacecraft, or B) large changes in relief, or C) very rough material on the meter scale, or some combination of the three. Such a failure of the scattering model can lead to very large errors in RRAD, SLOP, and RRHO. Obvious failures are deleted automatically by the fitting algorithms, and most of the remaining suspect values have been analyzed by hand.