[evla-sw-discuss] VLA archive records after the Modcomps

Ken Sowinski ksowinsk at aoc.nrao.edu
Wed Aug 3 16:11:27 EDT 2005


                       VLA Computing Memo #xxx
         Changes to the VLA Archive Record for the Interim VLA
                          August 3, 2005
                           K. Sowinski

In the era after the Modcomps and prior to WIDAR the VLA will continue
to produce data in a format very close to the traditional VLA archive
record as described in VLA Computing Memo #188.  This memo sets out
differences in format and semantics of the interim EVLA archive record
from the traditional VLA archive record.  Data from the WIDAR
correlator will be represented by some expression of the Science Data
Model.  The goal is to leave the existing archive record format and
contents as untouched as possible to relieve the burden of change upon
existing programs which read these records.  Some information will no 
longer be provided, but space in the record will remain and be filled 
with zeros or blanks as appropriate.  Other items may change slightly 
in meaning; these will be documented.  I don't expect to add anything 
new to the archive record.

Each archive record is a self contained logical record that fully 
describes and contains the result of a single VLA integration.  Each
such record is independent of all others.  Sequential records with 
the same start time comprise a 'scan'.  If multiple subarrays are in
use archive records for each subarray are interleaved.  Even a subarry
of a single antenna will produce an archive record, but it will have null
pointers to the visibility data.  The record is divided into four parts:
a Record Control Area (RCA) which contains bookkeeping information, time
and offsets to the other sections, a Subarray Data Area (SDA) which 
contains information describing the antenna independent parts of the 
record, Antenna Data Areas (ADA) for each antenna in the subarray which 
provide the antenna specific information, and depending upon correlator
mode up to four Correlator Data Areas (CDA) contain the visibility data.  
The results in the CDA can be interpreted only if the number of antennas 
in the subarray and the order of the ADAs is known.  A description of
the formatting of the visibility data, as well as all the other details
ignored here are available in VLA Computing Memo #188.

The archive record  will still be constructed in network (big-endian)
order so the usual byte shuffling will still be necessary on Intel
machines.  I am tempted to continue writing floating point numbers
in Modcomp rather than IEEE format so that programs I have written to
digest the records will not have to be changed.  The argument against
this is 1) the need to write IEEE to Modcomp conversion functions, and
2) the fact that it is about time isn't it?  I am open to arguments, or
merely pleading, in either direction.  

All the times that are described as IAT will likely be UTC if we 
continue to run the system on UTC as we have so far.  If UTC is
the time of the future we should consider modifying the existing system 
to run on UTC to ease the transition and to make comparisons easier
when testing the new system.  If I understand it rightly it requires 
only changing a value in a system file and a modifiable parameter in the
"IAT" clock.  Alternatively all the UTCs can be converted to IAT so
that any analysis programs that depend on time being expressed in IAT
will still do the right thing.

The visibility data is expected to be processed as we are used to with
the current system.  For VLA antennas the both the T_sys and round trip
phase corrections will be applied; for EVLA antennas we may apply a
backend T_sys correction and will apply a phase correction based on
the path length measurement in the L352 when we figure out how it
works and learn whether the changes are fast and large enough to warrant 
a correction.

The source of the data presented in the header records described below 
can be divided into four classes:
0.  Known to or calculable by the archive record creator when the record 
    is constructed,
1.  The executor, or other components of the system, know it and can
    present it to DCAF as needed,
2.  The information will be encoded in the script and passed, untouched
    to DCAF,
3.  The information is not explicitly present so will have to be 
    calculated from known quantities or teased out of CALC.

It is straightforward to determine the source of items in the first
three classes and class 2 will be discussed in more detail in a
separate document.  The items in class 3 deserve more explanation.  
I will provide that here and leave the complete list for the end of 
this memo.


SD.PDA          32     3       APPARANT RA AND DEC AT OBSERVATION
My understanding is that in the approach taken by CALC this is
irrelevant.  We will have to document the what this item means
for planetary observations, for both the old and the new systems.
I expect that this will be reported as zero or as the position
of epoch.

SD.SKY          56     3       SKY FREQUENCY AT BAND CENTRE  IN GHZ
The executor knows and will provide the signed sum of the LOs and 
the backend bandwidth.  If the net sideband is known this is easily
calculated.

SD.SRF          84     3       CURRENT SURFACE REFRACTIVITY  (N-1)
SD.ZAP          86     3       ESTIMATED ZENITH ATMOSPHERIC PHASE PATH  (NSEC
The are estimated by CALC and included in the pointing angle and
total delay.  So far as I know these have never been used except 
to be compared to independent calculations as a check.  If we can
get them out of CALC they will be provided, else they will be 
reported as zero.

SD.TRI          88     3       SIN AND COS OF EL, AZ AND ETA
These are the array-center angles.  The executor knows them only for 
individual antennas.  Eric says that sin(El) is used in AIPS; the
only other use I know of is to attach the Az and El at the time of
a pointing offset determination to the pointing measurement for PEEK.
To be consistent with current usage it must be the geometric Az and El
with no pointing model or refraction corrections included.  It is
probably sufficient to report the pointing angles for the innermost
of the antennas in the subarray.  Parallactic angle will not be
provided.

AD.U            28     3       U
AD.V            30     3       V
AD.W            32     3       AND W AT CENTER OF INTEGRATION FOR SPECIFIED E
These need to be calculated.  Eric says that they are needed in AIPS.
If CALC will not provide this, then we will need the source position
at epoch, an accurate LST for the center of the integration interval
and antenna positions.  In any case, this must be U and V suitable
to make a map in J2000 coordinates.

AD.BL           34             Apparant BX, By and BZ (nsec) 
In the current implementation we report the nominal antenna locations
as given in the BASELINE file corrected for local differential earth
tides and the IERS estimate of the position of the pole.  The
calculation of apparant source position includes all the usual effects
(precession, nutation, aberrations, light bending) but not these two.
U and V are then calculated using the corrected antenna positions and
source positions and accounts for all the relevant effects.  The
approach taken by CALC is quite different.  I don't understand it well
enough to describe it with any certainty, but I expect that we will
only be able to record nominal antenna positions and source position
of epoch expressed in the equator and equinox of J2000.  The
astrometry community will have to tell us how important this is and
what ought to be recorded.  The simplest approach is to record the time
of observation, the nominal antenna positions, the source position of
epoch and say that CALC version x.y was used to calculate the delay.



*      FIRST THE RECORD CONTROL AREA  (RCA)
Name            Offset Class   Description
====            ====== =====   ===========
RA.LRL          0      0       LOGICAL RECORD LENGTH IN WORDS
RA.FMT          2      0       FORMAT TYPE  (THIS IS FORMAT 1)
RA.REV          3      0       REVISION LEVEL    (-1 WILL DO FOR NOW)
RA.MJD          4      1       DATE  (MJAD)
RA.INT          6      1       IAT TIME IN INTERRUPT COUNTS SINCE MIDNIGHT
RA.CP           8      0       CONTROL PROGRAM ID
RA.SDP          12     0       POINTER TO SUBARRAY DATA AREA  (SDA)
RA.ADP          14     0       POINTER TO ANTENNA DATA AREA  (ADA)
RA.ADL          16     0       LENGTH OF ADA IN WORDS
RA.NAN          17     1       NUMBER OF ANTENNAS IN THIS SUBARRAY
RA.C1P          18     0       POINTER TO FIRST CORRELATOR DATA AREA  (CDA)
RA.C1L          20     1       NUMBER OF CORRELATORS  (CH*BL)
RA.C2P          22     0       POINTER TO SECOND CDA
RA.C2L          24     1       NUMBER OF CORRELATORS
RA.C3P          26     0       POINTER TO THIRD CDA
RA.C3L          28     1       NUMBER OF CORRELATORS
RA.C4P          30     0       POINTER TO FOURTH CDA
RA.C4L          32     1       NUMBER OF CORRELATORS
RA.RAT          34     0       RATIO OF BLOCK SIZE TO 'ANSI' RECORD SIZE
RA.SUB          35     1       ACTIVE SUBARRAY MASK (BIT15 == SUB1, ETC)
RA.LEN          36             SIZE OF RCA

*      THE SD.XXX ARE IN THE SUBARRAY DATA AREA
SD.ID           0      1       SUBARRAY ID
SD.SOU          1      1       SOURCE NAME (ASCII)
SD.QUA          9      2       SOURCE QUALIFIER  (INTEGER)
SD.CFG          10     2       ARRAY CONFIGURATION  (ASCII)
SD.PID          11     2       OBSERVING PROGRAM ID  (ASCII)
SD.OID          14     2       OBSERVER'S DEC-10 NUMBER  (INTEGER)
SD.MOD          15     2       OBSERVING MODE AND SUBMODE  (ASCII, MORE OR LE
SD.ASB          17             ARRAY STATUS BYTE
SD.NCH          18     1       NUMBER OF COMPLEX CORRELATORS/BASELINE FOR EAC
SD.AVG          19     1       INTEGRATION TIME  (WG CYCLES)
SD.STO          20     1       LST STOP TIME  (TURNS OR RADIAMS?)
SD.STA          22     1       LST START TIME          "
SD.POS          24     1       RA AND DEC AT EPOCH
SD.PDA          32     3       APPARANT RA AND DEC AT OBSERVATION
SD.SLO          40     1       SIGNED SUM OF LOS (A-D)   IN GHZ
SD.SKY          56     3       SKY FREQUENCY AT BAND CENTRE  IN GHZ
SD.IAT          72     1       IAT AT END OF INTEGRATION INTERVAL  (RADIANS)
SD.LST          76     1       LST              "                      "
SD.GTI          80     1       IAT FOR GEOMETRY CALCULATIONS  (RADIANS)
SD.SRF          84     3       CURRENT SURFACE REFRACTIVITY  (N-1)
SD.ZAP          86     3       ESTIMATED ZENITH ATMOSPHERIC PHASE PATH  (NSEC
SD.TRI          88     3       SIN AND COS OF EL, AZ AND ETA
SD.BW           100    1       BACKEND BANDWIDTH
SD.FEF          101    1       FRONT END FILTER  (VLA antennas only)
SD.REC          102    1       RECIRCULATOR CONTROL CODE
SD.ZSP          103    2       ZERO SPACING FLUX
SD.UVL          105            UV LIMITS SUPPLIED FOR ANTSOL
SD.ACB          109    0,1     ARRAY CONTROL BITS
SD.WEA          111    1       FIVE NUMBERS OF WEATHER INFO
SD.VEL          121    1       VELOCITIES FOR IF A-D  (KM/SEC)
SD.LRF          137    1       LINE REST FREQUENCIES FOR IF A-D  (MHZ)
SD.VSC          153    1       VELOCITY SYSTEM CODE FOR IF A-D  (ASCII)
SD.CRM          157    1       CORRELATOR MODE  (ASCII)
SD.APO          159    1       AP OPTIONS FOR IF A-D  (ASCII)
SD.EPY          161    0,1     EPOCH YEAR  (INTEGER)
SD.SEL          162    1       DATA SELECTION PARAMETERS FOR FOUR IFS
SD.CHS          166    0       CHANNEL SEPARATION CODES FOR 4 SPECTRA
SD.LEN          170

*      AD.XXX ARE ALL IN THE ANTENNA DATA AREA
AD.ID           0      1       ANTENNA ID AND DCS ADDRESS  (ONE PER BYTE)
AD.CTL          1      0,1     ANTENNA CONTROL BITS
AD.IFS          3      1       IF STATUS  (FLAGGING INFORMATION)
AD.NOM          4      1       NOMINAL SENSITIVITY IF A-D  (UNITLESS)
AD.PDE          12     1       PECULIAR DELAY (NSEC)
AD.PPH          20     1       PECULIAR PHASE  (TIRNS   B+0)
AD.DEL          24     1       TOTAL DELAY (NSEC)
AD.U            28     3       U
AD.V            30     3       V
AD.W            32     3       AND W AT CENTER OF INTEGRATION FOR SPECIFIED E
AD.BL           34     1       APPARANT BX, BY AND BZ   (NSEC)
AD.BA           46     1       BA  (K-TERM)
AD.TSF          48     1       FE Tsys -- VLA antennas only
AD.TSB          56     1       Be Tsys
AD.IFC          64     0,1     OR OF I.CNTL FROM FOUR IFS
AD.LEN          68



More information about the evla-sw-discuss mailing list