A preliminary list of Radar headers for comments and suggestions
Data Acquisition Modes
There are several data
acquisition modes, some of which may be hidden. For example,
for short Round Trip Times, it is may be better to put the T/R
times in the per-record rather than per-file header.
File Headers for Arecibo Radar Data
I include the headers from various old (NURAD-era) files, and
anything else that seems reasonable. There were various
hardware-specific pieces of information in the old system
(frequencies of specific LOs) that will likely be different
now. Capital letters generally indicate that an old NURAD or
MAIN header exists
1996 December 5: I've gone through this and tried to consider
how it fits into the AO header scheme, as described in Phil
Perrilat's .h files. These will be marked NAO (new AO) for
now. Phil has libraries to read and write some / most of
1996 December 13: Don insists that the format must be one
continuous data stream, i.e., no separate files for transmit
and receive (which was what I suggested), and with the
ephemerides interspersed. Perhaps this could be a "group"?
with records ETRRRRRR ETRRRRRR, or something like that. He
also wants multiple copies of the xmit and ephemeris
information in case of tape problems.
What to do about day boundaries,
especially between transmit and receive???
Head-of-File Header Information
- Magic Number
- Some sort of "magic number" to identify
this an AO raw data file. (NAO:
- Header ID
- Version number for header so that, in
principle, you can determine the format from a database.
This may not be necessary if it's keyword-value. NAO: Each
piece of the header (main, RI, experiment) has a version
number, which means you have to be careful to start, but
maintainance should be fairly easy. (
- Tape ID number (The AO# if possible, like a tape label)
- File ID
- File sequence number. This is in the format
YYDDD.NNNNN, where YY is the last two digits of the year,
DDD is the DOY, and NNNNN increments from one each day.
There was a motion at a RA meeting to make the first 5
digits be MJD, but it was quashed.
- observer ID
- Operator ID
- The observer and operator ID could just
be an 80 character field for "who was there" The "cognizant
observer" and/or "cognizant operator" should be here.
- program ID/Experiment Number
- Time allocation ID (may be
useful to document observation or lack thereof for various
purposes) NAO: Is this
- Configuration File Date
- Date when setup file was
generated, or possibly submitted to the system (or both,
better yet). NAO: Does this go in the sec.inp? Or where?
- run number
- data file sequence number
- UT date of experiment start NAO:
- Julian date of experiment start
- Is Radar
- This is an AO radar experiment (AO transmitter
is used), and uses the XYZ transmitter. NAO:
- UT Time of receive start for first pulse
(used to be AST?) NAO: in decoder header
- Exposure time
- Length of time to exposure end
- Transmit start
- UT Time of transmit start for first pulse
- Transmit time
- length of transmit time
- Transmitter Location
- Which transmitter
- Transmit Ephemeris
- Transmit coords and Doppler
- Transmit Dropouts
- Times and reasons for transmit failure
- Receiver Bandwidth NAO:
- Base Transmit Frequency, either actual or (center or
first?) for frequency hopping.
- Base Receive Frequency
- Number of channels (what does this mean in the
- Number of polarizations. NAO:
- Circular or linear polarization?
- Acquisition time for one cycle. This is probably a
code length (for ranging), or a convenient interval for FT
for CW. In the old system, frequency-hopping arranged to have
an integral number of these, plus some slop on each side, in
- Samples * Channels / IPP
- Code Length
- Code length in bauds. 0 for CW, -N for
(2^N)-1 long code NAO:
- Code ID
- Some sort of identification for the code (0 for CW). NAO:
- Code start position
- Position in code when transmitter
started. -1 for "unknown"?? NAO:
- Baud Length
- Baud length in ns. NAO:
- Sample Gate Time
- Sample time in ns (probably 1 or 2
times Baud Length) Old system used both this and NSPB
(Number of samples per baud)
- Gate Delay
- Time between "tick" and sample start.
- Sampler samples per baud
- samples per baud, probably 1
or 2. NAO:
- Output samples per baud
- Samples ber baud output from
device: the decoder can combine 2 samples per baud except at 50ns. NAO:
- Number of different frequencies for frequency
- Delta-frequency for hopping;
- Time for each frequency
- Restart flag (so the system knows to resume an
- Ephemeris Delay
- Time interval between start of data
taking and expected arrival of first echo so that the turnon
- IDL In decnew. What is it?
- "Time from marker to data start" Time from tick
to code-start (measured?)
- The complete ephemeris, at 1 minute
intervals. Was that ever too long? Save XDoppler,
RDoppler, RTT, RA2000, Dec2000, Az, El
- Ephemeris Setup
- How to regenerate the ephemeris. If
it's a two-body ephemeris for the short term, this can just
be the elements. In that case, the elements ought to have a
version, like MPC number, or something like that.
- Interpolation Methods
- The ephemeris is
sub-interpolated. The current method is the 4th order
polynomial generated using 5 minutes of ephemeris and
Pascal's triangle (for Doppler), and linear interpolation (I
assume) for RTT and AzEl.
- Various things for cals??
- Transmitter power at various times
- Transmitter configuration
- What transmitter mode?
- Complex/Real Data
- Direct-sampled data records are
written as complex / real numbers
- FT Data
- FT's of direct-sampled records are written
- Frequency direction
- Direction of increasing frequency,
or, whether a high-side (inverting) or low-side
(non-inverting) LO is used.
- Decoded Data
- Decoded data are written
- Decoded Samples Retained
- Which delay bins are saved
from decoded data?
- Sampled Bits/sample
- Bits per sample
- Samples per written 32 bit word??
- How samples are interleaved: IQIQ or IIQQ or
I*recl Q*recl; LRLR or L*recl R*recl; which samples are in
which positions? Should be sample-wise, as word packing is separate
Per-Record Header Information
I assume each transmit cycle gets at least one record,
thought I suppose it's conceivable that there would be
occasions not to, particularly since the fastest likely data
rates for asteroids are the ones with the shortest RTTs.
- Record Number
- Record Number in file
- Record Length
- Record length
- Frequency Number
- Which transmitter frequency for
frequency-hopping. May also be relevant for multiple
- Transmit start
- Time of transmit start for this pulse
- Transmit time
- length of transmit time for this pulse
(Could be Transmit Stop instead)
- Recv start
- Time of receive start for this pulse
- receive time
- length of receive for this pulse (could be
receive stop instead)
- RTT for this observation
- Avg Doppler for this observation? Or is it
better to just go with the ephemeris?
- Attenuators, gains, levels
- These need to be recorded.
For comparison, look at the Green
Bank Header (a FITS BINTABLE) and the VLBA
Unfortunately, Also relevant is the Definition
of FITS. Yes, in 1996, there are still people using two
digits for the year, and they is us.
Well, in September 1997, a four-digit year format is close to
being adopted: Use
represent 2097 October 31.
will always mean 1997 October 31.
Return to NAIC Home
Last update: 1997 September 15