|
|
The Bordeaux VLBI Image Database content description
- Why the BVID ?
 Top
The VLBI (Very Long Baseline Interferometry) group at the Laboratoire d'Astrophysique de Bordeaux (LAB) collaborates to
the International VLBI Service for Geodesy and Astrometry (IVS). In this framework, one of its contribution consists
in producing VLBI images of extragalactic radio sources, and especially sources that are part of the International Celestial Reference Frame (ICRF).
The ICRF was the realization of the International Celestial Reference System (ICRS) at radio wavelengths,
adopted as the fundamental celestial reference frame by the International Astronomical Union (IAU) in 1997.
The ICRF consisted in the VLBI position of 717 extragalactic radio sources: 212 defining sources, that set the direction of the frame axes,
294 candidate, 102 other, and 109 new sources. Since 1 January 2010 the fundamental celestial reference frame recognized by
the IAU has been the ICRF2, which currently includes VLBI coordinates of 3414 sources, comprising 295 defining sources (Figure 1).
Figure 1: Sky map of the 295 defining sources of the ICRF2 (Credit: D. Boboltz / USNO).
The accuracy, for the majority of the ICRF2 source positions, is at the submilliarcsecond level.
This accuracy depends essentially on the positional stability and the compactness of the source: point-like sources
with no positional variation in time are ideal sources for constructing a reference frame. Unfortunately extragalactic sources (and so ICRF2 sources)
typically exhibit extended structures on VLBI scales, and in addition, structures that can fluctuate over time.
Images provided by the Bordeaux VLBI Image Database (BVID) are essential to monitor the structural evolution and positional stability of the reference frame sources, and finally
for maintaining and improving this frame. Additionally to these astrometric applications, the BVID images are also useful for astrophysical studies, e.g. for
investigating superluminal motions in extragalactic radio sources.
The BVID is complementary to the Radio Reference Frame Image Database (RRFID)
maintained at the United States Naval Observatory (USNO).
- The VLBI technique Top
- The VLBI observing technique Top
The radio interferometry techique consists in the combination (correlation) of signals received
by two radiotelescopes (antennas or stations) observing the same radio source. The resolution depends on
the wavelength λ and the distance B between the two antennas (the baseline length), as ~ 1.22 λ/B.
Some interferometers are connected: their antennas are physically linked by cables, optical fibers
or radio link ; the distance between antennas is limited to several kilometers ; the correlation is done in real-time
in a nearby correlator.
The Very Long Baseline Interferometry (VLBI) network is composed by independant antennas.
The longest possible VLBI baseline corresponds to the diameter of the Earth (~12000 km), and the highest
resolution can reach the submilliarcsecond level (Example: for λ= 4 cm and B= 12 000 km, the resolution
is about 0.8 mas). Nowadays, the VLBI technique one of the most accurate technique to measure positions on the sky.
Figure 2: The VLBI technique. A very distant quasar emits radio waves, which are
received by two stations on Earth, forming the baseline B. The radio signal is received at the station 2
with a time delay τ, mainly due to geometrical considerations. The two signals are recorded on disks,
then shipped to the correlator, where they can be played back and correlated (Source: NASA/GSFC).
The VLBI observing technique is described as follows, and illustrated by Figure 2:
- Two stations on Earth observe the same celestial object (quasar, AGN, ... see later for more details).
- Each station registers the radio signal on disk, along with the timing information, obtained thanks to a local hydrogen maser.
- The disks are sent to a remote correlator.
- At the correlator, the two signals are played back and multiplied (correlated).
Note: Recently, instead of being recorded on disks, the signals can be directly transfered from each station
to the correlator through the Internet, and correlated in real-time: this is the e-VLBI
(electronic-VLBI).
|
VLBI stations are organised and associated in several possible networks:
(click on the map to see the location on Earth of some VLBI stations)
|
Dedicated bi-monthly observing sessions with the global VLBI Network (10 VLBA stations and up to 10 EVN stations)
are used for the purpose of imaging the extragalactic radio sources: this is the so-called RDV sessions
(Research and Development sessions with the VLBA).
- VLBI measurements Top
- For geodesy and astrometry Top
The main measurement is the time delay of the radio source signal received at two antennas.
This delay, essentially geometrical (τ in Figure 2), depends on the radio source position and
the baseline vector. The phase delay is calculated from this time delay.
The other contribution to the delay are induced by:
- the atmosphere: the troposhere (comprising a dry and a wet component) and the ionosphere (delay
calculated from the total electronic content) ; the latter is removed thanks to the combination of
simultaneous observations at X-band and S-band
- the instrumental errors: clock errors, delay in cables and electronic components, antenna effects, ...
- the relativistic effects
- the source structure impact (see the source structure correction maps section)
Nowadays, by correcting from these components and adjusting the time delay model to the observations,
we can precisely determine the radio sources position, the antennas position or the Earth's
orientation parameters (polar motion, Earth's rotation, precession-nutation).
Note:
Since the phase is only known modulo 2π, the phase delay is ambiguous. On the contrary,
the group delay (or bandwidth synthesis delay),
which is the derivative of the phase delay with respect to the pulsation (or angular frequency:
ω= 2πν, where ν is the frequency), is perfectly known.
In practice, the bandwidth synthesis delay is determined by fitting a straight line to the phases
measured at different discrete frequencies. This is done during the RDV sessions using the
frequency channels (or intermediate frequencies IFs). For example, the X-band is
divided in four IFs of 8 MHz bandwidth at 8406, 8476, 8791 and 8896 MHz (session RDV72),
which provide four phases measurements.
- For imaging Top
The output of the correlator is a table of amplitude and phase measured on each baseline (with
a certain integration time). Amplitude and phase form a quantity called complex visibility which
is the Fourier transform of the spatial brightness distribution of the source.
Thanks to the complex visibility, and using the inverse Fourier transform, we
can deduce the image of the radio source.
Unfortunately, the VLBI (and the other interferometer) provides a poorly-sampled Fourier plane, known as the
(u,v) plane: the visibility is measured at some points of the (u,v) plane
(corresponding to the observing baselines) and has a null value elsewhere.
In this case, the reconstruction (the inverse Fourier transform) leads to an incomplete
and distorted image: the dirty image. This dirty image is in reality the convolution
of the source brightness distribution with the dirty beam, which is the Fourier
transform of the sampling function (i.e. the Point Spread Function of the network). The whole mechanism is
presented in Figure 3.
Figure 3: (Top) Imaging in the Fourier plane: the visibilities
measured by the interferometer are obtained by multiplying the true visibility plane by the sampling function.
(Bottom) Imaging in the real plane: the dirty map is the convolution of the source brightness
distribution by the dirty beam. Top and bottom-panel are linked by a direct (or inverse) Fourier
transform (from Garrington 2007).
In order to interpolate the visibility value at the unsampled point of the (u,v) plane, and then
try to reconstruct the “best” (or at least the more realistic) image,
we use non-linear deconvolution methods, as for example the CLEAN method or the Maximum
Entropy Method (MEM), along with some physical hypothesis about the image (the sky is positive,
mostly empty...etc).
- About extragalactic radio sources... Top
The celestial objects observed with the VLBI technique are mainly very bright distant active galaxies.
The radiation is produced by the accretion of dust and gas by a supermassive black hole
located in the inner region of the galaxy. This region is called Active Galactic Nucleus (AGN).
There are several types of active galaxies: radiogalaxies, quasars, Seyfert galaxies, BL Lac, Blazars, etc.
All type of AGN are described on Wikipedia.
One particularly well-studied type of AGN with the VLBI are the quasars (for quasi-stellar radio source), so called because they look
point-like at optical wavelengths (more details on Wikipedia).
- The BVID products Top
- VLBI images Top
The VLBI images reveal the radio structure of the ICRF2 sources
on milliarcsecond scales (typical VLBI scales). Such structures vary in both time and
frequency, and set limits on the accuracy of source positions determined with VLBI
astrometry unless their effects are accounted for.
Images available in the BVID come from VLBI observations at 2 GHz (λ= 13 cm)
[S-band], 8 GHz (λ=4 cm) [X-band], and for some sources, from VLBA
observations at 24 GHz [K-band] and 43 GHz [Q-band]. |
 |
The data issued from the correlator are calibrated using the Astronomical Image Processing System
(AIPS) software. Here is the simplified reduction path for VLBI data:
- Bad data are removed thanks to flag information recorded during the experiment
at each station (antenna not yet on source, position error, channel error, ...).
- The initial amplitude calibration is applied using the measured system
temperatures and the gain curve provided for each station.
- After the correlation, some residuals remain in the delay (and time derivatives of delay) due to
the atmosphere, clocks offsets, errors in the stations position, etc.
These residuals are removed by the fringe-fitting process, which consists
in the research of the fringe amplitude maximum by scanning the delay and delay rate parameters.
The fringe is considered as detected if the maximum amplitude is superior to a specified noise level
(defined by a SNR value).
- A refined amplitude calibration is then applied, based on the model fitting of calibration
sources (i.e. strong and point-like sources) observed in the same experiment. In a first time,
a source structure model is fitted on the measured visibilities. From this adjustment, amplitude gain
correcting factors are determined for each antenna (and each IF). In a second time, the same factors
are applied to all the sources of the same experiment.
- Data are exported as FITS files.
These data files are then processed to image the source by using the Caltech VLBI imaging software
DIFMAP in automatic mode (Shepherd et al. 1995). The procedure is based on the
hybrid-mapping technique (Readhead & Wilkinson 1978) with a point-source model at start.
Visibilities are self-calibrated (firstly just for phases, then for phases and amplitudes),
Fourier inverted, and CLEANed. Uniform weighting and, after several iterations, natural weighting,
are successively applied to derive the final image.
A detailed tutorial about VLBI data reduction is available at
Asian Radio Astronomy Winter School 2007 webpage.
- Structure correction maps and structure indices Top
|
The structure correction maps show the magnitude of intrinsic source structure
effects on the basic quantity measured in astrometric VLBI, i.e. the geodetic VLBI bandwidth
synthesis delay measurements, as a function of interferometer resolution.
Source structure corrections are deduced from the spatial brightness distribution of the source, i.e.
the structure visible in the VLBI image. This is done for a range of length and orientation
of VLBI baselines projected onto the sky, expressed in units of the wavelength (u, v coordinates).
|
The source structure correction or structural delay is plotted in u, v coordinates (expressed in million
of wavelength). The structural effect ranges from 0 to 100 picoseconds (ps). All corrections larger than
100 ps are plotted in red. The black circle drawn has a radius equal to the Earth diameter, which represents
the longest baselines that can theoretically be observed with VLBI stations on Earth.
Some statistics (mean, max, rms, median) are available for all baselines included within the
black circle.
The structure index is deduced from the median structure effect in the VLBI bandwidth
synthesis delay (information of the source structure correction maps) as presented in Table 1.
This index ranges from 1 to 4, and increasing values indicate increasing average structural VLBI delay
effects, approximatively on a logarithmic scale.
This means that the structure index can be used as an indicator of the astrometric quality of a source,
as shown in Figure 4.
All details about the calculation of structure corrections and structure indices are described in:
-
Fey, A. & Charlot, P. 2000, ApJS, 128, 17
| Structure index |
Astrometric quality |
Median effect (ps) |
Observed structure |
| 1 |
very good |
0-3 |
point-like |
| 2 |
good |
3-10 |
resolved |
| 3 |
use with caution |
10-30 |
extended |
| 4 |
not acceptable |
>30 |
very extended |
| Table 1: Link between the structure index value, the astrometric quality, the median effect in the VLBI bandwith
synthesis delay effect and the observed source structure. |
Figure 4: (Top) VLBI contour plots at X-band for sources 0544+273, 0138-097, 2201+315 and 0108+388
representative of each structure index class from 1 (left) to 4 (right).
(Bottom) Structure correction maps showing the magnitude of structural delays induced by the
source structure at X-band for the same sources. The delay (in picoseconds) is plotted as a function of the
u, v coordinates. The circle drawn represents the maximum baseline length that can theoretically
be observed with stations on Earth (Charlot 2008).
- Visibility maps and source compactness Top
The visibility maps show values of the visibility amplitude (normalized to the maximum)
as a function of interferometer resolution.
The normalized visibility is plotted in u, v coordinates (expressed in million of wavelength).
It ranges from 0 to 1 (no units). Once again, the black circle drawn has a radius equal to the
Earth diameter, which represents the longest baselines that can theoretically be observed with
the Earth-based VLBI. Statistics (mean, max, rms, median) are available for all baselines
included within the
black circle.
|
|
Visibility maps may be used to estimate source compactness (the median value of the visibility
within the black circle) and the correlated flux density for any given VLBI baselines.
They are also useful e.g. for scheduling purposes.
The BVID usage Top
- Queries Top
The BVID can be queried by three complementary ways: by source name, by date or experiment name, and by
coordinates.
- By source name Top
In the BVID, three different source designations can be queried:
- the IERS name: designation provided by the International Earth Rotation and Reference System Service,
constructed from B1950 coordinates like HHMM+DDd or HHMM-DDd. This name is the first
identifier in the BVID.
Examples: 0202+149, 1741-038, 0430+052
One can provide the full name (0202+149) or just a part of the name (0202, 02, +149,...). If there is a single
result, the page of the source is directly and automatically loaded. If there are multiple results, a list of
the matching source names, sorted by increasing right ascension (i.e. the first 4 digits of the name), is displayed.
- the ICRS name: designation of ICRF2 sources, constructed from J2000.0
coordinates like JHHMMSS.s+DDMMSS or JHHMMSS.s-DDMMSS.
Examples: J020450.4+151411, J174358.8-035004, J043311.0+052115
Once again, one can provide the full name (J020450.4+151411) or just any part of it, and if the query result is not
unique, a matching source name list is provided.
- An other name: any other designation of the source. This name is resolved by
Simbad.
Examples: 4C 15.05 or DA 67, NVSS J174358-035004, 3C 120 or 4C 05.20 or DA 140 or Mrk 1506
A complete name should be provided ; this name has to follow the rules of the Simbad nomination
(see Simbad dictionnary of nomenclature)
in order to be resolved. An appropriate error message will be displayed if the source name provided is not found by Simbad,
or if the source found by Simbad is not the BVID.
In addition, on the right-hand side of the page is displayed a list of the most queried sources.
Finally, two query shortcuts ('Direct Links') are available:
- Query all sources gives a list of all sources (with data) available in the BVID.
- Query all defining ICRF2 sources gives a list of all defining sources from the ICRF2 catalog. Sources with no data
are displayed in greyish style.
- By date or experiment code Top
An easy way to navigate through the available dates and/or experiments names is to use the chronological axis.
Different colors represent different frequency bands, which can be show or hide using the legend links. An experiment
can be selected by clicking on its color bar or on its name.
More specific boxes allow to search a specific date or experiment name.
- By date (year)
Enter the year (4 digits) in the appropriate field (Example: 1997), then click on the OK button.
A new pull-down menu appears (if there is at least one experiment in the chosen year), in which
one can pick the wanted date (the experiment name is then given as an indication) ; then click
on the Search button.
If no experiment has been observed during the specified year, just Clear the form of re-enter a new date.
- By experiment code
Enter the full experiment name in the field, then click on the OK button. You will automatically be
transfered to the list of sources included in this experiment.
Examples: RDV45, BL115B
- By choosing in an experiment code list
Choose the corresponding frequency band you want in the first pull-down menu (SX or KQ). Then, pick in the second
pull-down menu the full name of the experiment. Experiment names are sorted by alphabetical order. Once the experiment
name is chosen, click on the OK button to automatically load the sources list of this experiment.
Examples: SX then RDV45, KQ then BL115B
- By coordinates Top
J2000 coordinates (Right Ascension and Declination) should be supplied in the appropriate fields
and in sexagesimal format (not decimal !) with a separator that could be ' ' (space), ':' (colon),
or 'hms' for the Right Ascension and 'dms' for Declination (units).
Examples: 12 24 29.1, 07:38:10.3, 00h44m30.9s
Incomplete coordinates can also be provided: the missing part(s) will be filled with zeros.
Examples: 12 24, 00h or 12h24, 00
Resulting sources along with the distance from the provided coordinates are presented in a list sorted
by increasing distance, where the maximum distance is given by the radius.
Examples: 5 deg, 10 arc min
Note: There is no default value for the radius. However, a “good” starting
radius could be 5 degrees, since the distance between the observed sources and their nearest neighbor
are typically ranging up to several degrees.
- Query results Top
The result page is divided into three panels: the “source details” panel which
mostly presents generalities about the source, the “thumbnails” panel
which allows a direct and simple visualization of the most recent images of the source, and finally
the “experiments” panel which regroups all the relevant information (structure indices,
compactness, links to images,...) for each observing session.
- Source details Top
Three kind of information are proposed in this panel:
- Source information: the BVID main name (IERS name), some important aliases, its ICRF2 category
(if the source is part of the ICRF2), its position (the ICRF2 position for ICRF2 sources) and direct links to
Simbad and
NED pages.
The Query around button gives the possibility to find all sources in the neighborhood of the currently displayed
source (i.e. all sources within a circle with the specified radius, set to 5 degrees by default).
- Data of the most recent session that includes the source: the experiment date and name
is specified, along with the corresponding total VLBI flux, the structure indices and
the compactness that we determined.
- The choice of the observing frequency band. When the source has been
observed at several frequency bands (SX and KQ), a pull-down menu appears at the right-side
of the “Band” menu item (Example).
If the KQ frequency band is picked, the page will reload with the KQ data
(Example).
- Thumbnails Top
The thumbnails panel allows a direct and simple visualization of the most recent images
for the frequency band selected (or displayed) in the source details panel. This panel is composed of two
sub-panels: on top, the X (or Q) band images ; on bottom, the S (or K) band images.
Each sub-panel presents a thumbnail for the VLBI image (if the most recent session was processed at
the Laboratoire d'Astrophysique de Bordeaux) or a direct link to the USNO RRFID corresponding image
(in this case, the link is materialized by the USNO logo icon ;
Example).
Structure correction maps and visibility maps thumbnails are also available for the two frequencies.
- Experiments Top
The experiments table shows all experiments which comprise the source for the frequency band
selected (or displayed) in the source details panel. This table summerizes all the relevant and useful
information about the source (structure indices, compactness, links to the images, ...). The experiments
are sorted by decreasing year.
This table can be presented in two different ways:
- as a year-organized table (by default). Each year can be deployed by clicking on the
year, the number of experiments or the blue triangle (Example)
- as a simple list. In this case, the data is directly accessible
(Example)
The desired presentation can be chosen thanks to the red table top-link (Switch to the ...).
Two other switches are available at the bottom of the experiments table:
- Show all years or show only the last 3 years (by default). Some sources have been
observed many times. By default, the number of displayed experiments is limited to the three
years preceding the last experiment. This link allows to show all years or to hide experiments older
than three years (from the last session)
(Example).
- Explode experiments (only when the table is year-organized). Instead of deploying one
year at a time, this link allows to deploy all the visible years
(Example).
A slideshow functionality, accessible through the screen icon  , provides
an convenient way to quickly visualise all VLBI images and experiment data available for the displayed source (Example).
|
|
