AT
A
T
sys
=
1
t Av
SNR =
T
A
AT
A
=
T
A
T
sys
t Av
Radiometer equation
N / 1
For Gaussian noise, total error for N samples
is
that of single
sample
Basic Radio Astronomy
Sensitivity (Noise)
Aperture Synthesis
Signals at each point in the aperture are brought together in phase
at the antenna output (the focus)
Imagine the aperture to be subdivided into N smaller elementary
areas; the voltage, V(t), at the output is the sum of the contributions
AV
i
(t) from the N individual aperture elements:
V(t) = AV
i
(t)
Aperture Synthesis
Output of a Filled Aperture
The radio power measured by a receiver attached to the telescope
is proportional to a running time average of the square of the output
voltage:
Any measurement with the large filled-aperture telescope can be
written as a sum, in which each term depends on contributions from
only two of the N aperture elements
Each term (AV
i
AV
k
) can be measured with two small antennas, if we
place them at locations i and k and measure the average product of
their output voltages with a correlation (multiplying) receiver
P AV
i
( )
2
= AV
i
AV
k
( )
= AV
i
2
+ AV
i
AV
k
i=k
Aperture Synthesis
Aperture Synthesis: Basic Concept
If the source emission is unchanging, there is no need to measure
all the pairs at one time
One could imagine sequentially combining pairs of signals. For N
sub-apertures there will be N(N-1)/2 pairs to combine
Adding together all the terms effectively synthesizes one
measurement taken with a large filled-aperture telescope
Can synthesize apertures much larger than can be constructed as a
filled aperture, giving very good spatial resolution
Aperture Synthesis
Aperture Synthesis: Basic Concept
What is the interferometer response as a function of sky position, l = sino
?
In direction s
0
(o = 0) the
wavefront arriving at
telescope #1 has an extra
path b s
0
= b sinu to
travel relative to #2
The time taken to traverse this
extra path is the geometric
delay, t
g
= b s
0
/c
This delay is compensated for
by inserting a signal path
delay for #2 equivalent to t
g
Aperture Synthesis
A Simple 2-Element Interferometer
At angle o
,
a wavefront has an extra path x = u sino = ul to travel
Expand to 2D by introducing |
orthogonal to o, m = sin|,
and v orthogonal to u, so
that in this direction the
extra path y = vm
Write all distances in units of
wavelength, x x/, u u/,
etc., so that x and y are
now numbers of cycles
Extra path is now ul + vm
V
2
= V
1
e
-2ti(ul+vm)
Aperture Synthesis
Response of a 2-Element
Interferometer
The output from the correlator (the multiplying and time-averaging
device) is:
For (l
1
=l
2
, m
1
=m
2
) the above average is zero (assuming mutual sky),
so
C = V
1
V
2
= V
1
(l,m)dldm
}}
V
2
(l,m)dldm
}}
C = V
1
(l,m)V
2
(l,m)dldm
}}
= V
1
(l,m)V
2
(l,m) dldm
}}
= V
1
(l, m)
2
e
2ti(ul +vm)
dldm
}}
= I(l, m)e
2ti(ul +vm)
dldm
}}
Aperture Synthesis
Correlator Output
Thus, the interferometer measures the complex visibility, V, of
a source, which is the FT of its intensity distribution on the
sky:
u,v are spatial frequencies in the E-W and N-S directions,
are the projected baseline lengths measured in units of
wavelength, i.e., B/
l, m are direction cosines relative to a reference position in
the E-W and N-S directions
(l = 0, m = 0) is known as the phase center
the phase | contains information about the location of
structure with spatial frequency u,v relative to the phase
center
V (u,v) = Ae
i|
= I(l, m)e
2ti(ul +vm)
dldm
}}
Aperture Synthesis
The Complex Visibility
This FT relationship is the van Cittert-Zernike theorem, upon
which synthesis imaging is based
It means there is an inverse FT relationship that enables us to
recover I(l,m) from V(u,v):
The correlator measures both real and imaginary parts of the
visibility to give the amplitude and phase:
V (u,v) = I(l, m)e
2ti(ul +vm)
dldm
}}
I(l, m) = V (u,v)e
2ti(ul +vm)
}}
dudv
A = 9
2
+
2
| = tan
1
9
|
\
|
.
|
9
A=|V |
|
Aperture Synthesis
The Complex Visibility
Fourier theory states that any signal (including images)
can be expressed as a sum of sinusoids
Jean Baptiste
Joseph Fourier
1768-1830
signal 4 sinusoids sum
Aperture Synthesis
The Fourier Transform
the Fourier Transform contains all information of the original
(x,y) plane and (u,v) plane are conjugate coordinate systems
I(x,y) V(u,v) = FT{I(x,y)}
I(x,y)
narrow features transform to wide features (and vice-versa)
Amp{V(u,v)}
Gaussian
o Function
Constant
Gaussian
Aperture Synthesis
Some 2-D Fourier Transform Pairs
elliptical
Gaussian
sharp edges result in many high spatial frequencies
elliptical
Gaussian
Disk
Bessel
Aperture Synthesis
More 2-D Fourier Transform Pairs
I(x,y)
Amp{V(u,v)}
complex numbers: (real, imaginary) or (amplitude, phase)
amplitude tells how much of a certain spatial frequency
component
phase tells where this component is located
I(x,y)
Amp{V(u,v)}
Pha{V(u,v)}
Aperture Synthesis
Amplitude and Phase
complex numbers: (real, imaginary) or (amplitude, phase)
amplitude tells how much of a certain spatial frequency
component
phase tells where this component is located
I(x,y)
Amp{V(u,v)}
Pha{V(u,v)}
Aperture Synthesis
Amplitude and Phase
The FT of a single visibility measurement is a sinusoid with spacing
1/u = /B between successive peaks, or fringes
Build up an image of the sky by summing many such sinusoids
(addition theorem)
FT scaling theorem shows:
Short baselines have large
fringe spacings and measure
large-scale structure on the sky
Long baselines have small
fringe spacings and measure
small-scale structure on the sky
+ + + Fringe Sign
/B rad.
Source
brightness
Aperture Synthesis
Picturing the Visibility: Fringes
34
sample V(u,v) at enough points to synthesize the equivalent large
aperture of size (u
max
,v
max
)
1 pair of telescopes 1 (u,v) sample at a time
N telescopes number of samples = N(N-1)/2
fill in (u,v) plane by making use of Earth rotation:
Sir Martin Ryle, 1974 Nobel Prize in Physics
reconfigure physical layout of N telescopes for more
2 configurations
of 8 SMA antennas
345 GHz
Dec = -24 deg
Sir Martin Ryle
1918-1984
Aperture Synthesis
Aperture Synthesis
Imaging
in aperture synthesis, V(u,v) samples are limited by number of
telescopes, and Earth-sky geometry
high spatial frequencies:
maximum angular resolution
low spatial frequencies:
extended structures
invisible
(aka. only a max scale can
be imaged; also ``zero-
spacing problem = no large
scales)
irregular within high/low limits:
sampling theorem violated
still more information
missing
Imaging
(u,v) Plane Sampling
sample Fourier domain at discrete points, i.e.,
the inverse Fourier transform of this is
But the convolution theorem tells us
where (the point spread function)
Fourier transform of sampled visibilities yields the true sky brightness
convolved with the point spread function
(the dirty image is the true image convolved with the dirty beam)
Formal Description
I
I
I
Imaging
B(u,v)
I
D
(x,y)
(dirty
image)
b(x,y)
(dirty
beam)
I(x,y)
Dirty Beam and Dirty Image
Imaging
2 Antennas
Dirty Beam Shape and N Antennas
Imaging
3 Antennas
Dirty Beam Shape and N Antennas
Imaging
4 Antennas
Dirty Beam Shape and N Antennas
Imaging
5 Antennas
Dirty Beam Shape and N Antennas
Imaging
6 Antennas
Dirty Beam Shape and N Antennas
Imaging
7 Antennas
Dirty Beam Shape and N Antennas
Imaging
8 Antennas
Dirty Beam Shape and N Antennas
Imaging
8 Antennas x 6 Samples
Dirty Beam Shape and N Antennas
Imaging
8 Antennas x 30 Samples
Dirty Beam Shape and N Antennas
Imaging
8 Antennas x 60 Samples
Dirty Beam Shape and N Antennas
Imaging
8 Antennas x 120 Samples
Dirty Beam Shape and N Antennas
Imaging
8 Antennas x 240 Samples
Dirty Beam Shape and N Antennas
Imaging
8 Antennas x 480 Samples
Dirty Beam Shape and N Antennas
Imaging
uv plane analysis
best for simple sources, e.g., point sources, disks
image plane analysis
Fourier transform V(u,v) samples to image plane, get I
D
(x,y)
but difficult to do science on dirty image
deconvolve b(x,y) from I
D
(x,y) to determine (model of) I(x,y)
visibilities dirty image sky brightness
How to analyze interferometer data?
Imaging
Visibility Weighting in the FT:
Including weighting function W to modify dirty beam sidelobes:
natural weighting: density of uv-coverage = highest compact
flux sensitivity
uniform weighting: extent of uv-coverage = highest resolution
robust weighting: compromise between natural and uniform
tapering: downweights high spatial frequencies = higher
extended flux sensitivity
Weighting and Tapering
Imaging
imaging parameters provide a lot of freedom
appropriate choice depends on science goals
NB: Primary Beam FWHM is the Field-of-View of a single-
pointing interferometric image
Robust 0
+ Taper
0.77x0.62
o = 1.7
Natural
0.77x0.62
o = 1.0
Uniform
0.39x0.31
o = 3.7
Robust 0
0.41x0.36
o = 1.6
Weighting and Tapering: Examples
Imaging
Deconvolution
difficult to do science on dirty image
deconvolve b(x,y) from I
D
(x,y) to recover I(x,y)
information is missing, so be careful! (theres noise, too)
dirty image CLEAN image
Deconvolution
Deconvolution
Deconvolution:
uses non-linear techniques effectively interpolate/extrapolate
samples of V(u,v) into unsampled regions of the (u,v) plane
aims to find a sensible model of I(x,y) compatible with data
requires a priori assumptions about I(x,y)
CLEAN (Hgbom 1974) is most common algorithm in radio astronomy
a priori assumption: I(x,y) is a collection of point sources
variants for computational efficiency, extended structure
deconvolution requires knowledge of beam shape and image noise
properties (usually OK for aperture synthesis)
atmospheric seeing can modify effective beam shape
deconvolution process can modify image noise properties
Deconvolution Algorithms
Deconvolution
1. Initialize
a residual map to the dirty map
a CLEAN component list
2. Identify strongest feature in
residual map as a point source
3. Add a fraction g (the loop gain) of
this point source to the clean
component list (g ~ 0.05-0.3)
4. Subtract the fraction g times b(x,y)
from residual map
5. If stopping criteria
*
not reached, go
back to step 2 (an iteration), or
6. Convolve CLEAN component (cc)
list with an estimate of the main
dirty beam lobe (i.e., the CLEAN
beam) and add residual map to
make the final restored image
b(x,y)
I
D
(x,y)
Basic CLEAN Algorithm
I(x,y)
Deconvolution
* Stopping criteria = N x rms (if noise limited), or I
max
/N (if dynamic
range limited), where N is some arbitrarily chosen value
restored
image
residual
map
CLEAN
model
I
D
(x,y)
CLEAN
Deconvolution
CLEAN beam size:
natural choice is to fit the central peak of the dirty beam
with elliptical Gaussian
unit of deconvolved map is Jy per CLEAN beam area
(= intensity, can convert to brightness temperature)
minimize unit problems when adding dirty map residuals
modest super resolution often OK, but be careful
photometry should be done with caution
CLEAN does not conserve flux (extrapolates)
extended structure missed, attenuated, distorted
phase errors (e.g. seeing) can spread signal around
Restored Images
Deconvolution
dynamic range
ratio of peak brightness to rms noise in a region
void of emission (common in astronomy)
an easy to calculate lower limit to the error in
brightness in a non-empty region
fidelity
difference between any produced image and the correct image
a convenient measure of how accurately it is possible to make an
image that reproduces the brightness distribution on the sky
need a priori knowledge of correct image to calculate
fidelity image = input model / difference
fidelity is the inverse of the relative error
Measures of Image Quality
Deconvolution
(Sub)mm Issues
(sub)mm issues
(PWV = Precipitable Water Vapour)
Opacity as a Function of PWV
Variations in the amount of precipitable water vapour (PWV) cause phase
fluctuations, which are worse at higher frequencies, resulting in
Low coherence (loss of sensitivity)
Anomalous pointing offsets
Anomalous delay offsets
Patches of air with different water vapor content
(and hence index of refraction) affect the
incoming wave front differently.
Simplifying assumption:
The timescale for changes in the water
vapor distribution is long compared to
time for wind to carry features over the
array
V
w
~10 m/s
Atmospheric Phase Fluctuations
Can be ameliorated by short integration
times, fast switching, WV radiometry
(sub)mm issues
Pointing: for a 10 m antenna at 350 GHz, primary beam is ~20
- a 3 error A(Gain) = 5% at pointing center, 22% at PB FWHM
- need ~1 pointing accuracy, ALMA pointing goal ~ 0.6
Aperture efficiency, q: Ruze formula gives q = exp([4to
rms
/]
2
)
- for q = 80% at 350 GHz, need a surface accuracy, o
rms
, of 30 m
- ALMA surface goal of 25 m
Au = angular separation between
source & calibrator, can be
large in mm/sub-mm
Ab = baseline error
Baseline determination: phase errors due to errors in the telescope
positions are given by:
A| = 2t Ab Au
to keep A| < Au need Ab < /2t
e.g., for = 1.3 mm need Ab < 0.2 mm
(Sub)mm Antenna Requirements
(sub)mm issues
Instrument stability:
Increases linearly with frequency (delay lines, oscillators, etc)
Millimeter/sub-mm receivers:
SIS mixers, cryogenics needed to achieve low noise
characteristics
Correlators:
Broad bandwidth needed for sensitivity to thermal continuum and
phase calibration
For 300 km s
-1
line, 1.4 MHz @ 1.4 GHz but 230 MHz @ 230 GHz
Limitations of existing and future arrays:
Small FoV of 12 m antenna @ 230 GHz is ~ 30 mosaics
Limited uv-coverage with small number of elements, ALMA helps!
Other (Sub)mm Issues
(sub)mm issues
Radio Telescopes are cool
Single-dish telescopes have fat beams making details hard to see
Interferometers use optics to achieve high resolutions
Antenna pairs sample the FT of the image plane, an inverse FT of
the ensemble of visibilities returns the image
Resulting images are spatially filtered; only compact emission seen
Dirty images can be deconvolved (with care), e.g., using CLEAN
Submillimetre/millimetre interferometry must contend with issues
High atmospheric opacities and phase variability
Given higher frequencies, all instrumentation must be more precise
Summary