This action might not be possible to undo. Are you sure you want to continue?
EuroSummer School
Observation and data reduction with the Very Large Telescope Interferometer
Goutelas, France
June 416, 2006
Guy Perrin
Observatoire de Paris
10 June 2006
Disclaimer
This is not a good lecture !
Spatial frequency
¢( f ) = +(t)e
÷2itft
dt
÷·
+·
}
Frequency f and time t are conjugated variables by the
Fourier transform.
v=1/T
Fourier
transform
t f
÷1/T
+(t) = cos(2t
t
T
)
¢( f ) =
1
2
o( f ÷
1
T
) +o( f +
1
T
)

\

.

Angular coordinates are natural when the source is at a large distance from the
diaphragm (or pupil). This is in particular true in astronomy for which objects are
located at quasi infinite distances.
Spatial frequencies are not so obvious but this notion is not so difficult. A spatial
frequency is the reciprocal of a characteristic scale of an object :
Spatial frequency coordinates have reciprocal dimensions with respect to
coordinates in direct space (m m
1
).
Direct coordinates and Fourier space coordinates are conjugated by the Fourier
transform.
o
x
1
z
x
y
P
f
x
f
y
+1/P 1/P
Fourier
Transform
Diffraction, imaging, Optical Transfer Function
Fraunhofer Diffraction
Hypotheses :
• 1. Small diffraction angles :
• 2. Diaphragm and screen are small w.r.t. z : 1/s ≈ 1/z
• 3. z much larger than the Rayleigh distance:
z
R
=
x
0
2
+ y
0
2
( )
max
ì
cos_ ~1
_
(rad
1
or arcsec
1
)
(rad or arcsec)
u =
x
0
ì
v =
y
0
ì
¦
´
¦
¦
¹
¦
¦
o =
x
1
z
 =
y
1
z
¦
´
¦
¦
¹
¦
¦
¦
´
¦
¦
¦
¦
¦
¹
¦
¦
¦
¦
¦
¢
1
(o,) ~
e
it
z
ì
2+o
2
+
2
 
i
z
ì
+
0
(ìu, ìv)e
÷2it ou+v  
diaph
}}
dudv
¢
1
(x
1
, y
1
) ~
e
2it
ì
z
iìz
.e
it
ìz
x
1
2
+y
1
2
 
+
0
(x
0
, y
0
)e
÷
2it
ìz
x
0
x
1
+y
0
y
1
 
diaph
}}
dx
0
dy
0
Traditionally, variables are changed into :
Pupil plane
Image plane
The diffracted field is proportional to the Fourier Transform of the field in
the pupil plane :
¢(o,) ·TF +(u, v) { }
I(o,) = +(o,)
2
t
÷ +(o,)
2
The image is the convolution of the source spatial intensity distribution by the point
spread function
o

Pupil function
P
Image plane
O(o
0
,
0
)
Im(o,) = O(o
0
,
0
).I(o÷o
0
, ÷
0
) do
0
d
0
}}
=OI (o,)
Point source response:
Extended object:
Image spectrum :
The optical system is a lowpass filter with a cutoff frequency D/ì
Image frequency contents
S(u,v) = FT Im(o,) { }= FT O(o,) { }.OTF(u,v)
Object
x
FT
Object spatial spectrum
=
Image spatial spectrum
f
f D
ì
D
ì
The same image formation theory applies to the interferometer
The interferometer as a bandpass filter
+D/ì D/ì
+D/ì D/ì B/ì +B/ì
1
1/2
Pupil Optical transfer function
+D/ì D/ì B/ì +B/ì
1
Source spatial spectrum
(Visibility)
Source starts to be resolved
What happens for
an extended source ?
D
B
f
f
f
Fourier plane sampling
© C.A. Haniff
(u,v) tracks
(u,v) tracks are ellipses whose center is on the v axis. (u East, v North)
General equation for (u,v) tracks :
B
x
, B
y
and B
z
are the coordinates of the baseline vector projected onto the axes
pointing towards East, North and the meridian, respectively
Particular cases :
 o = 0° : (u,v) tracks are straight lines parallel to the u axis
 o = 90° : (u,v) tracks are circles centered on the origin
u
2
+
v ÷ B
z
ì
( )
coso
sino

\

.

2
=
B
x
2
+ B
y
2
ì
2
Examples of (u,v) coverages at IOTA
(Arizona)
Three 45 cm relocatable siderostats
Lattitude = 31.4°
Hour angle range : 4h , +4h
80
80
60
40
20
0
20
40
60
80 80 60 40 20 0 20 40 60
(u,v) t races
o = 0°
Examples of (u,v) coverages at IOTA
(Arizona)
Three 45 cm relocatable siderostats
Lattitude = 31.4°
Hour angle range : 4h , +4h
o = 90°
80
80
60
40
20
0
20
40
60
80 80 60 40 20 0 20 40 60
(u,v) t races
Examples of (u,v) coverages at IOTA
(Arizona)
Three 45 cm relocatable siderostats
Lattitude = 31.4°
Hour angle range : 4h , +4h
o = +45°
80
80
60
40
20
0
20
40
60
80 80 60 40 20 0 20 40 60
(u,v) t races
Coherence
Coherence of light waves
Let be a light wave.
As for random variables, a correlation can be defined between fields at
different times or at different locations.
In the first case, this is called temporal coherence:
In the second case, it is spatial coherence:
E(P ,t)
Corr(E(P , t), E(P , t + t)) =
E(P , t) ÷ E(P , t)
( )
. E(P , t + t) ÷ E(P , t + t)
( )
*
E(P , t) ÷ E(P , t)
2
E(P , t + t) ÷ E(P , t + t)
2
=
E(P , t).E
*
(P , t + t)
E(P , t)
2
E(P , t + t)
2
Corr(E(P , t), E(P + AP , t)) =
E(P , t).E
*
(P + AP , t)
E(P , t)
2
E(P + AP , t)
2
Temporal coherence
Measuring the temporal coherence of the electric field means measuring:
How shall we do?
E(P ,t).E
*
(P ,t +t)
At >>1/ v
The Michelson interferometer
At optical wavelengths?
The Michelson interferometer
(polychromatic case)
Fields at different wavelengths are not coherent.
The interferogram measured with the Michelson interferometer is therefore the
sum of mochromatic interferograms in the band:
(o = 1/ì is the wavenumber)
The complex degree of coherence in the polychromatic case is therefore:
(proof: WienerKhintchine theorem):
The interferogram of the Michelson setup therefore allows to measure the
spectrum of the source principle of the Fourier transform spectrometer
I
tot
(t ) = I
tot
(ì,t )
Aì
}
dì = I
tot
(o, t )
Ao
}
do
= 2I + 2Re I(o)exp(2ito.ct )do
Ao
}
¸
(
¸
(
¸(t) =
I(o)exp(÷2ito.ct) do
Ao
}
I(o) do
Ao
}
Coherence length
l
c
=
1
Ao
=
ì
2
Aì
Number of fringes
=
l
c
ì
=
ì
Aì
Opd in units of ì
Spatial coherence
Measuring the spatial coherence of the field means measuring:
How shall we do?
E(P ,t).E
*
(P + AP ,t )
At>>1/ v
The Young slit experiment
At optical wavelengths?
The spatial interferometer
(extended source)
The source is extended and uncoherent (i.e. the fields emitted by
individual points are not correlated).
The intensity of the field in the focal plane is:
Which yields for the complex coherence factor:
I
tot
(B ,t) = E(P, S , t) + E(P + B , S , t + t)
2
source
}
d
2
S
= 2 I(S , ì)
source
}
d
2
S + 2Re I(S , ì)
source
}
exp i et ÷2tS .
B
ì

\

.

¸
(
¸
(
(
d
2
S

\

.


= 2 I(S , ì)d
2
S
source
}
+ 2 I(S , ì)d
2
S
source
}

\

.

Re µ
12
B
( )
exp iet
( )
( )
µ
12
(B ) =
I(S , ì)
source
}
exp ÷2itS .
B
ì
¸
(
¸
(
d
2
S
I(S , ì)d
2
S
source
}
• This result is known as the ZernikeVan Cittert theorem : the complex
coherence factor is equal to the Fourier transform of the spatial intensity
distribution of the source.
• The conjugated coordinates are the angular direction S and the spatial
frequency B/ì.
• Measuring a value of the complex coherence factor yields a value of the
spatial spectrum of the source.
• The interferometer is therefore a bandpass filter (as opposed to a single
pupil which is a lowpass filter) giving access to the very high spatial
frequencies of the source.
• Measuring the complex coherence factor at several spatial frequencies allows
to restore the spatial intensity distribution of the source.
> aperture synthesis technique
V
The complex coherence factor is usually called the complex visibility:
• the modulus is the fringe contrast of the interference pattern
• the phase is derived from the position of the central fringe with respect
to the zero optical path difference (zero opd or zopd):

¢ =
2to
ì
Image plane
f
D
Angular resolution :
ì
D
f ×
ì
D
> ì
Interferometry
ì D
Image plane
f
B
Angular resolution :
ì
B
f ×
ì
D
Interferometry
f ×
ì
B
< ì
Image plane
f
B
Angular resolution :
ì
B
f ×
ì
D
Interferometry
f ×
ì
B
ì
Image plane
f
B
Angular resolution :
ì
B
f ×
ì
D
Interferometry
f ×
ì
B
ì B
V
cosx dx = 0
÷¢
+¢
}
when ¢ = t
hence ¢ = t =
2t
ì
×
BAo
2
¬Ao =
ì
B
> ì
The optical interferometer
Astronomical interferometry in practice
S
variable delay (B.S)
Delay line
B
Model fitting and visibility models
Two examples of visibility functions
Uniform disk: with u the angular diameter.
The visibility function is equal to:
Visibility function (modulus) :
I(S ) = H(
S
u
)
V(B ) =
2J
1
tuB
ì

\

.
tuB
ì
0
0,2
0,4
0,6
0,8
1
V
i
s
i
b
i
l
i
t
é
Base B
1,22 ì/ u
u
Binary star:
The visibility function equals:
Visibility function (modulus) for B//o :
I(S ) =o S ÷
1
2
o

\

.
+o S +
1
2
o

\

.
V(B ) = cos
to .B
ì

\

.

o
0
0,2
0,4
0,6
0,8
1
V
i
s
i
b
i
l
i
t
é
Base B
ì/ 2 o
Quiz
Modulus Phase
Object 1
Object 2
Object 3
Atmospheric turbulence
Atmospheric turbulence
atmosphere
Temporal and spatial
fluctuations of T and
therefore n(T) alter light
paths
Spatial scale of correlation
of phase: r
0
(Fried
diameter), a few cm at
visible wavelengths (~ì
6/5
)
distorted wavefront
incident wavefront
We lose twice:
1. The object is apparently more
extended hence less spatially coherent.
2. The object looks different in the two
telescopes poorer correlation.
The issue of atmospheric turbulence
Spatial coherence loss
AO corrected image
Phase: closure, differential
Atmosphere Corrupts the Phase
The absolute phase is illdefined
© J.D. Monnier
Differential Phase
Wavelength
F
l
u
x
only emission continuum line
~ ~ ~
V V V
'
+
'
=
'
Prime indicates corrupted
by atmospheric piston
Measured Interpolated
If you make assumption
about V_continuum (e.g.,
unresolved), then
V_emission can be
determined fully
© J.D. Monnier
Differential Phases with VLTIAMBER:
What might this be?
Visibilities Differential Phase
© J.D. Monnier
The “Closure Phase” Is Not Corrupted
© J.D. Monnier
How Much Phase Information?
Closure Phases are not all independent from each other.
Number of Closure Phases Number of Fourier Phases
Number of Independent Closure Phases
© J.D. Monnier
Summary of a few Important Points
The closure phases are independent of all telescopespecific phase errors.
The closure phases are all 0 or 180 degrees for sources with point
symmetry.
Object must be resolved (~> half fringe spacing B/ì) to have nonzero CP
 CP · (baseline)^3
 Phase · (baseline)
Based on J.D. Monnier
Calibration
Calibrator Sources
Formally, anything can be a calibration source.
However, assessing our measurement accuracies we must account for
uncertainties in our ability to predict the properties of the calibration
source:
Traditionally (realistically) this has meant that we choose calibrators
whose properties are as simple as possible – single stars!
V
trg
2
= (V
model÷cal
2
/V
meas÷cal
2
)V
meas÷trg
2
oV
trg
2
·(cV
model÷cal
2
/cmodel) o
model
oV
trg
2
·(cV
model÷cal
2
/cO) o
O
Choose star diameter as small as
possible as then the derivative is 0
Based on A. Boden
Estimating Angular Sizes of Stars (3)
• Because stellar size is proportional to (sqrt
of) brightness, bright & “pointlike” does not
formally exist
• Source resolution is a question of instrument
resolution and sensitivity
• For given apparent brightness, “hotter” stars
appear smaller
Homework: Recompute for
Apparent K Magnitude
2
) ( 2 . 0
4
5800
10 17 . 8
4
÷
+ ÷
(
¸
(
¸
  =
= O
K
T
mas
T
F
eff BC V
eff
Bol
o
• So estimating apparent size is easy:
all you need is bolometric flux
and effective temperature!
• In this sense effective temperature
is just a proxy for surface brightness
© A. Boden
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.