Transcript Optical Monitoring

FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
OPTICAL MONITORING
by Luis J. Goicoechea (UC)
● The optical monitoring of GLQs began just after the discovery of the first
gravitational mirage (Walsh, Carswell & Weymann 1979, Nature 279, 381)
A
Q0957+561
(Nordic Optical Telescope/GLITP)
B + cD galaxy
●● The structure of the light curves of the components A and B is an important
tool to decide on the nature of the phenomenon. Apart from an offset in flux
(magnitudes) and a time delay, the macrolens scenario predicts similar light
curves of both components. The brightness record of B must be a replica of the
brightness record corresponding to A.
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
I. OPTICAL MONITORING (OM) PROGRAMS
 THE FIRST DECADE (1980-1990)
For Q0957+561 (double), the first OM programs were conducted by Lloyd
(1981, Nature 294, 727), Keel (1982, ApJ 255, 20), Florentin-Nielsen (1984,
A&A 138, L19), Schild & Cholfin (1986, ApJ 300, 209), Vanderriest et al. (1989,
A&A 215, 1) and Schild (1990, AJ 100, 1771). On the other hand, the system PG
1115+080 (quad) was discovered in 1980 (Weymann et al. 1980, Nature 285,
641), and the pioneering analyses of variability were carried out by Vanderriest
and collaborators (Vanderriest et al. 1986, A&A 158, L5: in French). The
brightness changes in the double system UM 673 = Q0142-100 (Surdej et al.
1987, Nature 329, 695) was followed up by Surdej et al. (1988, A&A 198, 49),
and finally, the variability of the quad system Q2237+0305 (Huchra et al. 1985,
AJ 90, 691) was studied by Irwin et al. (1989, AJ 98, 1989: microlensing!)
CASTLES (http://cfawww.harvard.edu/glensdata)
2
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
 CfA-FLWO (USA) PROGRAM: Q0957+561 FOREVER
Rudy Schild has devoted a big effort to follow up the flux fluctuations of the two
components of Q0957+561. All the data have been taken at Fred Lawrence
Whipple Observatory (1.2 m telescope) in the R band. Although the light curves
(Schild & Thomson 1995, AJ 109, 1970) show a strange behaviour: “for no value
of time delay do the fluctuations align” (Schild 1996, ApJ 464, 125), they permit
to obtain an average delay of about 14 months: 423  6 days (Pelt et al. 1996,
A&A 305, 97: dispersion method), 425  17 days (Pijpers 1997, MNRAS 289,
933: SOLA method) and 424.9  1.2 days (Ovaldsen et al. 2003, A&A 402, 891:
re-reduction of data: dispersion & c2 methods).
 CALTECH-Princeton-APO (USA) PROGRAM: Q0957+561 AND
Q2237+0305
The Gravitational Lens Monitoring Project at Apache Point Observatory (3.5 m
telescope) was not successful with the Einstein Cross (Q2237+0305). However,
the collaboration got to catch two very prominent twin events in the light curves
of Q0957+561A and Q0957+561B (in the g band), which led to an accurate time
delay estimation of ≈ 417 days (Kundic et al. 1997, ApJ 482, 75: c2, optimal
3
reconstruction, dispersion & cross-correlation methods).
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
A
B
BA  gravity (Shapiro
effect) + geometry
VERY
PROMINENT
TWIN EVENTS
B = red
A = blue
4
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
 FLWO-Potsdam (USA/Germany) PROGRAM
An ongoing program with the 1.2 m telescope (Fred Lawrence Whipple
Observatory). Several gravitationally lensed quasars (Q0957+561, SBS
0909+523, H1413+117, B1422+231, PG 1115+080, RXJ 0911+0551 and SDSS
1004+4112) are observed in different optical filters (wavelengths).
 Wise Observatory (Israel) PROGRAM
Using the Wise Observatory 1m telescope, the Tel-Aviv University astronomers
are monitoring in the V and R bands about 30 systems of lensed quasars, in order
to detect the time-delay between the images. Recently, they measured the time
delay between the two components of HE1104-1805 (Ofek & Maoz 2003, ApJ
594, 101: c2 & cross-correlation methods). The new value of ≈ 160 days strongly
disagrees with the previous one of about 310 days (Gil-Merino, Wisotzki &
Wambsganss 2002, A&A 381, 42: dispersion & cross-correlation methods), and
this system merits more attention. There is significant uncorrelated variability,
and it seems that most of the uncorrelated (extrinsic) variability occurs in
component A (the one nearest to the lens): microlensing fluctuations?
5
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
A = red , B = blue: time delay + linear gradient in the flux ratio B/A
6
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
 Belgian-Nordic Group
The Belgian-Nordic group carried out a very intense activity during the past five
years. They participated in several monitoring projects and measured several time
delays at optical wavelengths.
PG 1115+080
(quad): BC = 23.7  3.4 days, AC ≈ 9.4 days (Schechter et al.
1997, ApJ 475, L85)
7
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
(double): 51  4 days (Burud et al. 2000, ApJ 544, 117: SOLA,
dispersion, c2 & iterative methods). They used the Nordic Optical Telescope
(NOT) and the I optical filter.
B1600+434
A
B
Problems:
• Time dependent offset: microlensing?
• Main event in B: another
microlensing?
• Main events seem to be shifted in time
8
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
(double): 103  12 days (Burud et al. 2002, A&A 383, 71: c2
method). Using the Danish 1.5 m telescope at La Silla Observatory (ESO, Chile)
and the V optical filter.
HE 2149-2745
No problems!
9
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
(quad): time delay of 146  8 days between A = A1 + A2 + A3
and B (Hjorth et al. 2002, ApJ 572, L11: iterative method). Using the NOT at
Roque de Los Muchachos Observatory (Spain) and the I optical filter.
RXJ 0911.4+0551
Offset by values between
-1.95 mag and – 2.05 mag
10
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
(double): 130  3 days (Burud et al. 2002, A&A 391, 481:
iterative method). Using the NOT at Roque de Los Muchachos Observatory
(Spain) and the R optical filter.
SBS 1520+530
11
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
(double): 16  2 days (Jakobsson et al. 2004, astroph/0409444). Using the NOT at Roque de Los Muchachos Observatory (Spain)
and the R optical filter.
FBQ 0951+2635
Problems… linear gradients… more
problems…
Only the last 300 days are
considered…problems in the beginning:
microlensing, multiple delays?
12
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
 Maidanak Collaboration (Russia/Ukraine/Uzbekistan + Germany)
The former Soviet Union group and the Potsdam group carry out an important
activity at the Maidanak Observatory (Uzbekistan), by using the 1.5 m telescope.
Joint efforts in Potsdam (Germany), Tashkent (Uzbekistan), Moscow (Russia)
and Kharkov (Ukraine) permit to develop a GLQ monitoring program. The
targets are: Q2237+0305 (microlensing!), SBS 1520+530, Q0957+561, SBS
0909+523, H1413+117, B1422+231, PG 1115+080, RXJ 0921+4528 and UM
673 = Q0142-100.
 Spanish Collaboration
The Spanish collaboration began in the summer of 1995, and it is currently
conducted by the IAC (Tenerife), UC (Santander) and UV (Valencia) groups.
BEFORE 1995…
IN A NEAR FUTURE…
13
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
From two prominent twin events in the light curves of Q0957+561A and
Q0957+561B (in the R band), the collaboration estimated an accurate time delay
of 425  4 days (Serra-Ricart et al. 1999, ApJ 526, 40: d2 method). They used the
Spanish 82 cm (IAC-80) telescope at Teide Observatory (Tenerife, Spain).
NO PROBLEMS, NO
EVIDENCES IN FAVOUR
OF MICROLENSING
A (time delay-advanced) = green circles
B (magnitude-shifted) = red squares
14
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
What is the time delay in Q0957+561? averaged ≈ IAC events ≈ 423-425 days or
APO main events ≈ 417 days? …
417 days
424 days
417 days
Coming back to
the APO data in
the g band …
(Goicoechea
2002, MNRAS
334, 905)
432 days
d(BA) =
(Dd/cDds) (1 + zd)
[(qA – qB) . dr]
15
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
Other things:
of Q2237+0305 (Alcalde et al. 2002, ApJ 572, 729) and
Q0957+561 (Ullán et al. 2003, MNRAS 346, 415). Using the NOT, and the V
and R filters
● GLITP monitoring
UC group also observed Q0957+561 and SBS 0909+523 from 2003
March to 2003 June. The new VR observations were made with the 1.52-m
Spanish Telescope at Calar Alto Observatory, Almeria (Spain). The analysis and
interpretation of the VR light curves are in progress (in collaboration with the
Oslo group)
●● The
●●● Finally, the
Spanish groups are involved in two monitoring programs with the
2.00-m Liverpool Telescope, which is a fully robotic telescope at the
Observatorio del Roque de Los Muchachos, La Palma (Spain).
P1: Daily monitoring of SBS 0909+523 and Q0957+561
P2: Photometry and spectroscopy of several GLQs
16
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
II. OPTICAL PHOTOMETRY (or MAGIC?)
Due to the usual small angular separation between the lensed components and
the proximity of one or several components to the lens galaxy, from groundbased frames, the photometry of a multiple QSO is remarkably complex (a
magical task?). The size of the seeing disk (atmospheric effect) is of about 1-2”,
i.e., similar to the typical angular size of the gravitational mirages.
Double star Z Aquarii, which has a
separation of 2” (Sky & Telescope,
“Beating the Seeing” by A. M.
MacRobert)
Q0957+561
Q2237+0305
2”
6”
17
A1+A2+A3+Lens
B+Lens
A+B+C+D+Lens
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
SOME PHOTOMETRIC TECHNIQUES (sorry if your own
technique is not included here)
 Deconvolution (Magain, Courbin & Sohy 1998, ApJ 494, 472: BelgianNordic group). This task combine all the frames (optical images) obtained at
different epochs to determine the numerical light distribution of the extended
source (lens galaxy) as well as the positions of the point sources (QSO
components), since these parameters do not vary with time. The flux of the point
sources are allowed to vary from image to image, which produces the light
curves.
 Aperture/PSF photometry (Serra-Ricart et al.
1999, ApJ 526, 40; Ovaldsen et al. 2003, A&A
402, 89: large separation double systems, i.e., >>
1”). For example, the PHO2COM task (by Miquel
Serra) works in the following way: if the FWHM
of the seeing disk is q, we initially take two circles
of radius q, which are centered on the two
components A and B. A reference star (reference
candle) S is also encircled by a ring of radius q.
2q
A
B+lens
18
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
PHO2COM:
The reference star total flux is extracted through aperture
photometry with an aperture of radius 2q, i.e., we integrate the instrumental flux
(counts) within a circle of radius 2q (using a clean frame, without background
signal).
central
axis
2q
FS(2q) = pixels  2q FS(pixel)
S
CCD pixels
PSF (point spread function) fitting photometry, within the initial circle of
radius q, is applied to all the objects (QSO components and reference star).
1D: central
axis for S (FS)
FWHM = q
Fit in 2D (e.g., Gaussian) 
Integration in 2D 
FS*(q), Fi*(q) (i = A,B)
We compute an aperture correction f = FS(2q)/FS*(q) and QSO total
fluxes: Fi(2q) = f  Fi*(q).
19
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
PHO2COM:
Finally, one can compare the QSO component fluxes with that of
the reference star. As the magnitude of an object is defined by m = - 2.5 logF
+ k, the component-to-star flux ratios are equivalent to magnitude differences:
mi – mS = - 2.5 log[Fi(2q)/FS(2q)].
Although we dealth with a large separation (between components) system, one
of its components could be close to the lens galaxy. If the component B and
the deflector are close enough, and the lens galaxy is relatively bright, then the
PHO2COM flux of the contaminated component (B) will be overestimated.
q
central
axis
Excess of flux!
B + Lens
Fit in 2D 
Integration in 2D 
FB**(q) = FB*(q) +
contamination
Therefore, FB(2q) = f  [FB*(q) + c] and mB = mS - 2.5
log[Fi(2q)/FS(2q) + c/FS(2q)] = mS - 2.5 log[Fi(2q)/FS(2q)] – C (C > 0)
20
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
For example, the contamination of Q0957+561B in the R band at the NOT is
given by a law C = mB(true) – mB(PHO2COM) = a  q + b.
Both the galaxy/component B confusion and the bias in the
component flux from PHO2COM will depend on the seeing (FWHM
= q). The contamination by galactic light can be derived from
simulations or from another data processing technique (which
avoids that contamination).
21
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
 Image subtraction (Alard 2000, A&AS 144, 363; Wozniak et al. 2000, ApJ
529, 88; Alcalde et al. 2002, ApJ 572, 729: bright galaxy lens). Observations of a
multiply imaged QSO with a bright lens galaxy are well adapted for optimal
image subtraction designed to efficiently remove the galaxy contribution without
modeling its photometric profile. This task co-adds the frames to built up a deep
reference frame (reference image). This reference frame is subtracted
subsequently from each of the individual CCD frames in order to obtain
differential images. The lens galaxy disappears “totally” after the optimal
subtraction, and the observed residuals essentially correspond to the variable
lensed QSO images. In the last steps (apart from the comparison with the
reference star and the error estimates), one must perform a simultaneous N-PSF
fitting photometry (N is the number of QSO components) of all differential
frames obtained by optimal subtraction in order to measure for each of the N
lensed components the respective flux differences between each individual frame
and the reference one. In order to get the absolute fluxes at all observing epochs
for each of the N components, we also need to perform direct photometry on the
reference frame.
22
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
OGLE collaboration: nine difference images of Q2237+0305 at various epochs
(V-band)
23
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
 PSF fitting (e.g., McLeod et al. 1998, AJ 115, 137; Ullán et al. 2003, MNRAS
346, 415: CASTLES and Spanish collaborations). For example, the PSFPHOT
task is useful for extracting clean QSO fluxes (free of background signal, crosscontamination between the components and contamination by the galaxy light).
The technique uses an observationally motivated 2D galaxy profile (e.g., de
Vaucouleurs profile), which is convolved with a PSF to reproduce the observed
lens galaxy profile. Thus the instrumental fluxes of the N QSO components and
the reference star (S) can be measured by means of PSF fitting. Obviously,
constant backgrounds are included to model the two regions of interest: the lens
system and the S star. In this photometric procedure, the clean two-dimensional
profiles of the field stars are used as empirical PSFs. Analytical PSFs are not
considered (e.g., Gaussian law and so on).
STEP-TO-STEP
1.- To determine the relevant information on the galaxy, the method
uses the best images in terms of seeing values (superbimages).
Therefore, the code is applied to each image with a seeing (FWHM)
better than qlim, considering the PSF of the brightest field star and
allowing all parameters to be free.
24
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
For some frames, the method could not be able to accurately extract several
physical parameters of the lens galaxy, leading to results in apparent
disagreement with the global distributions. Thus, in the estimation of each
parameter, it is followed a scheme with two parts. First, the values with a
deviation (= value - average) exceeding the standard one are dropped. Second,
the parameter is inferred from the average of the "surviving" values. Finally, one
obtains the morphological parameters of the galaxy (e.g., the effective radius,
Reff, the ellipticity, e, and the position angle, P.A.), the relative position of the
galaxy (position relative to the brightest QSO component) and the relative flux G
= Flens/FS.
2.- The code is applied to all images (whatever their seeings), setting the
galaxy parameters to those derived in the previous step (e.g., Reff, e,
P.A., relative position and G), using galaxy fluxes given by Flens = G  FS
and allowing the remaining parameters to vary. In this last iteration,
although all the available PSFs are used, the fluxes from the PSF of the
brightest field star are regarded as the standard ones. For the icomponent (i = 1,N):
yi = mi – mS = - 2.5 log(Fi/FS).
25
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
The light curves yi(t) show global behaviours gi(t) that are caused by true
intrinsic or extrinsic phenomena (e.g., linear trends or prominent events), as
well as day-to-day variabilities, ri(t) = yi(t) – gi(t), whose interpretation is not
easy: true daily fluctuations or observational noise?. For example …
eA  <rA2>1/2 ≈ 5 mmag
gA(t)
(in principle, the
residual signal contains
both observational noise
and true variability)
rA(t)
eE = <(yE - <yE>)2>1/2 ≈ 4 mmag
NO true variability
eA ≈ 5 mmag
26
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
In some systems, the situation is relatively “simple”. If the gravitational mirage
consists of two QSO components, and the lens galaxy is clearly fainter than the
multiple QSO, each frame (region including the system) can be explained in
terms of the background signal (free parameter #1) and two light distributions
with the shape of a stellar PSF [free parameters: flux of A (#2), flux of B (#3),
2D position of A (#4-5) and 2D position of B (#6-7)].
Each night, to estimate a reliable error, one may take several consecutive
frames, and obtain the average value and the standard deviation of the
consecutive measurements yi(1), …, yi(M). This is a statistically and
physically conservative approach, because the s value is not divided by M1/2
and any possible true intraday (hourly) variation is neglected. Of course, the
intrinsic uncertainty only can be tested by using complementary photometric
tasks.
SBS 0909+523
CASTLES (http://cfawww.harvard.edu/glens
data) - HST image
27
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
The system SBS 0909+523: frame taken with the 1.52-m Spanish Telescope at
Calar Alto Observatory, Spain
c2 fit
The code starts with
a set of seven initial
values (initial
model):
ptsrc[0].intens
ptsrc[1].intens
ptsrc[1].x
ptsrc[1].y
background
x0
y0
Model frame
Difference frame
28
FIRST ANGLES SCHOOL: MEASURING THE HUBBLE CONSTANT AND LENS MASS MODELLING
Santander (Spain), 10 - 14 December 2004
III. TIME DELAY ESTIMATION
 DISPERSION METHOD (D2): Pelt and collaborators (Pelt et al. 1994, A&A
256, 775; 1996, A&A 305, 97) developed this statistical technique that is based
on the minimum dispersion between the two brightness records
 ITERATIVE METHOD: the two light curves are correlated through an
iterative procedure (e.g., Burud et al. 2000, ApJ 544, 117)
 CROSS-CORRELATION METHODS: there are several variants. The most
sophisticated are: discrete cross-correlation function (Edelson & Krolik 1988,
ApJ 333, 646), ZDCF (Alexander 1997, in Astronomical Time Series,
Kluwer, p. 163) and d2 test (e.g., Serra-Ricart et al. 1999, ApJ 526, 40). This
last technique (d2) is based on the expected similarity between the discrete
autocorrelation function (DAC) of the light curve of one component and the
discrete cross-correlation function (DCC), and it is useful to derive a very
accurate delay, provided that a first (rough) estimate is known. Updated
information on the d2 test and other UC group materials (software, images and
so on) is available at https://grupos.unican.es/glendama/ 29