Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004

Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Cluster Colour-Magnitude Diagrams

Abstract
The aim of this project was to construct a colour-magnitude diagram of one selected open
cluster, M67, from just supplied CCD images taken through different filters. CCD images
are an invaluable current source of gathering useful scientific information, but in order to
maximize its potentially, the data conversion process must be carefully performed.

After photometric data have been properly extracted, it was converted into standard
system values that finally allowed the construction of the cluster diagram. From its
analysis, it was possible to infer an estimation about M67 distance and age. The
comparison to published values finally confirmed the validity of the overall executed
procedure.

Introduction
This project report will begin with a succinct analysis of the particularities that cluster
diagrams show, underlying that just from the morphological analysis of each diagram
many essential cluster features can be found out.

Next, it will be presented a detailed description about the complete translation process
from raw CCD images to photometry data expressed in an “universal” system. This will
include the fundamental topics of data reduction, aperture photometry, and
standardization.

Having analyzed the conversion process, its practical application for the open cluster M67
will immediately follow up. It will be described the provided raw CCD images, used
software, found peculiarities, obtained data and calculated errors. Some processed images,
and a spreadsheet containing all processed data will be attached.

The gathered information will finally allow the construction of the M67 CMD. An analysis
of the physical significance of its constituents will be done, resulting in an estimation
about M67’s age and distance, as well as expected uncertainties involved. Both estimations
will be compared to published data. Lastly, a discussion about the found discrepancies will
propose some possible source of errors and the suggested way to avoid them.

Page 1
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

1) Why star cluster diagrams are so particular?

Star clusters are groups of stars that appear concentrated in small regions of the sky. As it
can be assumed that all star members of a cluster has been formed essentially at the same
time from a single primordial cloud of gas and dust, those stars achieve three fundamental
common features: (1) they lay at the same distance from us; (2) they have the same age;
and (3) they have the same chemical composition. Therefore, stellar clusters become ideal
specimens to study the evolution of stars [1].

In order to facilitate any data analysis, a proper graph allows to visualize peculiar features
of the sample, such as preferred values, distribution tendencies, prohibited relations, etc.
For stellar study, this graph is the classic Hertzsprung-Rusell (H-R) diagram, which plots
star luminosity (or absolute magnitudes) against their surface temperatures (or spectral
types). In particular, if the horizontal variable is expressed as the radiation difference
measured from two different wavelengths (the colour index ), such a diagram is referred to
as a colour-magnitude diagram (CMD).

Luminosity or absolute magnitude values implies the necessity of knowing the actual
distances to stars. On the other hand, surface temperatures or colour indexes can be
properly estimated independently of distance. Hence, the practical difficulty for
constructing general H-R diagrams is always to know the corresponding distance-
dependent values.

For stars lying at the same distance from us (as it certainly is the case for stellar clusters),
their apparent magnitudes share a common factor of proportionality with their absolute
magnitudes. Therefore, the H-R diagram for a cluster will show the same relative data
positions by using distance-dependent absolute magnitudes or the straightforward
measured, distance-independent, apparent magnitudes.

This great simplification in the construction of cluster H-R diagrams or cluster CMDs,
while still maintaining all their full analysis potentiality, makes them a fundamental tool in
both quantitative and qualitative studies of star evolution. The particular morphology of
any cluster CMD does reveal a lot about its own characteristics.

2) What features do a cluster CMD reveal?

All stars evolve, some very fast, some very slowly, critically depending on their masses.
The more massive a star is, the lesser its lifetime. Mass defines how long any star is going
to live and the kind of process to perform in its final period of life.

But independently of mass, from birth up to 90% of its lifetime, each star evolves in a
similar manner. Stars spend most of their lives burning hydrogen slowly in a state of stable
equilibrium, and this is the reason why any H-R diagram do not show randomly scattered
points over the entire graph, but mostly concentrated on a band stretching diagonally. This
band is called the Main Sequence, and goes from the upper left (hot, luminous, blue stars)
to the lower right (cool, dim, red stars).

Page 2
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

When the hydrogen supply in the core of the star is depleted, hydrogen burning is no
longer possible. This ends the main sequence phase of the star’s life, and hence it moves
out from its long during location on the H-R diagram towards the upper right. This new
accessed region is known as the Red Giant Branch, and corresponds to the initial stages of
the helium burning phase.

Therefore, any considered sample of stars of the same age will necessarily integrate a quite
particular stellar diagram. Figure 1 shows a typical CMD of a star cluster, where the
massive shorter-lifetime stars have already evolved abandoning their Main Sequence
location towards the Red Giant Branch, while many less-massive, long lifetime stars are
still placed at the Main Sequence.

Figure 1
A typical cluster colour-magnitude diagram

Considering star clusters, the point on a H-R diagram or CMD at the top of the still
surviving Main Sequence, where the Red Giant Branch seems to begin, is called the turn-
off point. Knowing its precise location on the diagram gives an important clue to the age of
the cluster, as it is possible by comparison to find out the mass (MMS ) of stars that have just
left the Main Sequence and hence the desired time (t) can be estimated from

t ≈ (MMS) -2.5 (1)

where both magnitudes are expressed relative to the Sun’s mass and lifetime [2].

Page 3
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Another straightforward way to estimate the cluster age is just by comparison the absolute
magnitude of the turn-off point to isochrones (diagrams plotting absolute magnitude versus
Main Sequence lifetime derived from theoretical considerations), as Figure 2 depicts.

Figure 2
Absolute Visual Magnitude of the main sequence turnoff from Hodge's (1983) calibration,
from Van den Berg's (1985) isochrones, and from models from Chiosi et al. (1986)

The distance to the star cluster can be find out by comparing its CMD with a standard
Main Sequence CMD (a general graph of absolute magnitude versus colour index, as
Figure 3 shows) of the same vertical and horizontal scales.

The two diagrams have to be overlapped, specifically observing the right correspondence
of the colour index values from each graph. The diagrams are moved vertically until the
bottom edge of the standard Main Sequence falls along the surviving Main Sequence stars
of the cluster. For any star of that common portion, its known apparent magnitude (m)
from the cluster CMD corresponds to the absolute magnitude (M) that can be read in the
standard diagram.

Therefore, the distance modulus (m – M) becomes known, and then the distance to the
cluster (d, in parsecs) can be calculated from

 m −M + 5 
 
d= (2)
10  5 

Page 4
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Figure 3
Left: A general CMD of near stars
Right: A sketched CMD only showing the Main Sequence

In finding this preliminary distance from equation (2) it has been omitted the reddening
effect of the interstellar dust, which certainly causes a wavelength variable absorption.
Although this extinction factor can be considered, thus improving the accuracy of the final
“corrected” distance, the achieved improvement is usually negligible compared to the error
introduced in the absolute magnitude of the turn-off point by the applied method.

Finally, the construction of a cluster CMD also offers information about its metallicity,
that is, an idea about the relative abundance of its constituent elements [3].

Page 5
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

3) Obtaining the required data for cluster CMD construction

As previously mentioned, for constructing a cluster CMD it is necessary to find out the
colour index and apparent magnitude of each star member, as precisely as possible. Those
values can be accurately obtained from photometric measurements derived from CCD
images. However, those images must be carefully processed in order to assure that the
derived data is a faithful representation of the real world.

a) Data Reduction
Before attempting any photometry measurement, the CCD stellar images (generically
called “science” images, as they contain scientific useful data) have to be “normalized”.
This normalization task refers to the elimination (or at least, to its maximal reduction) of
the noise that both the CCD camera and the overall optical system has certainly introduced
per se in the resulting science images.

In the case of CCD professional cameras, which operate at very low temperatures, the
main causes of noise come from non-uniform responses from pixels and electronics. The
optical system also introduces noise due to non-uniform responses from optical
irregularities or randomly distributed dirties.

The electronic noise introduced by the CCD camera can be easily removed from any
image by subtracting a bias frame, that is, an exposure of “zero duration” taken with the
camera shutter closed. This special frame represents the zero-point or base-line signal
from the CCD camera [4].

All the non-uniform responses, from both CCD and optical system, can be eliminated from
any image by dividing it by a flat field, that is, an image of a perfectly even illuminated
field. Any feature appearing in this special frame represents distortions introduced by the
proper imaging system.

As the use of each different filter actually means a different overall optical system (each
filter has its own irregularities), a flat field for each filter must be obtained.

Therefore, any image must undergo a two-stage process prior to any other action: firstly,
the bias frame has to be subtracted; secondly, the resulted image has to be divided by the
corresponding flat field.

It is highly recommended to average a set of many bias frames, in order to obtain a master
bias that will result uncontaminated by any possible transitory effect (as the frequent
appearing of cosmic rays). Usually, at least five bias are combined to obtain the master
bias. In a similar manner, many flat fields of the same filter should be combined to assure
the achievement of a more “representative” master flat for that filter.

b) Aperture photometry

Page 6
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Having the science images been reduced, any specific imaging processing software can be
applied in order to obtain photometry values for each desired star. This is automatically
performed by means of a procedure called aperture photometry. It consists in the
following steps [5]:

1. The operator establishes the radius in pixels (the “aperture”) of a region, centered
in the star, thought to contain almost all its light.
2. The operator establishes the inner and outer radii in pixels (the “annulus”) of a
region, centered in the star, thought to contain no starlight at all, just “background”.
3. The operator selects an approximate center of the star image.
4. The program computes the exact centroid of the star image.
5. The program determines which pixels lie within the aperture, sums their counts,
and computes the average pixel value in the aperture.
6. The program determines which pixels lie within the annulus, sums their counts, and
computes the average pixel value in the annulus.
7. The program subtracts the average pixel value in the annulus from the average
pixel value in the aperture, and gives a final photometric value.

The obtained photometric values are measurements of the counts from pixels only
belonging to any considered star. This process permits to obtain raw instrumental
magnitudes, that is, just comparative magnitude values that has not been yet corrected for
the particular characteristics of the used telescope and CCD detector that captured the
measured images, nor for the particular atmospheric absorption at the time of each image.

In order to select an optimum radius aperture, that is, one that provides the best signal-to-
noise measurement, it is recommendable to previously analyze the photometric curve of
growth. This is a graph that shows how the raw instrumental magnitude of a star varies
with changes of the radius aperture used to determine it. The graph usually starts out low
and rises to a plateau 1. The aperture should be selected balancing between a larger value
(which assures that no starlight will be missed) and a lower one (which assures the lesser
possible measurement errors due to the background contribution) [6].

c) Standardization
In order to find out objective stellar apparent magnitudes, that is, a standard system, it is
necessary to transform the obtained raw instrumental magnitudes, eliminating all the
particularities of the measurement process. As said, this implies to consider the distortions
introduced by the proper imaging system and by the atmosphere as well.

Therefore, the equations to be resolved 2 are [7]:

mB = B + x1 (B) + x2 (B) * (B – R) + x 3(B)*X(B) (3)

mR = R + x1 (R) + x2 (R) * (B – R) + x 3(R)*X(R) (4)

1 For good photometry results, the plateau region of the curve of growth should achieve an almost constant
value (variation lesser than ± 0.05 magnitudes) [6].
2
The particular case to be analyzed is for transforming instrumental magnitudes mB and m R into standard
magnitudes B and R, but the equations are basically the same for whatever other pair of used filters.

Page 7
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

where: mB is the star’s raw instrumental magnitude measured by the software on an image
obtained through a standard B (blue) filter;
mR is the star’s raw instrumental magnitude measured by the software on an image
obtained through a standard R (red) filter;
B is the star’s true apparent magnitude measured through a standard B (blue) filter;
R is the star’s true apparent magnitude measured through a standard R (red) filter;
x1 (B) is the constant offset between the arbitrary assigned instrumental magnitudes
and the corresponding standard magnitudes for the standard B (blue) wavelengths;
x1 (R) is the constant offset between the arbitrary assigned instrumental magnitudes
and the corresponding standard magnitudes for the standard R (red) wavelengths;
x2 (B) is the blue color term introduced by the used optical system, which accounts
for its particular response at standard B (blue) wavelengths;
x2 (R) is the red color term introduced by the used optical system, which accounts
for its particular response at standard R (red) wavelengths;
x3 (B) is the blue airmass term, which accounts for the particular attenuation of the
atmosphere regarding standard B wavelengths;
x3 (R) is the red airmass term, which accounts for the particular attenuation of the
atmosphere regarding standard R wavelengths;
X(B) is the airmass at the moment of the obtained image through the standard B
filter;
X(R) is the airmass at the moment of the obtained image through the standard R
filter.

The usual way to resolve this system and find out the unknown terms is by also taking
images of reference stars lying as close as possible to the stars going to be studied.
Reference stars means stars of known apparent magnitude values accepted as standards by
the astronomical community. Those reference stars has been catalogued from high
precision photometry works, such as those performed by Arlo Landolt or Arne Henden.

By far, the most used reference catalog is the one developed by Landolt, which not only
employed the Johnson-Cousins colour system filters, but has become the “practical
definition” for the UBVRI standard system, which waveband characteristics is shown in
Figure 4. The Landolt fields, as have been generically called, are a series of star regions
near the celestial equator with well known magnitudes in at least the B, V and often R
bands [8].

Therefore, knowing mB , mR , B, R, X(B), X(R), and assuming also known both x3 (B) and
x 3(R)3, the system (3) and (4) becomes reduced to just four unknown terms: x1 (B), x 2(B),
x 1(R) and x2 (R). Having at least four sets of values from measured standard stars (two
through the blue filter and two through the red one), the system can be resolved. For
instance, equations (3) and (4) can be rewritten as

mB - B - 0.263*X(B) = x2 (B) * (B – R) + x 1(B) (5)

mR - R - 0.159*X(R) = x2 (R) * (B – R) + x 1(R) (6)

For each measured standard star at B band, the known left side of the equation (5) can be
plotted against its corresponding (B – R) value. Therefore, for just the two minimum
required data from the blue filter, it becomes possible to graphically obtain the unknown
3 They both are quite constant, being x3(B) =0.263 and x3(R) = 0.159.

Page 8
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

terms x2 (B) and x1 (B), which respectively happen to be the angular coefficient and the
intercept value in the vertical axis of the line that joins them. Proceeding in a similar way
with equation (6) it can be found out the terms x2(R) and x1 (R).

Figure 4
The UBVRI standard system bands

The final accuracy of the graphically obtained terms highly depends on the quality
(measurement precision) of the measurements of mB and mR from the reference stars. It
becomes quite obvious that by using as many as possible measurements from different
reference stars instead of just the minimum required pair, term errors will be dramatically
reduced.

Hence, the usual way is to find the correction terms is by making a spread plot including
all possible data, and then use the least square linear solution. Their final accuracy can be
estimated by computing the standard deviation and the median of the differences between
reference (Landolt) values and the resulting values derived from just their application to
instrumental magnitudes, that is those values obtained from:

m B − m R − [x1 (B ) − x1 (R )] − [ x3 (B ) * X (B ) − x3 ( R ) * X ( R )]
B−R= (7)
1 + x 2 ( B ) − x 2 (R )

R = m R − x 1 ( R ) − x 2 ( R ) * ( B − R ) − x 3 ( R ) * X (R ) (8)

In particular, the standard deviations of the difference between the Landolt magnitudes and
those derived from (7) and (8) will give the errors of the measurement.

Once all the “x” terms are known, and having been checked their correct accuracy, the
instrumental magnitude of any star of the remaining science images can be properly
converted into a standard magnitude by applying equations (7) and (8). Usually, the
typical error in the conversion will be greater for faint objects than for bright ones.

Page 9
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

4) The practical work

All the needed raw images for this project, including both sky and calibration frames, were
supplied via Internet. Appendix 1 shows the complete list of provided images, all of them
obtained from the McDonald Observatory in Texas, USA, and supplied without any
further improvement process.

After all the images were uncompressed, they were ready to be processed by means of the
application of the “AIP4WIN”4 software. All images were 2048 x 2048 pixels.

The first performed task was to cut overscan region off of all images, by applying the
corresponding crop tool. All the images were cropped to 2045 x 2048 pixels.

Next, the image reduction process was executed according to:

1) Bias correction: The five supplied bias frames were averaged (the usual
recommended process when the images to be combined are less than 7), resulting
the master bias. This master bias was then immediately subtracted to each one of
the rest of images (science and calibration), which became bias corrected.

2) Master flats: Each set of the five flat field images corresponding to the same filter
were separately averaged, thus finally obtaining three independent master flats.

3) Final reduction: Each science bias corrected image was divided by the master flat
of the corresponding filter.

Appendix 2 shows the master bias; Appendix 3, the master flat B; and Appendix 4, the
final processed M67 R image.

Now all the science images were ready to aperture photometry. This application began by
the supplied Landolt field images in order to find out the term corrections. All the
photometry measurement was performed by applying “AIP4WIN” software.

First of all, it was necessary to identify each reference star by comparing the images to the
corresponding Landolt reference celestial charts obtained from Internet [9]. As all the
science images had the left and right sides mirrored compared to the Landolt reference
celestial charts, it was necessary to perform a flop transformation in order to facilitate the
identification of each reference star.

Next, after having been analyzed many photometric curves of growth for those particular
science images, and also considered their overall stellar concentration, it was selected an
aperture of 6.0 radius, and a sky annulus of 9.0 and 15.0 radius for its inner and outer
dimensions. Arbitrary, the zero point constant was chosen as 20.0.

The instrumental magnitudes of the reference stars were then obtained. An excel
worksheet (attached to this report) was constructed in order to gather all the information.
The Data Provided page includes all the data corresponding to the reference field images.
4 “Astronomical Image Processing for Windows” is a program specifically designed for the case.

Page 10
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Each row represents each reference star, and the columns contain the star identification
(column A), the used filter and its corresponding imaging time and airmass (columns B, C
and D for the blue filter), the obtained instrumental magnitude (column E for the blue
filter), the reference Landolt values (column S for the blue filter, column V for the colour
index BR), the left side of equation (5) (column AA), the computation of BR from equation
(7) (column AP), and the difference between the Landolt BR value and the obtained from
equation (7) (column AT). Likewise, there are similar columns for data corresponding to
the red and infrared filters.

With those values it was possible to resolve graphically equation (5) by plotting together
columns V and AA, and equation (6) from columns V and AE. Some data (those appearing
with blank row values) was specifically eliminated due to showing great relative
discrepancies with the rest of the sample (that is, they laid notoriously apart than the
overall tendency of the rest of data). Several reasons could explain such deviations
(overexposure of the relative brighter stars, measurements performed to stars lying too
close one from another, irregularities in the field images such as bad columns, etc).

Finally, the unknown terms became:

x1(B) = -3.1098 and x 2(B) = -0.0934 (9)

x1(R) = -1.873 and x 2(R) = 0.0907 (10)

Applying those correction terms to recalculate the standard magnitude values [equations
(7) and (8)], resulted in median errors of –0.010 and 0.011, and standard deviations of
0.082 and 0.072, respectively for the colour index BR and the standard magnitude R.
Being the medians lower than ± 0.08, and standard deviations lower than 0.2, the values
for the correction terms proved to be satisfactory.

Next, it was done the same aperture photometry measurement to the stars of M67
contained inside a square area of 18 arcmin by side, centered in RA = 08h 51m 24.10s,
DEC = +11° 48’ 48’’ (J 2000.0), for both the B and R images. This area was arbitrarily
chosen because it seemed to contain at least all the core of the cluster. The yellow square
shown in Figure 5 represents such measured area.

The M67 B negative image was printed in order to allow the identification of each one of
the stars inside the selected area. It was found 492 stars, many of them appearing very
close from other stars, and a few even partially overlapping themselves.

The 492 stars from the M67 B image were attempted to be measured in the same aperture
photometry conditions as before, followed by the same operation for the M67 R image.
Many of the stars resulted unable to be measured, and even a few faint stars were simply
“not found” in the M67 R image (probably due to introduced noise). Finally, 457 stars
were actually measured from both B and R images.

Applying the same correction terms (9) and (10) to equations (7) and (8) it was obtained
the corresponding M67 apparent magnitudes on a standard system. However, the resulting
colour index BR values were not satisfactory, as all of them seemed to have been shifted to
higher values. The origin of this problem is still unknown. A possible explanation would

Page 11
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

be that the provided field images were not obtained by using the same equipment as the
M67 images.

W E

Figure 5
The measured area at the core of M67

The solution to overcome this difficult this was to obtain new correction terms from the
very same M67 images, by applying the same procedure as before to a sample of stars of
well known standard magnitudes. Luckily, it happens to be that there are several reference
stars measured from M67 by Arne Henden, although its quantity -just 20 stars- is quite
limited [10].

After having identified those 20 stars, their instrumental magnitudes was then obtained and
collected in the spreadsheet at the Observed page, as well as all the corresponding data, in
the same way as the Data Provided page. Data from the three brighter stars was finally not
taken into account (they present noticeable deviations). From the remaining 17 reference
stars the correction terms resulted:

x1(B) = -2.0976 and x 2(B) = -1.1819 (11)

x1(R) = -1.8287 and x 2(R) = -0.7035 (12)

As before, they were checked by computing medians and standard deviations (respectively
0.00 and 0.12 for the colour index BR, and –0.03 and 0.15 for the R standard magnitude).

Page 12
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

The complete information for the surveyed stars of M67 (including data, measurements,
and calculations performed with the latter correction terms) was collected in the C-M
Diagram page of the same spreadsheet.

5) The obtained M67 CMD

From a spread plot of columns M and N of the C-M Diagram page, it was finally obtained
the following M67 CMD:

B-R
-1.0 0.0 1.0 2.0 3.0 4.0
8

10

12
R magnitude

14

16

18

20

Figure 6
The obtained M67 colour-magnitude diagram
The diagram certainly had an overall shape as expected, but also appeared a lot of points
that notoriously were “out of correspondence”, basically those laying lower than the 16th
magnitude, and also those being below and to the left of the Main Sequence. The obvious
explanation was that an important percentage of included data actually do not correspond
to M67 stars.

Page 13
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Inside the measured area (324 arcmin 2) centered in M67 it was counted 492 stars, thus
giving an image star density of 1.52 star/arcmin 2 . As Figure 7 shows, the star counts for
the same area at each corner of the M67 image gave 192, 261, 274, and 219 stars, although
the two counts at the right side still seemed to be “contaminated” by the near presence of
M67. For this reason, it can be considered a value of 206 stars (the average of the two left
counts) as the image star average count for a 324 arcmin 2 area in the nearby region to M67
(giving an image star density of 0.63 star/arcmin 2 ).

Statistically arguing, this calculation implies that about 200 stars that appeared inside the
yellow square actually do not belong to M67, being either foreground or background stars.
As those “interloper” 200 stars (40 % of the sample) do not lay at the same distance as the
truly cluster stars, their locations in the M67 CMD became spread following unknown
patterns (although basically displaced in the vertical orientation, each shift actually
depending on the true distance of the considered star).

Figure 7
The star count for the same area in five different sectors

Another source of data error was possible due to the fact that the core of M67 is a very
crowded area, which implied greater inaccuracies in the performed aperture photometry.
(specially regarding that the applied “AIP4WIN” may not be the best tool for the job).

Page 14
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Having done the previous observations, the obtained M67 CMD was “cleaned” up to its
most fundamental form, shown in Figure 8. In this basic sketch clearly appeared the Main
Sequence, the Red Giant Branch, and the “golden” final objective: the turn-off point
location (roughly estimated on standard magnitude R = 13, colour index BR = 0.9).

Figure 8
A schematic simplification of the M67 CMD of Figure 6

Now the road was all clear to determine M67’s age and distance.

6) Deriving the age and distance to M67

The short length of Main Sequence in the M67 CMD announced that this open cluster is
very old indeed. The age estimation came derived from the turn-off point location.

The sketched M67 CMD was overlapped to a similar Main Sequence CMD but plotted in
absolute magnitudes 5 , both in the same scale (each vertical increment corresponded to the
same magnitude variation for both graphs) as shown in Figure 9.

As mentioned at the theoretical description of this procedure, a well done fitting would
have implied that not only the short M67 Main Sequence perfectly matched the long
general Main Sequence plot, but also for the same corresponding colour indexes. As this

5
It was applied the CMD shown at the right side of Figure 3.

Page 15
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

double condition turned out impossible to achieve simultaneously, the Main Sequence
coincidence was priority considered.

Such found impossibility suggested some problem with the obtained colour index values
(perhaps some offset was unnoticed introduced). The fact that the M67 CMD was
expressed in colour index BR while the Main Sequence CMD was in colour index BV o
certainly introduced some error, but not enough to justify such difference. The BV o scale
had to be expanded up to a factor of 1.7 in order to allow the proper fitting of the Main
Sequence portions.

Figure 9
The sketched M67 CMD overlapped to the sketched MS CMD shown in Figure 3

Finally, the best fitting for the Main Sequence overall shape gave for the M67 turn-off
point an absolute magnitude of 3.5.

This 3.5 result, applied to the Van den Berg's isochrones shown in Figure 2, converted it to
an estimated age of 4,000 years for M67.

There are several intrinsic inaccuracies associated to this found age. First of all, the
difficult in determining the “exact” turn-off point; secondly, the achievement of the proper
magnitude values for it; and finally, the translation -via stellar isochrones- to the final age.

Page 16
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Regarding the contaminated M67 CMD, the performed simplification to place the turn-off
point, and the problem with the color index, it could be conservatively estimated an overall
error not less than 50 %.

The age of M67 is currently known with relative great uncertainty (its published values
varying from 3.2 billion years [11] up to 9.4 billion years [12]) 6. Hence, it could be said
that the obtained age was acceptable.

As the best fitting between the Main Sequence diagram and the Main Sequence portion of
the sketched M67 CMD had been achieved for a vertical displacement that leveled
apparent magnitude values to absolute magnitude values plus a 9.5 offset, this number
became M67’s distance modulus, and the distance was obtained from (2)

 13. 0−3.5+5 
 
= 794 parsecs = 2,590 light years
d = 10  5 

The error in the found distance was basically associated to the difficult in achieving a
correct fitting of the M67 Main Sequence to the general CMD.

Figure 10
A M67 CMDs comparison between the obtained and a published one

The best current M67’s distance modulus is 9.7 (Dinescu, 1995) [13], so the found value is
amazingly good (difference 2%), considering that (a) it has been derived by a comparative
method which implies difficulties in finding the best fit; (b) the M67 CMD has been

6
This explains the eagerness of Dr. John Joseph Bochanski Jr. for using the Hubble Space Telescope to
construct a new M67 CMD [3].

Page 17
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

obtained from a highly contaminated sample of interloper stars; and (c) the M67 CMD
plots apparent magnitudes at R band, instead of apparent magnitudes at V band.

A head to head comparison between the obtained M67 CMD and a published M67 CMD is
shown in Figure 10. They have been respectively placed in order to have the same vertical
scale (although the first is for R magnitudes, and the latter for V magnitudes). The
comparison shows the correctness of the found Main Sequence shape and turn-off point
magnitude, although the index colour, as said, does not coincide (it respectively
corresponds to BR = 0.9 and BV = 0.6).

The obtained Main Sequence appears notoriously broadened, while the Red Giant Branch
almost coincide up to magnitude 11. The left CMD gets very fuzzy down the 16th
magnitude, and this is due to the already mentioned interloper “contamination”. Some stars
that seem to “continue” the Main Sequence (up and to the left) may be blue-straggler,
although this area must be necessarily so contaminated as the rest.

More accurate results would have been obtained if all surveyed stars were positively
members of the cluster. This condition would allow a “clear” CMD, and hence a lesser
uncertainty in the correct determination of the Main Sequence shape and turn-off location.
The practical way to perform a “clear” cluster CMD would be to identify each star inside
the field and delete all data corresponding to foreground or background stars.

Conclusion
Star clusters are the key to enter into the world of stellar evolution. The membership
condition that all those stars share -being born at the same time from the same origin, and
still remaining together, that is, laying at the same distance from us-, make them the place
to look for development, confirmation or rejection about any theoretical conception about
stellar evolution.

CMDs are the proper tool to analyze data from clusters. They achieve, at the same time,
two fundamental properties: (1) they are quite simple to construct, just derived from
photometric data; (2) they can supply a lot of important information.

From CCD images it is possible to obtain accurate photometric data. But such CCD raw
information must be precisely processed in order to really become quality data. This
implies not only to correct all possible introduced noise, but to make data independent of
measurement particularities from both equipment and atmosphere as well.

As a practical exercise, it was constructed the CMD for the cluster M67 from provided
images via Internet. The final diagram allowed the estimation of both its age and distance,
achieving satisfactory results. This confirmed the tremendous potentially of cluster CMDs,
becoming a high recommendable investment considering time and resources consumption
versus the essence of the deduced information.

Page 18
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

References

[1] “Measuring a Globular Star Cluster’s Distance and Age” by ESA, The ESA/ESO
Astronomy Exercise Series 4, 2001, page 6
[2] “Universe: Stars and Galaxies” by Roger A. Freedman and William J. Kaufmann III,
R.R. Donnelley et Sons Company, 6 ed, 2002, page 481
[3] “Determining the Age of the Old Open Cluster M67” by Dr. John Joseph Bochanski
Jr., page 1, at http://www.astro.washington.edu/stinson/HST/boo.pdf
[4] “Use of CCD Cameras” by SAO, CD-ROM HET 609 (Module 8, Activity 2), 2004
[5] “The Handbook of Astronomical Image Processing” by Richard Berry & James
Burnell, Willmann-Bell, Inc., 2000, pages 257-259
[6] “Handbook of CCD Astronomy” by Steve B. Howell, Cambridge University Press,
2000, page 89
[7] “Standard Calibrations” by Pamela Gay, page 9, at
http://stryder.fas.harvard.edu/ExcerptChptr2.pdf
[8] “A Practical Guide to Lightcurve Photometry and Analysis” by Brian D. Warner, Bdw
Publishing, 2003, page 23
[9] “UBVRI photometric standard stars in the magnitude range 11.5-16.0 around the
celestial equator” by Arlo U. Landolt, Astronomical Journal 104, 1992, pages 340-371
[10] “Arne Henden M67 file”, at ftp://ftp.nofs.navy.mil/pub/outgoing/aah/m67
[11] “Companion - SkyAtlas 2000.0” by Robert A Strong and Roger W Sinnot, 2ed,
Cambridge University Press, 2000, page 54
[12] “WEBDA’s Cluster page for M67” by Jean-Claude Mermilloid, at
http://obswww.unige.ch/webda/cgi-bin/ocl_page.cgi?dirname=ngc2682
[13] “Determining the Age of the Old Open Cluster M67” by Dr. John Joseph Bochanski
Jr., page 3, at http://www.astro.washington.edu/stinson/HST/boo.pdf

All images for this project have been provided from:

http://astronomy.swin.edu.au/sao/students/projects/PamelaGay/

Figures:
1 - “Star clusters” at http://zebu.uoregon.edu/~js/ast122/lectures/lec12.html
2 - http://orca.phys.uvic.ca/astrocourses/a120/A120/lab5.html
3 - left: http://orca.phys.uvic.ca/astrocourses/a120/A120/lab5.html ; right: “Ages and
Distances of Clusters” at http://www.physics.sfasu.edu/markworth/ast105/09agdist.htm
4 - http://orca.phys.uvic.ca/astrocourses/a120/A120/lab5.html
5 - by the author
6 - by the author
7 - by the author
8 - by the author
9 - by the author
10 - left: by the author; right: http://orca.phys.uvic.ca/astrocourses/a120/A120/lab5.html

Page 19
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Appendix 1

The provided images from Internet

Image code Image Date and UT time Airmass

Swin81_0001 B flat 2001-04-26 08:32
Swin81_0002 B flat 2001-04-26 08:33
Swin81_0003 B flat 2001-04-26 08:34
Swin81_0004 B flat 2001-04-26 08:35
Swin81_0005 B flat 2001-04-26 08:35
Swin81_0006 R flat 2001-04-26 08:44
Swin81_0007 R flat 2001-04-26 08:45
Swin81_0008 R flat 2001-04-26 08:46
Swin81_0009 R flat 2001-04-26 08:47
Swin81_0010 R flat 2001-04-26 08:47
Swin81_0011 I flat 2001-04-26 08:56
Swin81_0012 I flat 2001-04-26 08:57
Swin81_0013 I flat 2001-04-26 08:58
Swin81_0014 I flat 2001-04-26 08:59
Swin81_0015 I flat 2001-04-26 09:00
Swin81_0022 SA 104 R 2001-04-24 03:04 1.087
Swin81_0023 SA 104 B 2001-04-24 03:09 1.083
Swin81_0024 SA 104 I 2001-04-24 03:14 1.078
Swin81_0025 M67 R 2001-04-24 03:21 1.073
Swin81_0026 M67 B 2001-04-24 03:27 1.071
Swin81_0027 M67 I 2001-04-24 03:32 1.069
Swin81_0028 NGC 4147 R 2001-04-24 03:38 1.075
Swin81_0029 NGC 4147 B 2001-04-24 03:44
Swin81_0030 NGC 4147 B 2001-04-24 03:49 1.058
Swin81_0031 NGC 4147 I 2001-04-24 03:58 1.051
Swin81_0032 SA 104 R 2001-04-24 04:11 1.089
Swin81_0033 SA 104 B 2001-04-24 04:16 1.095
Swin81_0034 SA 104 I 2001-04-24 04:22 1.106
Swin81_0035 NGC 4147 R 2001-04-24 05:44 1.042
Swin81_0036 NGC 4147 B 2001-04-24 05:48 1.047
Swin81_0037 NGC 4147 I 2001-04-24 05:56 1.055
Swin81_0038 SA 107 R 2001-04-24 06:06 1.417
Swin81_0039 SA 107 B 2001-04-24 06:11 1.396
Swin81_0040 SA 107 I 2001-04-24 06:16 1.375
Swin81_0041 SA 107 R 2001-04-24 06:53 1.266
Swin81_0042 SA 107 B 2001-04-24 06:57 1.255
Swin81_0043 SA 107 I 2001-04-24 07:03 1.244
Swin81_0047 SA 107 R 2001-04-24 08:26 1.166
Swin81_0048 SA 107 B 2001-04-24 08:30 1.166
Swin81_0049 SA 107 I 2001-04-24 08:36 1.167
Swin81_0056 Bias
Swin81_0057 Bias
Swin81_0058 Bias
Swin81_0059 Bias
Swin81_0060 Bias

Page 20
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Appendix 2

The master bias image

Page 21
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Appendix 3

The master flat B image

Page 22
Project: Cluster Colour-Magnitude Diagrams (Supplied Data) - Course: HET 609, June 2004
Supervisor: Dr. Pamela Gay - Student: Eduardo Manuel Alvarez

Appendix 4

The M67 R final processed image

Page 23