Finding the peaks isn’t difficult, as they are
largely where we expect
Measuring the slight differences is far more
tricky
Gaussians
Measure peak placement precisely involves fitting a model to the
peak
One of the most common is that of a Gaussian: \[ g(x) = H +
A\exp\left(\frac{-(x-x_0)^2}{2\sigma^2}\right) \] where
\(H\) is the offset or shift of the
baseline
\(A\) is the peak height above the
baseline (can be negative)
\(x_0\) is the \(x\) at which the peak is centered
\(\sigma\) is the spread of the
peak
The Continuum
Peaks often show up on areas of the blackbody spectrum that are
heavily sloped
Fitting them well requires “flattening”, or normalizing out this
background continuum
Most common approach is to fit a line or polynomial to just the
background (not the peak!) and then divide the entire background by this
signal
If done well, you should get a clean peak sitting on a level
surface, ready for fitting
For velocities or identifying chemical composition, determining
\(x_0\) is the most important, as that
is the wavelength the peak is centered at
Example Using Solar Spectra
To illustrate this approach, let’s investigate the \(H_\alpha\) line in the solar spectra given
here
Want to:
Establish there is a peak approximately where we expect it
Normalize the background
Fit a gaussian to determine peak location
Velocity Computing
You can technically compute a velocity for each peak
individually
The spectrum will generally have many peaks
Don’t just average them! Fit a line to them: \[\lambda_{obs} = \left(\frac{V}{c} +
1\right)\lambda_{rest} \]
Luminosity
So what can we observe?
Light from the stars tells us:
Their location in the sky
Their overall brightness
Their intensity at different wavelengths (their spectrum)
From these, we can determine:
Surface temperature
Radial motion
Distance (sometimes)
Size (in a fashion)
Power output or Luminosity
Mass (sometimes)
Luminosity
We measure the apparent brightness \(B\) of an object here at Earth (area under
spectra)
Like ripples around a dropped rock though, brightness falls off with
distance:
Unlike pond ripples, the waves spread out radially, so the energy
gets spread over a sphere
Thus the luminosity is: \[ L = 4\pi d^2
\times B \] where \(d\) is the
distance
The range of possible stellar luminosities is huge
\(L_{sun} = L_\odot = 4 \times
10^{26}\) W
Dimmest at around \(0.000001L_\odot\)
Brightest around \(100000L_\odot\)
Energy at the source is spread over ever
larger spheres
Accounting for Distance
Stellar Distances
Initially, coming from parallax
Shifts of the foreground relative to the background when the
viewpoint changes
Parallax effects are larger for closer objects, and stars are
far away
Need as large a baseline as possible: observing during 6 month
intervals to be on opposite sides of the Sun
Parallax effects from stars are still tiny:
generally less than an arcsecond
A parsec is the distance that corresponds to a parallax
angle of 1 arcsecond
Equivalent to 3.26 light-years, or 3.26\(\times\) the distance light travels in a
year
Measuring the parallax angle \(p\)
in arcseconds gives the distance in parsecs \(d\): \[ d_{pc} =
\frac{1}{p_{asec}} \]
Absolute Magnitude
Astronomers will also use absolute magnitude as a proxy for
luminosity
A star’s absolute magnitude (commonly denoted \(M\)) is the magnitude it would seem to have
if it was 10 parsecs away
Still requires knowing the distance to the star to compute: \[ m - M = 5\log_{10}\left(\frac{d_{pc}}{10}\right)
\] where \[ \begin{aligned}
M &= \text{ absolute magnitude} \\
m &= \text{ apparent magnitude} \\
d_{pc} &= \text{ distance in pc}\\
\end{aligned} \]
Classifying Stars
Star Types
Stars were originally classified by the strength of their Hydrogen
lines
The strongest were classified type A, all the way down to type O,
which showed virtually no hydrogen lines
Original Star Types
Scrambling the System
As more spectra were observed, the H lines were proving to be less
reliable in predicting similar properties
Enter Annie Cannon
Hired as one of the Harvard Computers
Classified some 350,000 stars (yikes!)
Drastically simplied the system and eliminated many classes,
focusing mainly on the Balmer line transitions
Once the relationship between spectra lines and temperature was
understood, the letters were reordered to match the temperature
trend
HR Diagrams
HR Diagram
HR Trends
Star Sizes
Given certain names, you can perhaps guess how stellar size varies
in an HR diagram
But why?
Recall that total brightness over some interval of wavelength is
measured in watts per square meter
This would be the area under a spectra curve
This is why brightness drops off as it travels away from the star to
us
This also means though that the total energy emitted from the
surface of the star will depend on the star’s size!
The area under the curve depends (heavily) on the temperature \[ L = 4\pi R^2_s \times \sigma T^4 \]
Size Trends
The total energy output of the star thus depends on both its size
(radius) and its temperature
Cooler stars need to be much larger to have the same luminosity
output!
Hot stars can be smaller
HR Diagram Size Dependence
Mass and HR Diagrams
What about patterns in the mass of stars on the HR
diagram?
Globally, there is no obvious trend
There do appear to be trends within the subgroups
though:
Main sequence stars decrease in mass from upper
left to lower right
White dwarfs are generally fairly low in mass
Giants and supergiants can vary wildly
Mass determines many of the equilibrium points in
stars, so no clear trend is interesting!