Chapter 9

Doppler and Zeeman Doppler Imaging of Stars

Oleg Kochukhov

Abstract In this chapter we discuss the problem of reconstructing two-dimensional

stellar surface maps from the variability of intensity and/or polarisation profiles
of spectral lines. We start by outlining the main principles of the scalar Doppler
imaging problem concerned with recovering maps of chemical spots, temperature
or brightness from the intensity spectra. After presenting the physical and mathe-
matical foundations of this remote sensing method, we review its applications to
mapping different types of spots in early-type chemically peculiar and late-type
active stars, and non-radial pulsations in early-type stars. We also discuss an exten-
sion of Doppler imaging to the problem of recovering vector distributions of stellar
magnetic fields from spectropolarimetric observations and review applications of
this Zeeman Doppler imaging technique in the context of stellar magnetism studies.

9.1 Introduction

Stars exhibit different types of inhomogeneities on their surfaces. In many cases,

including the presence of cool spots on the solar surface, magnetic fields are
responsible for this structure formation. In other situations lateral inhomogeneities
may be related to non-radial pulsations or surface convection. Investigation of the
formation, evolution and mutual interaction of different stellar surface structures
represents an essential part of stellar physics, which has profound consequences
for understanding the stellar evolution in general and the phenomena of mass loss,
angular momentum evolution, planet formation and habitability of exoplanets in
particular. Detailed information about geometrical distribution of stellar surface
inhomogeneities and its temporal evolution comprises a critical input required for
developing realistic theoretical models of these phenomena. For instance, historical
observations of sunspot cycles and more recent discovery of the link between cool
spots and magnetic fields was a prerequisite for understanding the solar activity in
the context of a dynamo theory.

O. Kochukhov ()
Department of Physics and Astronomy, Uppsala University, Box 516, SE 75120 Uppsala, Sweden
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 177

J.-P. Rozelot, C. Neiner (eds.), Cartography of the Sun and the Stars,
Lecture Notes in Physics 914, DOI 10.1007/978-3-319-24151-7_9
178 O. Kochukhov

For the Sun a bewildering complexity of surface structures is accessible to direct

imaging with many ground-based instruments and space missions. But, with a few
exceptions of interferometric studies of nearby and intrinsically large stars (see
chapters by P. Kervella and J. Monnier), the disks of stars other than the Sun cannot
be resolved and hence cannot be studied using direct imaging. In that case the only
viable option to obtain information about the stellar surface structures is to apply
some form of an inverse remote sensing method of indirect imaging, capable of
recovering a stellar surface map from spatially unresolved stellar observations.
In this chapter we will discuss two powerful remote sensing techniques: Doppler
imaging (DI) and Zeeman Doppler imaging (ZDI). The first method uses time
series observations of the intensity line profiles recorded at high spectroscopic
resolution to reconstruct scalar star spot maps. The second technique utilises circular
polarisation signatures inside line profiles or, more generally, high-resolution
spectropolarimetric observations in all four Stokes parameters to recover topology
of vector magnetic field at the stellar surface. In the following sections we outline
key physical principles and mathematical methods used by DI and ZDI techniques.
This discussion is supplemented with a brief overview of the applications of indirect
imaging to different types of surface structures in both hot and cool stars.

9.2 Doppler Imaging with Intensity Spectra

9.2.1 Main Principles of DI

An inhomogeneity on the stellar surface changes the local emergent line and
continuum radiation. For example, a local enhancement of the concentration of
some chemical element increases intensity of its absorption lines; the presence of a
cool dark spot changes the local spectrum and suppresses the continuum radiation
relative to immaculate photosphere. The flux spectrum of an unresolved stellar disk
represents a weighted average of all local spectral contributions, Doppler-shifted
according to the local projected rotational velocity relative to the observer. If the
surface inhomogeneities are sufficiently large and have a high contrast, their spectral
signatures will be visible in the disk-integrated line profiles as distortions—either
“emission” bumps for cool spots or an extra absorption features for spots with an
enhanced element concentration (see Fig. 9.1a, b)—superimposed onto the regular
Doppler-broadened line profile. The velocity of the distortion relative to the line
centre is given by the longitude of the surface feature, reckoned from the central
stellar meridian.
The latitude position of a star spot cannot be inferred from a single observation.
Instead, one uses time-series behaviour of the spot signatures to recover the latitude
information. As the star rotates, spectral distortions corresponding to each spot first
appear on the blue side of the spectral line profile and then gradually move to the
red side. As illustrated by the dynamic spectra in Fig. 9.1c, temporal behaviour of
9 Doppler and Zeeman Doppler Imaging of Stars 179

Fig. 9.1 Illustration of the main principles of Doppler imaging. The local stellar spectrum formed
in a spot differs from the photospheric spectrum. This leads to a distortion in the disk-integrated
stellar line profile. This distortion is Doppler-shifted according to the position of the spot relative to
the disk centre. The two spherical plots and corresponding line profiles illustrate spectral signatures
of (a) a single spot with reduced continuum brightness typical of late-type active stars and (b) a
spot with enhanced line strength typical of early-type chemically peculiar stars. As the star rotates,
the spot signature moves across the line profile from blue to red. The rectangular panel (c) shows
the dynamic difference spectrum as a function of the rotational phase. In this case, the stellar
surface has four small spots at latitudes 30; 0; C30, and C60 degrees. This plot demonstrates
how temporal variation of the spot signatures depends on their latitude position

the spot signatures depends significantly on their latitudes. For the intermediate
inclination of the stellar rotational axis (i D 50o ) shown in this figure, the line
profile distortions produced by the spots with latitudes from 0o to  C30o span
the largest range of relative velocities and are visible roughly half of the rotational
period. On the other hand, signatures of structures at higher latitudes move over a
smaller velocity range and are visible over a larger fraction of the rotational cycle.
Spots with a latitude above i remain visible all the time; their line profile signatures
exhibit faint “backward” red-to-blue progression during some part of the rotational
180 O. Kochukhov

cycle. Spots below the equator are visible very briefly and leave a weak signature in
the disk-integrated line profile.
To summarise, the amplitude and position of stellar surface structures are
encoded in the variability of distortions observed in Doppler-broadened line profiles.
Given high-quality observations of stellar spectra, obtained with a sufficiently dense
rotational phase coverage and a signal-to-noise (S/N) ratio high enough to clearly
detect line profile distortions due to spots, it should be possible to reconstruct a two-
dimensional map of the stellar surface. This is the scope of the inversion technique
known as Doppler imaging.

9.2.2 Spatial Resolution of DI

A classical method of estimating resolution at the stellar surface provided by DI is to

compare the rotational Doppler broadening with the stellar line width in the absence
of rotation. The latter is usually dominated by the instrumental broadening. Then,
the angular size of the resolution element at the stellar equator is given in degrees
by

 vc
ı` D 90o ; (9.1)
 ve sin i

where R D = is the instrumental resolution of the spectrograph, ve sin i is

the projected rotational velocity, and vc is the speed of light. For example, for
R D 65 000 and ve sin i D 150 km s1 we obtain ı` D 2:7o , which corresponds
to 133 resolution element along the equator. The distance to the star is not directly
entering Eq. (9.1); but of course the star has to be sufficiently bright to allow
obtaining high S/N ratio spectra within a time interval small compared to the
rotational period. Considering, for instance, a main sequence early B-type star at
1 kpc (with V magnitude  6 it is easily accessible to high-resolution spectroscopy
at 2–4 m class telescopes), we can infer that ı` D 2:7o corresponds to an angular
resolution of 0.3 arcsec. This is orders of magnitude better than any direct imaging
technique can provide at the moment or in the foreseeable future.
A common misconception is to use Eq. (9.1) for estimating the lower ve sin i
limit of DI, which leads to an argument that DI requires ve sin i  vc =R. On the one
hand, it is certainly true that the spatial resolution of this remote sensing method
gradually decreases as the Doppler broadening becomes comparable to the intrinsic
line width. On the other hand, we should not forget that Eq. (9.1) estimates spatial
resolution based on a single snapshot observation. In reality, DI operates with time
series spectra, gaining significant additional resolution from rotational modulation.
The same type of information is used by the photometric mapping methods (see
the chapter by A. Lanza) to recover coarse brightness maps of stellar surfaces from
broad-band light curves. Similarly, DI and especially its extension to the polarisation
spectra—ZDI—is capable of recovering maps of large-scale surface features for
9 Doppler and Zeeman Doppler Imaging of Stars 181

extremely slowly rotating stars (e.g. Petit et al. 2008) independently of the actual
ve sin i value, provided that the input time series data exhibit a significant phase-
dependent variation.

9.2.3 DI as an Ill-Posed Inverse Problem

As discussed above, DI provides a method to relate the observed line profile

variability to the underlying geometry of stellar surface structures. However, a
key question is how to mathematically implement the inverse problem of deriving
a 2-D surface map from a given spectroscopic observational data set. Early
studies of chemically peculiar stars and late-type active stars attempted to fit
observations with parameterised maps, typically consisting of a small number of
circular spots (Mihalas 1973; Vogt and Penrod 1983b) or employing a low-order
spherical harmonic expansion (Mégessier 1975). But it was quickly realised that
the problem of finding parameters of these spots is in practice mathematically ill-
posed, meaning that an infinite number of very different solutions can fit a given
data set. A breakthrough was achieved when Goncharskii et al. (1977) suggested
to employ regularisation methods to ensure uniqueness of the DI solution. In other
words, besides observations themselves, one introduces some additional criterium
of simplicity to limit the family of possible solutions. Mathematically this is
implemented by finding a solution x that minimises the sum of 2 of the fit to
observations and regularisation function R
X
ŒSobs  Smodel .x/2 =obs
2
C R.x/ ! min; (9.2)

where regularisation parameter  is determined empirically by trial and error or is

set by requiring a certain target 2 .
In the early studies of chemical spots in Ap stars (e.g. Khokhlova et al. 1986) DI
problems were regularised with the Tikhonov functional

R.x/ D krxk: (9.3)

Later Vogt et al. (1987) introduced the maximum entropy method (MEM) in the
context of the application of DI to temperature mapping of cool active stars. In this
case,
X xi xi
R.x/ D log ; (9.4)
i
x0 x0

where x0 is some default value of the map. Subsequently both regularisation

methods were used for mapping chemical structures in Ap stars and temperature
spots in late-type stars (Piskunov et al. 1990; Hatzes 1991).
182 O. Kochukhov

The two regularisation approaches differ slightly in their interpretation of the

concept of “simplicity” of a surface map. The Tikhonov regularisation favours
solutions with the least local gradient. On the other hand, MEM prefers a map
with the least deviation from the default value. In practice, the two methods
produce compatible results when applied to the same observations of sufficiently
high quality (Strassmeier et al. 1991). However, it cannot be taken for granted
that the two regularisation schemes are always equivalent and equally applicable
to any DI problem. In particular, MEM fails whenever the surface structure does not
have a natural default value. For example, this problem is encountered in the ZDI
reconstruction of large-scale multipolar magnetic fields.
Finally, it should be added that a number of auxiliary stellar parameters are
required to perform a DI inversion. They include the stellar rotational period Prot ,
inclination of the stellar rotational axis relative to the line of sight i, the projected
rotational velocity ve sin i, and the radial velocity of the star Vr . Of these parameters,
Prot and Vr are usually known prior to a DI analysis. The projected rotational velocity
can be accurately determined with the help of DI inversions themselves by finding
a ve sin i value that yields the best fit to observations. An incorrect ve sin i produces
characteristic axisymmetric artefact features in the surface maps (Vogt et al. 1987;
Rice et al. 1989). On the other hand, DI inversions are relatively insensitive to the
choice of i, with errors of 10–15o leading to negligible distortions of the surface
map. The relation

Prot ve sin i
sin i D (9.5)
50:613R?

can be employed to constrain inclination if the stellar radius R? is known. In this

equation ve sin i is measured in km s1 , Prot in days, and R? in solar units.
In addition to all these parameters specific to DI, one has to know parameters
that determine the shape and strength of the local line profiles. Depending of the
modelling approach, these parameters can comprise either a few numbers required
to specify some analytical line profile function (e.g. a Gaussian profile) or a
complete set of stellar atmospheric parameters and chemical abundances for the
most realistic spectrum synthesis representation of stellar observations.

9.2.4 DI Applications

Chemical Mapping of Ap and HgMn Stars

Chemical mapping of the upper main sequence chemically peculiar A and B stars
was the first application of Doppler imaging (Goncharskii et al. 1983; Khokhlova
and Pavlova 1984; Khokhlova et al. 1986). These objects, comprising a small
fraction of all A and B stars, are distinguished by a host of peculiarities including
anomalously strong spectral lines of iron-peak and rare-earth elements, a strongly
9 Doppler and Zeeman Doppler Imaging of Stars 183

non-uniform (with contrasts of up to several orders of magnitude) vertical and

horizontal chemical distributions, and the presence of strong (typically 1–10 kG),
globally organised magnetic fields. These so-called magnetic Ap stars exhibit a
well-defined periodic variability of line profiles, spectral energy distribution, and
magnetic field strength. Since the seminal paper by Stibbs (1950), this behaviour
is interpreted in the context of the oblique rotator model. This phenomenological
framework attributes all types of stellar variability to the rotational modulation of
the aspect angle at which the surface spot structure and magnetic field are seen by
a distant observer. Consequently, Ap stars vary in a strictly periodic manner and do
not exhibit any intrinsic evolution of their surface structure. These properties make
them ideal DI targets.
Early abundance DI studies of Ap stars focused on mapping a small number
of chemical elements (Khokhlova et al. 1986; Ryabchikova et al. 1996; Rice and
Wehlau 1991; Hatzes 1991, 1997), trying to relate their distributions to the process
of atomic diffusion in magnetic field thought to be responsible for the formation
of horizontal chemical inhomogeneities (Michaud et al. 1981). Although some
encouraging results were obtained, particularly for Si (Alecian and Vauclair 1981),
these interpretation efforts were hampered by the lack of detailed information about
magnetic field geometries of the target stars.
More recent abundance DI studies of Ap stars provided maps of up to a dozen
chemical elements (Kochukhov et al. 2004; Lüftinger et al. 2010; Nesvacil et al.
2012), taking advantage of a wide wavelength coverage of modern echelle spectrom-
eters. Some of these maps were reconstructed simultaneously, by modelling blends
containing contributions of several elements. A typical example of such abundance
DI maps is shown in Fig. 9.2.
The comprehensive multi-element DI studies uncovered a complex and diverse
behaviour that did not agree with expectations of simple atomic diffusion models for
a star with a dipolar-like magnetic field. In particular, only a few chemical elements,
notably Li (Polosukhina et al. 1999) and O (Rice et al. 1997, 2004), showed a surface
distribution with a systematic relationship to the underlying magnetic field topology.
Other chemical elements, for example Ca, Fe, Cr, Ti, typically show little or no
such relation. There is also a great deal of diversity in the surface distributions of
the same elements in stars with very similar fundamental and magnetic parameters
and in the surface patterns of elements with similar properties (e.g. different rare-
earth elements) in the same stars. This non-systematic behaviour suggests that some
hitherto unknown structure formation mechanism contributes to shaping of chemical
maps of Ap stars.
A new domain of applications of abundance DI was opened by the discovery
of chemical spots in late-B HgMn stars (Ryabchikova et al. 1999). These objects
also belong to the group of chemically peculiar stars. But, unlike magnetic Ap
stars, HgMn stars lack detectable large- or small-scale magnetic fields (Aurière
et al. 2010; Bagnulo et al. 2012; Kochukhov et al. 2013). Despite this, several
chemical elements which are most overabundant in these stars (Hg, Y, Pt) show
subtle line profile variations indicative of low-contrast non-uniform chemical
distributions (Kochukhov et al. 2005; Hubrig et al. 2006; Folsom et al. 2010).
184 O. Kochukhov

Fig. 9.2 Typical chemical abundance distributions reconstructed with DI for a magnetic Ap star. In
this case, the maps of Li, O, Fe, and Eu are shown for the cool Ap star HD 83368. The scale bars to
the right indicate abundance values in logarithmic units log.Nel =Ntot ). The magnetic field structure
of this star is approximately dipolar, with a large magnetic obliquity (ˇ  90o ). The “C” and
“o” symbols indicate positions of the positive and negative magnetic poles, while the dashed line
corresponds to the magnetic equator. Some chemical elements have simple distributions correlating
with the dipolar magnetic field geometry. Other elements exhibit complex or simple maps, which
show no apparent relation to the underlying magnetic field structure. Adapted from Kochukhov
et al. (2004)

These inhomogeneities were mapped with DI in several HgMn stars (Adelman

et al. 2002; Briquet et al. 2010; Makaganiuk et al. 2011, 2012; Korhonen et al.
2013). Remarkably, it was discovered that chemical spots in HgMn stars evolve
slowly with time (Kochukhov et al. 2007)—something that has never been observed
in magnetic Ap stars. Figure 9.3 shows an example of such an evolution, on a
time scale of several years, for the mercury distribution in the brightest HgMn
star ˛ And. No conclusive theoretical explanation of the origin of these spots and
reasons for their slow variation has been proposed. However, there are indications
that a non-equilibrium, time-dependent atomic diffusion may play some role in these
phenomena (Alecian et al. 2011).
9 Doppler and Zeeman Doppler Imaging of Stars 185

Fig. 9.3 Slow evolution of mercury overabundance spots on the surface of non-magnetic chem-
ically peculiar star ˛ And. (a) The Hammer-Aitoff projection of the Hg DI maps reconstructed
at three different epochs spanning 7 years. The sidebar gives the local element abundance in
logarithmic units relative to the solar concentration of mercury. (b) The corresponding pairwise
difference maps. Adapted from Kochukhov et al. (2007)

Temperature Mapping of Active Late-Type Stars

In parallel with the DI studies of chemically peculiar stars significant efforts were
devoted to reconstruction of temperature inhomogeneities on the surfaces of cool
active stars. In these objects the surface activity is invariably associated with an
enhanced dynamo action amplified by a rapid stellar rotation. This makes these stars
much more active and consequently more spotted than the Sun. Several classes of
late-type active stars exhibit these characteristics: single rapidly rotating pre-main
sequence (including classical and weak-line T Tauri) and young main sequence
stars, members of close binary systems spun up by tidal interaction (RS CVn stars),
various classes of rapidly rotating giants (e.g. FK Com-type stars). The presence
of large spots on the surfaces of these stars and significant rotational broadening
of their spectral line profiles facilitate application of DI. However, unlike static
chemical maps of Ap stars, temperature distributions of active cool stars evolve with
time. This imposes a significant limitation of the observational data: a complete
rotational phase coverage has to be achieved within  10 stellar rotations, which
typically corresponds to a single observing run.
Temperature maps have been published for  100 cool stars. A catalogue and
detailed review of these studies can be found in Strassmeier (2009, 2011). A few
active stars were targeted repeatedly by DI (Vogt et al. 1999; Kovári et al. 2004;
Korhonen et al. 2007; Hackman et al. 2012) in an effort to reveal activity cycles and
compare them with predictions of dynamo theories. Despite collecting a valuable
data base of temperature maps, these investigations achieved a rather limited success
in establishing connections with dynamo theory and with well-known cyclic activity
behaviour of the Sun. This is either because the most active cool stars usually chosen
for DI do not exhibit well-defined activity cycles or because a long-term behaviour
was sampled by DI studies with too few maps to reach definite conclusions about
the nature of temporal evolution of the surface structure.
186 O. Kochukhov

Methodologically, the temperature mapping of cool stars is commonly performed

using a small number of diagnostic lines or even a single spectral feature. These
spectral lines are modelled by assigning individual temperature values to each
stellar surface element and using model atmospheres from standard grids with
Teff corresponding to these local temperatures to calculate local continuum and
line intensities. Reliability of temperature distributions, especially the spot to
photosphere temperature contrast Ts , can be improved by considering atomic lines
with different temperature sensitivity or by incorporating molecular bands in the
temperature inversion (Rice et al. 2011). Summarising studies with the most careful
determination of Ts , Berdyugina (2005) found that there exists a systematic trend
of the spot contrast with the stellar Teff . The maximum Ts  2000 K is found for
early G-type stars while the minimum contrast of only a few hundred K is seen in
active M dwarfs.
Reconstruction of the stellar surface map in terms of brightness distribution or
fractional spot coverage instead of local temperature is an alternative method of
performing cool star DI (Collier Cameron 1998). These studies often completely
neglect variation of the local line intensities with temperature, attributing all disk-
integrated line profile variability to changes of the local continuum brightness. In
other cases spectra of slowly rotating cool and hot template stars are employed
to approximate the local spot and photospheric profiles of the DI target (Unruh
et al. 1995). Due to this simplified spectrum synthesis approach results of the
brightness DI studies cannot be directly interpreted in terms of physical parameters,
such as spot temperatures. This type of DI also requires some external information
(e.g. setting the spot to photosphere brightness contrast or choosing appropriate
template stars) that can only be provided by more physically detailed studies. On the
other hand, this method is well adapted to using high S/N ratio mean line profiles
constructed by combining information from thousands of individual metal lines.
This enables a fast and very precise (but not necessarily accurate) reconstruction
of the stellar surface map, which is particularly suitable for detection of subtle
secondary effects such as differential rotation (Donati et al. 2003; Barnes et al.
2005).

Mapping of Stellar Non-radial Pulsations

Stellar non-radial pulsations (NRP) represent another well-known cause of line pro-
file variability. Pulsational perturbation on the stellar surface produces an alternating
pattern of zones receding and approaching to the observer. The resulting velocity
shifts are superimposed on the rotational Doppler shifts, producing a characteristic
periodic variability of the disk-integrated line profiles (Vogt and Penrod 1983a).
Typically, the stellar NRP pattern is parameterised with the spherical harmonic
functions and is fully described by specifying the ` and m numbers of each pulsation
mode, and a ratio of the vertical to horizontal pulsation amplitude. However, in
some particularly interesting cases the NRP geometry is significantly and non-
trivially distorted by a rapid stellar rotation or a strong magnetic field (Lee and Saio
9 Doppler and Zeeman Doppler Imaging of Stars 187

1990; Saio and Gautschy 2004). In such cases a mono-periodic pulsation cannot be
described with a single combination of ` and m values. Instead, one can attempt to
reconstruct the surface pulsation pattern by solving an ill-posed DI problem. For
example, Berdyugina et al. (2003) mimicked the line profile variation due to NRP
with temperature spots. They performed a pseudo-temperature DI for the B-type
pulsating star ! 1 Sco with a dominant sectoral mode (` D m). In this particular
case pulsations can be described as a fixed surface pattern that rotates with a period
determined by the stellar rotation period and the m value (which has to be guessed)
of the sectoral pulsation mode.
In another implementation of the NRP DI problem, Kochukhov (2004a) devel-
oped a more physically realistic description of an arbitrary pulsational velocity
perturbation in terms of two surface maps, effectively representing the surface
distribution of the pulsational amplitude and phase. This study demonstrated that
it is possible to reconstruct both maps with the help of DI methodology, provided
that the stellar rotational and pulsational periods are known. This DI technique was
applied to the rapidly oscillating Ap star HR 3831 (Kochukhov 2004b), yielding the
first DI NRP velocity map and providing a unique insight into the interplay between
p-mode pulsations and a global magnetic field.

9.3 Zeeman Doppler Imaging with Polarisation Spectra

9.3.1 Detection and Diagnostic of Stellar Magnetic Field

The presence of a magnetic field leads to splitting of the atomic energy levels due
to the Zeeman effect. Consequently, individual spectral lines corresponding to the
transitions between Zeeman-split levels separate into groups of so-called and
 components. The magnitude of this separation depends on the magnetic field
strength, magnetic sensitivity of a given spectral line (characterised by the mean
Landé factor), and the central wavelength of this line. Then, a magnetic field at the
stellar surface can be detected by two basic effects: the splitting of magnetically
sensitive lines and the presence of polarisation in Zeeman components.
Typical magnetic fields of non-degenerate stars produce Zeeman splitting which
is much smaller than the intrinsic line width. Only very strong ( 1–2 kG) magnetic
fields of slowly rotating Ap stars and active M dwarfs can be diagnosed by the
Zeeman splitting or broadening of spectral lines in high-dispersion optical spectra
(Mathys et al. 1997; Reiners and Basri 2007). At the same time, magnetic field is
normally the only cause of polarisation in spectral lines. This means that the mere
presence of a systematic line polarisation signal is a signature of magnetic field.
The full state of polarisation of stellar radiation is characterised by the four
Stokes parameters: Stokes I (total intensity), V (circular polarisation), and QU
(linear polarisation). The line profile shape in Stokes I is primarily sensitive to the
field modulus. The amplitude of the corresponding circular polarisation signal is
188 O. Kochukhov

given by the line of sight projection of the magnetic field vector. The magnitude of
the transverse field component and its orientation in the plane of the sky determines
the Stokes QU parameters. Thus, all four Stokes parameters are in principle needed
for a complete diagnostic of stellar magnetic field. However, the Zeeman effect
produces circular polarisation that is up to 10 times stronger than linear polarisation.
In addition, the local Zeeman circular polarisation profile has a simpler morphology
(S-shape, double-lobe signature) compared to the linear polarisation profiles (M
or W-shape, three-lobe signature).1 For these reasons, the vast majority of stellar
magnetism studies rely exclusively on the Stokes V observations both for the field
detection and modelling.
Even constrained to the Stokes V parameter, stellar spectropolarimetry is a
challenging task owing to low amplitudes of the typical polarisation signals. For
example, the global kG-strength magnetic fields of Ap stars produce circular
polarisation signatures with amplitudes of a few % of the unpolarised continuum.
These signatures can be observed in individual spectral lines (e.g. Silvester et al.
2012) provided that the spectra have a S/N ratio of 300–500. Much weaker magnetic
fields of active cool stars yield disk-integrated signals at the level of 103 to 104
in Stokes V. A detection of 1–10 G global fields of moderately active or inactive
stars, such as the Sun, requires polarimetric precision of 105 –106 . Obviously,
such signals are impossible to detect in individual spectral lines. Instead, efficient
multi-line polarisation methods were developed to average polarisation signatures
over all suitable spectral lines. These procedures reduce the photon noise to the
level required for the detection of extremely weak polarisation signals. The most
commonly used multi-line method of least-squares deconvolution (LSD, Donati
et al. 1997; Kochukhov et al. 2010) incorporates de-blending and recovers an
average profile shape that in the context of DI can often be treated as a single spectral
line with mean parameters.
A fundamental difference between magnetic field signatures in the intensity
and polarisation is that the Zeeman splitting in the Stokes I spectra depends on
the field modulus while the Stokes QUV profiles are highly sensitive to the field
orientation. On the one hand, this means that polarisation observations are providing
rich information about field geometry. On the other hand, any analysis of complex
magnetic topologies inevitably suffers from a cancellation of polarisation signals in
the disk-integrated spectra due to addition of polarisation signatures with opposite
signs corresponding to the regions of opposite field polarity.

1
We refer to the textbook by Landi Degl’Innocenti and Landolfi (2004) for a comprehensive review
of spectral line polarisation theory.
9 Doppler and Zeeman Doppler Imaging of Stars 189

9.3.2 Mapping of Stellar Magnetic Field Topologies

Magnetic field is a vector quantity. Therefore, to fully describe the stellar surface
magnetic field topology one has to specify three scalar two-dimensional maps of
the radial, meridional, and azimuthal magnetic field components. Simultaneous
reconstruction of these distributions from spectropolarimetric observations is a
formidable task, generally requiring data in all four Stokes parameters. Such
observations are currently available for a small sample of strongly magnetic Ap
stars (Wade et al. 2000; Silvester et al. 2012; Rusomarov et al. 2013) and for only
one bright active RS CVn star (Rosén et al. 2013).
Figure 9.4 gives an example of the Stokes IQUV profile variation for a star with a
strong dipolar magnetic field. These theoretical profiles demonstrate that magnetic
field influences the line shapes in intensity as well as gives rise to a stronger
circular and weaker linear polarisation signatures. Piskunov and Kochukhov (2002)
and Kochukhov and Piskunov (2002) showed that based on an observational data
set comprising 10–20 phase of high-quality Stokes IQUV spectra it is possible to
recover a vector magnetic field map using the same basic mathematical formulation
of the DI problem as described in Sect. 9.2.3 and applying Tikhonov regularisation
to each of the three magnetic field component maps individually. In their magnetic
inversion method Piskunov and Kochukhov (2002) modelled the Stokes parameter
profiles of individual spectral lines with realistic numerical polarised radiative
transfer calculations and implemented a self-consistent and simultaneous magnetic

Fig. 9.4 Stokes profile variation of a star with an oblique dipolar magnetic field. The spherical
plots show the surface magnetic field distribution for three different rotational phases. The
underlying colour image represents the field modulus. The vector map shows the field orientation
with different colours highlighting areas of the inward and outward directed magnetic field. The
Stokes IQUV profiles are illustrated below. The scale of the polarisation profiles is the same for
Stokes V and QU but is expanded by a factor of 1.4 relative to Stokes I. These calculations are
for 4 kG dipolar field inclined by ˇ D 50o with respect to the stellar rotational axis. The latter is
inclined by i D 60o with respect to the line of sight. The projected stellar rotational velocity is
30 km s1
190 O. Kochukhov

and chemical spot inversions. According to the numerical tests by Kochukhov and
Piskunov (2002), this general ZDI methodology is successful in recovering both
global and structured magnetic field topologies. This ZDI method is now routinely
used to study the field topologies and chemical spot distributions in magnetic Ap
stars (see Sect. 9.3.3).
However, due to lack of linear polarisation observations for cool magnetic stars,
a different restricted and less sophisticated ZDI technique is usually applied. Brown
et al. (1991) and Donati et al. (1997) developed and tested a MEM-based inversion
procedure relying on the Stokes V time series, which still aims at recovering all
three components of the stellar magnetic field distribution. Figure 9.5 gives an idea
of how one can constrain the field orientation with the help of rotational modulation
of the Stokes V signatures of magnetic spots. In this highly simplified example of
isolated circular magnetic spots, the radial field spot produces a Stokes V signature
moving across the stellar line profile from blue to red and showing a maximum
amplitude when the spot is at the disk centre (Fig. 9.5a). The meridional field of the
same strength produces a much weaker Stokes V signal that has a minimum at the
disk centre (Fig. 9.5b). On the other hand, the azimuthal field exhibits a noticeably
different Stokes V signature, which changes sign at the disk centre (Fig. 9.5c). Based
on this behaviour, one expects that the azimuthal field can be readily distinguished
from the radial and meridional fields. But the latter two magnetic components are
difficult to disentangle from each other over at least some part of the stellar surface.
Reconstruction of the meridional field is going to be the least reliable. Detailed ZDI
numerical tests by Brown et al. (1991), Kochukhov and Piskunov (2002) and Rosén
and Kochukhov (2012) confirmed this assessment.
Returning to the question of regularisation in ZDI, Brown et al. (1991) found
that the restricted Stokes V inversions produce reasonable results for isolated radial
field spots but fail for global magnetic field distributions such as a dipolar field. This
problem was attributed to the fundamental assumption of the MEM regularisation
(Piskunov and Kochukhov 2002): it requires setting a “default” value that cannot
be easily and uniquely defined for a global magnetic field geometry. In contrast,
the Tikhonov regularisation performs much better for global magnetic geometries.
More recent ZDI studies switched to using a spherical harmonic representation of
the stellar magnetic field (Donati et al. 2006; Kochukhov et al. 2014). In this case
a ZDI inversion code recovers spherical harmonic coefficients corresponding to the
poloidal and toroidal magnetic field components rather than the local magnetic field
values. This modification allows one to model both global and highly structured
magnetic topologies with an added benefit of being able to characterise the field in
detail (e.g. assess poloidal vs. toroidal or axisymmetric vs. non-axisymmetric field).
In this case the regularisation is accomplished by limiting the maximum angular
degree ` of the harmonic expansion and by adding to chi-square an additional
9 Doppler and Zeeman Doppler Imaging of Stars 191

Fig. 9.5 Signatures of magnetic spots in the circular polarisation line profiles of a rotating star.
Different rows show the Stokes V profiles corresponding to the radial (a), meridional (b), and
azimuthal (c) magnetic spots with the same field strength. In each case the star is shown at three
rotational phases, separated by 0.125 of the rotational period
192 O. Kochukhov

penalty function which discourages the inversion code to introduce high-order

harmonic modes not justified by the observational data.2
The restricted Stokes V inversions are usually applied to the LSD circular polari-
sation profiles of cool active stars using a simplified polarised line formation models.
Typically, a Gaussian local line profile together with the weak-field approximation
(Marsden et al. 2011) or the Milne-Eddington analytical solution of the polarised
radiative transfer equation (Morin et al. 2008) is used. Except for the recent series
of studies of global magnetic fields of classical T Tauri stars (Donati et al. 2010) and
a couple of exploratory analyses of other types of active stars (Carroll et al. 2012;
Kochukhov et al. 2013), these ZDI studies completely ignore the effect of cool spots
on the circular polarisation profiles. Consequently, the field strength in the vicinity
of dark features is at least significantly underestimated. In the worst-case scenario
when all magnetic field is concentrated within cool spots magnetic field distributions
inferred by this inversion methodology are largely spurious (Rosén and Kochukhov
2012).
A new self-consistent ZDI inversion method based on detailed polarised radiative
transfer modelling of the LSD profiles in two or all four Stokes parameters was
introduced by Kochukhov et al. (2014). This technique is computationally expensive
but overcomes most of the shortcomings of the restricted ZDI based on LSD profiles
and can be applied to both early- and late-type magnetic stars.

9.3.3 ZDI Applications

Magnetic Mapping of B and A Stars

Historically information about magnetic field topologies of Ap stars was obtained

by fitting the phase curves of integral magnetic observables (longitudinal magnetic
field, mean field modulus, etc.) with low-order multipolar geometries (Landstreet
and Mathys 2000; Bagnulo et al. 2002). These studies established that the majority
of early-type magnetic stars possess nearly dipolar fields, with occasional deviations
from axisymmetry or a quadrupolar contribution (Kochukhov 2006; Bailey et al.
2012). Initial applications of ZDI to high-resolution Stokes I and V observations
of Ap stars validated this picture (Kochukhov et al. 2002, 2014; Lüftinger et al.
2010). Even using detailed circular polarisation profile modelling of individual
spectral lines in the framework of the general ZDI methodology discussed above
little deviation from the basic dipolar field topology was found for A-type magnetic
stars.

2
Somewhat confusingly, such ZDI with the spherical harmonic regularisation is still commonly
called “maximum entropy” inversion although the employed regularisation is quite different from
the original MEM-based ZDI scheme described by Brown et al. (1991).
9 Doppler and Zeeman Doppler Imaging of Stars 193

A qualitatively different picture emerged when ZDI was applied, for the first
time, to the full Stokes vector observations of Ap stars obtained with the MuSiCoS
spectropolarimeter (Wade et al. 2000). In particular, Kochukhov et al. (2004) and
Kochukhov and Wade (2010) showed that satisfactory fits to the Stokes QU (linear
polarisation) observations of Ap stars 53 Cam and ˛ 2 CVn require a considerably
more complex field topologies than is apparent from the Stokes IV profiles of
these stars. Figure 9.6 shows a comparison of the ZDI results by Kochukhov and
Wade (2010) for ˛ 2 CVn obtained by considering all four Stokes parameter spectra
(Fig. 9.6a) and excluding the linear polarisation data from inversion (Fig. 9.6b).
Evidently, the full Stokes vector ZDI is able to recover the small-scale magnetic
field structures—essentially comprising several horizontal field spots—that remain
undetected by the inversions limited to Stokes IV data.
Subsequent investigation of ˛ 2 CVn using high-quality Stokes IQUV observa-
tions secured with a new generation of spectropolarimeters (ESPaDOnS and Narval,
see Silvester et al. 2014) confirmed the ZDI results of Kochukhov and Wade (2010)

Fig. 9.6 Comparison of the ZDI reconstruction of the surface magnetic field topology of the Ap
star ˛ 2 CVn from observations in all four Stokes parameters (a) and from a data set limited to
Stokes I and V (b). In both cases the rows show spherical plots of the magnetic field modulus
(upper row) and field orientation (lower row). The star is shown at five rotational phases and at
the inclination angle of i D 120o . The contours are plotted with a 0.5 kG step in the field strength
map. It is evident that the magnetic inversion limited to the circular polarisation spectra is unable
to recover the small-scale magnetic features. Adapted from Kochukhov and Wade (2010)
194 O. Kochukhov

Fig. 9.7 Long-term stability of the global and small-scale magnetic field topology of the Ap star
˛ 2 CVn. Rectangular plots show surface distributions of the radial, meridional, and azimuthal field
components. The two rows correspond to (a) magnetic field map derived by Kochukhov and Wade
(2010) from the MuSiCoS observations collected in 1997–1999 and (b) an equivalent magnetic
field distribution obtained by Silvester et al. (2012) from the ESPaDOnS and Narval four Stokes
parameter data acquired in 2006–2007. In both cases magnetic inversions were carried out using
all four Stokes parameters. Adapted from Silvester et al. (2012)

and demonstrated that no significant evolution of either the global or small-scale

magnetic field of this star has occurred during  10 years that have passed between
acquisition of the two polarisation data sets (see Fig. 9.7).
Thus, the four Stokes parameter ZDI studies of intermediate-mass Ap stars
53 Cam and ˛ 2 CVn indicated that, although the overall magnetic field structure
defined by the radial field component is dipolar-like, there are also smaller scale
magnetic features on the stellar surface. Since very few Ap stars were studied with
four Stokes parameter ZDI, it is difficult to ascertain how this picture depends on
the stellar mass and age. To this end, recent results seem to hint on the mass (or
age) dependence of the level of complexity of Ap star magnetic fields. Rusomarov
et al. (2015) reported no evidence of deviations from a dipolar field topology
based on the ZDI analysis of very high quality HARPSpol Stokes IQUV spectra
of the low-mass (and old) Ap star HD 24712. On the other hand, a couple of
magnetic early B stars (hence massive and young) were found to possess remarkable
non-dipolar magnetic geometries from a complex phase variation of their circular
polarisation profiles (Donati et al. 2006; Kochukhov et al. 2011). ZDI results for
one of these extraordinary stars, HD 37776, are presented in Fig. 9.8. In this case
the field structure does not resemble a dipole or quadrupole even remotely. So far
this extreme degree of the field complexity has only been seen in the most massive
early-type magnetic stars.
Several ongoing four Stokes parameter ZDI studies will increase the sample of
Ap stars with detailed information on the surface magnetic field structure, helping
to clarify the emerging trend of the field complexity with stellar parameters.
9 Doppler and Zeeman Doppler Imaging of Stars 195

Fig. 9.8 Unusually complex magnetic field structure of the B-type star HD 37776 reconstructed
with ZDI. The spherical maps show surface distributions of (a) the field modulus and (b) field
orientation. This magnetic field geometry clearly deviates from a simple axisymmetric dipole or
dipole+quadrupole topology. From Kochukhov et al. (2011)

Magnetic Mapping of Late-Type Active Stars

First ZDI maps of the magnetic field topologies of cool active stars were published
by Donati and Collier Cameron (1997), Donati (1999) and Donati et al. (1999,
2003). These authors studied a couple of rapidly rotating single dwarf stars (AB Dor,
LQ Hya) and the primary giant component of the RS CVn binary HR 1099. For all
three objects inversions were performed for several epochs, giving an idea of the
field evolution. These ZDI studies employed LSD profiles for the restricted ZDI
reconstruction (see above) of the brightness map from Stokes I and the magnetic
field distribution from Stokes V data.
Magnetic geometries of cool active stars turned out to be qualitatively different
from that of the Sun. Rather than featuring a system of bipolar regions with mostly
radial field orientation, as was expected by some “active Sun” models (Schrijver
and Title 2001), the first ZDI targets showed significant amounts of large-scale
horizontal magnetic fields. This field was found to be arranged in azimuthal bands
which evolved with time (Donati 1999). Another surprising result was the lack of
correlation between the dark photospheric regions and magnetic features, suggesting
that the strong magnetic fields inside star spots are not resolved or that the restricted
ZDI technique is not sensitive to such fields (Donati and Collier Cameron 1997). In
general, the local magnetic field strengths inferred by ZDI studies occasionally reach
1 kG in the strongest magnetic concentrations but, more typically, amount to only a
few hundred G. However, extrapolating from the physics of cool spot formation on
the solar surface, multi-kG fields should be ubiquitous in active stars with large star
spots.
Subsequently ZDI analyses were applied to many other classes of late-type
active stars, ranging from F stars to M dwarfs. A comprehensive review of these
investigations can be found in Donati and Landstreet (2009). One of the most
impressive achievements was an extension of the magnetic mapping to G-type
196 O. Kochukhov

stars with the overall activity levels and rotation rates comparable to the Sun (Petit
et al. 2008). This work demonstrated that magnetic inversions can constrain the
global field topologies even for stars with very small ve sin i. As discussed above
(Sect. 9.2.2), in that case information is primarily extracted from the rotational
modulation of the Stokes V signal rather than from Doppler shifts. For a sample
of four solar analogues Petit et al. (2008) found that the balance between poloidal
and toroidal contributions to the global magnetic field geometry depends on the
stellar rotation rate. Stars rotating faster than 12 days show predominantly toroidal
field and stars with a slower rotation exhibit poloidal field topology, reminiscent of
the global structure of the solar magnetic field.
A few active stars were systematically followed by ZDI over several years
(Fares et al. 2009; Morgenthaler et al. 2012; Kochukhov et al. 2013). In most
cases a significant change of the global field topology was detected. For example,
Fig. 9.9 illustrates the magnetic field reversal for the RS CVn star II Peg studied by
Kochukhov et al. (2013). It remains to be seen how these direct observations of the
magnetic field cycles relate to the behaviour of indirect magnetic proxies such as
the X-ray and Ca H&K emission measures. So far no clear link between the cycles
in direct magnetic observations and proxy indicators was found.
A new type of stellar magnetic field topologies was identified in active M dwarfs
stars with the help of ZDI (Morin et al. 2008; Donati et al. 2008). It turns out that the
convective dynamo mechanism operating in mid- and late-M dwarfs produces fairly
strong ( 500 G), large-scale magnetic fields. For the majority of these stars the field
topology is dipolar and aligned with the stellar rotational axis. On the other hand,
early M dwarfs tend to exhibit more complex, weaker and non-axisymmetric fields.
At the same time, it also became clear that the Stokes V ZDI of M dwarfs misses up
to 95 % of the magnetic flux (Reiners and Basri 2009) because the field modulus
measured from the Stokes I spectra of the same objects indicates 2–4 kG fields
(Johns-Krull et al. 1999; Reiners and Basri 2007), which is much larger than the
global field strength inferred by polarimetry. Presumably, these strong fields have a
complex structure and therefore cancel out in the disk-integrated polarisation signal.
Quantitative field topology models simultaneously reproducing both the Stokes I
and V observations of M dwarfs are yet to be developed.
Over the past few years significant efforts were made to test key assumptions of
the restricted ZDI inversions and to introduce more realistic polarisation modelling
methodologies in the ZDI of cool stars. Limitations of the traditional single-
line interpretation of the Stokes V LSD profiles were explored by Kochukhov
et al. (2010). Based on the experience gained from the general ZDI of Ap stars,
Kochukhov et al. (2013) performed ZDI inversions for the RS CVn star II Peg using
detailed self-consistent polarised radiative transfer calculations. According to the
numerical tests by Rosén and Kochukhov (2012), this should have been sufficient for
identifying strong unipolar magnetic fields inside cool spots. However, such fields
were not detected suggesting that the dark spots recovered by DI are not monolithic
but are composed of numerous bipolar groups whose polarisation signals cancel out
in the disk-integrated Stokes V spectra.
9 Doppler and Zeeman Doppler Imaging of Stars 197

Fig. 9.9 Self-consistent ZDI reconstruction of the vector magnetic field and temperature distribu-
tion for the RS CVn star II Peg at two different epochs. For each epoch the two columns on the
left compare the observed (symbols) and theoretical (lines) LSD Stokes I and V profiles. The four
rectangular maps illustrate distributions of the radial, meridional, and azimuthal field components,
and temperature. Significant evolution of the surface structure over a time span of 2 years is evident.
From Kochukhov et al. (2013)

In another development, Kochukhov et al. (2011) and Rosén et al. (2013)

obtained the first spectropolarimetric observations of cool active stars in all four
Stokes parameters. It was demonstrated that the full Stokes vector Zeeman spec-
tropolarimetry is feasible using LSD, at least for a handful of brightest objects. A
ZDI investigation based on these data was published by Rosén et al. (2015). Their
results indicate that, similar to the situation with ZDI of Ap stars, simultaneous
modelling of the Stokes IQUV spectra reveals considerably more complex magnetic
fields than suggested by the traditional Stokes V inversions.
198 O. Kochukhov

9.4 Conclusions

Doppler and Zeeman Doppler imaging have proven themselves as a powerful remote
sensing methods of obtaining detailed maps of spot distributions and magnetic field
topologies for unresolved stellar surfaces. DI and ZDI have been applied to a large
number of stars, leading to several important breakthroughs in our understanding
of the stellar magnetism and the processes of surface structure formation. As
a conclusion of this review we take a look into the future, summarising key
development directions of Doppler inversion studies of early- and late-type stars.
• DI of chemical spots Doppler inversions of chemical abundance inhomo-
geneities in chemically peculiar stars recover horizontal distributions under the
assumption of no significant vertical abundance variation. However, Ap stars
are known to exhibit vertical stratification of chemical elements (Ryabchikova
et al. 2002) and the apparent lateral inhomogeneities may well be a consequence
of the variation of vertical stratification over the stellar surface (Alecian and
Stift 2010). Therefore, chemical abundance DI should eventually incorporate the
vertical dimension in the inversion process, ultimately providing 3-D chemical
spot maps.
• DI of temperature spots This application of DI would benefit from an increase
of reliability of reconstruction of the physical properties of star spot interiors.
This can be accomplished by a systematic incorporation of the molecular
indicators in the temperature DI modelling and combining the optical and near-
infrared spectroscopic diagnostics. At some stage, the question of the vertical
temperature and pressure structure of magnetised star spots needs to be addressed
and dedicated models of spot atmospheres be developed and incorporated in DI.
• ZDI of early-type magnetic stars The methodology of self-consistent four
Stokes parameter inversions using polarisation profiles of individual lines is well-
established and thoroughly tested. However, this technique requires Stokes IQUV
observational data of superb quality that is unavailable for all but brightest stars.
It is of interest to pursue development of multi-line four Stokes parameter ZDI
methodology, using the S/N gain advantages of LSD but without compromising
on the detailed polarised radiative transfer calculations. First steps in this
direction were taken in the Stokes IV inversions described by Kochukhov et al.
(2014), but the method is yet to be tested for a full Stokes vector data set.
• ZDI of cool active stars Despite numerous applications of magnetic mapping
to different classes of active stars, there remain fundamental questions about
reliability of the ZDI inversions of complex field topologies using only Stokes IV
observations. Furthermore, the consequences of numerous simplifications of the
restricted ZDI (analytical line profiles, lack of self-consistency between spot and
magnetic field modelling, etc.) remain poorly explored. Occasional comparisons
of the ZDI reconstructions by independent inversion codes reveal uncomfortably
large discrepancies (see Skelly et al. 2010; Carroll et al. 2012). Methodological
improvements of the cool star ZDI are essential for understanding the absence
9 Doppler and Zeeman Doppler Imaging of Stars 199

of correlation between temperature and magnetic maps and the apparent lack of
strong magnetic fields that should be associated with cool star spots.

References

Adelman, S. J., Gulliver, A. F., Kochukhov, O. P., & Ryabchikova, T. A. (2002, August). The
variability of the Hg II 3984 line of the mercury-manganese star ˛ andromedae. Astrophysical
Journal, 575, 449–460.
Alecian, G., & Stift, M. J. (2010, June). Bi-dimensional element stratifications computed for
magnetic Ap star atmospheres. Astronomy and Astrophysics, 516, A53
Alecian, G., Stift, M. J., & Dorfi, E. A. (2011, December). Time-dependent diffusion in stellar
atmospheres. Monthly Notices of the Royal Astronomical Society, 418, 986–997.
Alecian, G., & Vauclair, S. (1981, August). Element stratification in the atmospheres of main
sequence stars - The silicon accumulation. Astronomy and Astrophysics, 101, 16–25.
Aurière, M., Wade, G. A., Lignières, F., Hui-Bon-Hoa, A., Landstreet, J. D., Iliev, I. K., et al. (2010,
November). No detection of large-scale magnetic fields at the surfaces of Am and HgMn stars.
Astronomy and Astrophysics, 523, A40.
Bagnulo, S., Landi Degl’Innocenti, M., Landolfi, M., & Mathys, G. (2002, November). A
statistical analysis of the magnetic structure of CP stars. Astronomy and Astrophysics, 394,
1023–1037.
Bagnulo, S., Landstreet, J. D., Fossati, L., & Kochukhov, O. (2012, February). Magnetic field
measurements and their uncertainties: the FORS1 legacy. Astronomy and Astrophysics, 538,
A129.
Bailey, J. D., Grunhut, J., Shultz, M., Wade, G., Landstreet, J. D., Bohlender, D., et al. (2012,
June). An analysis of the rapidly rotating Bp star HD 133880. Monthly Notices of the Royal
Astronomical Society, 423, 328–343.
Barnes, J. R., Collier Cameron, A., Donati, J.-F., James, D. J., Marsden, S. C., & Petit, P.
(2005, February). The dependence of differential rotation on temperature and rotation. Monthly
Notices of the Royal Astronomical Society, 357, L1-L5.
Berdyugina, S. V. (2005, December). Starspots: A key to the stellar dynamo. Living Reviews in
Solar Physics, 2, 8.
Berdyugina, S. V., Telting, J. H., Korhonen, H., & Schrijvers, C. (2003, July). Surface imaging of
stellar non-radial pulsations. II. The ˇ Cephei star ! 1 Sco. Astronomy and Astrophysics, 406,
281–285.
Briquet, M., Korhonen, H., González, J. F., Hubrig, S., & Hackman, T. (2010, February).
Dynamical evolution of titanium, strontium, and yttrium spots on the surface of the HgMn
star HD 11753. Astronomy and Astrophysics, 511, A71.
Brown, S. F., Donati, J.-F., Rees, D. E., & Semel, M. (1991, October). Zeeman-Doppler imaging
of solar-type and AP stars. IV - Maximum entropy reconstruction of 2D magnetic topologies.
Astronomy and Astrophysics, 463–474.
Carroll, T. A., Strassmeier, K. G., Rice, J. B., & Künstler, A. (2012, December). The magnetic
field topology of the weak-lined T Tauri star V410 Tauri. New strategies for Zeeman-Doppler
imaging. Astronomy and Astrophysics, 548, A95.
Collier Cameron, A. (1998). Stellar tomography. Astrophysics and Space Science, 261, 71–80.
Donati, J., & Collier Cameron, A. (1997, October). Differential rotation and magnetic polarity
patterns on AB Doradus. Monthly Notices of the Royal Astronomical Society, 291,1–19.
Donati, J.-F. (1999, January). Magnetic cycles of HR 1099 and LQ Hydrae. Monthly Notices of
the Royal Astronomical Society, 302, 457–481.
200 O. Kochukhov

Donati, J.-F., Cameron, A. C., Semel, M., Hussain, G. A. J., Petit, P., Carter, B. D., et al. (2003,
November). Dynamo processes and activity cycles of the active stars AB Doradus, LQ Hydrae
and HR 1099. Monthly Notices of the Royal Astronomical Society, 345, 1145–1186.
Donati, J.-F., Collier Cameron, A., Hussain, G. A. J., & Semel, M. (1999, January). Magnetic
topology and prominence patterns on AB Doradus. Monthly Notices of the Royal Astronomical
Society, 302, 437–456.
Donati, J.-F., Collier Cameron, A., & Petit, P. (2003, November). Temporal fluctuations in the
differential rotation of cool active stars. Monthly Notices of the Royal Astronomical Society,
345, 1187–1199.
Donati, J.-F., Howarth, I. D., Jardine, M. M., Petit, P., Catala, C., Landstreet, J. D., et al. (2006,
August). The surprising magnetic topology of Sco: fossil remnant or dynamo output? Monthly
Notices of the Royal Astronomical Society, 370, 629–644.
Donati, J.-F., & Landstreet, J. D. (2009, September) Magnetic fields of nondegenerate stars. The
Annual Review of Astronomy and Astrophysics, 47, 333–370.
Donati, J.-F., Morin, J., Petit, P., Delfosse, X., Forveille, T., Aurière, M., et al. (2008, September)
Large-scale magnetic topologies of early M dwarfs. Monthly Notices of the Royal Astronomical
Society, 390, 545–560.
Donati, J.-F., Semel, M., Carter, B. D., Rees, D. E., & Collier Cameron, A. (1997, November).
Spectropolarimetric observations of active stars. Monthly Notices of the Royal Astronomical
Society, 291, 658–682.
Donati, J.-F., Skelly, M. B., Bouvier, J., Gregory, S. G., Grankin, K. N., Jardine, M. M., et al.
(2010, December). Magnetospheric accretion and spin-down of the prototypical classical T
Tauri star AA Tau. Monthly Notices of the Royal Astronomical Society, 409, 1347–1361.
Fares, R., Donati, J.-F., Moutou, C., Bohlender, D., Catala, C., Deleuil, M., et al. (2009, September)
Magnetic cycles of the planet-hosting star Bootis - II. A second magnetic polarity reversal.
Monthly Notices of the Royal Astronomical Society, 398, 1383–1391.
Folsom, C. P., Kochukhov, O., Wade, G. A., Silvester, J., & Bagnulo, S. (2010, October). Magnetic
field, chemical composition and line profile variability of the peculiar eclipsing binary star AR
Aur. Monthly Notices of the Royal Astronomical Society, 407, 2383–2392.
Goncharskii, A. V., Ryabchikova, T. A., Stepanov, V. V., Khokhlova, V. L., & Yagola, A. G. (1983,
February). Mapping of elements on the surfaces of Ap-stars - part two - distribution of eu Sr and
Si on alpha-2-canum-venaticorum Ch-serpentis and Cu-virginis. Soviet Astronomy, 27, 49–53.
Goncharskii, A. V., Stepanov, V. V., Kokhlova, V. L., & Yagola, A. G. (1977). Reconstruction of
local line profiles from those observed in an Ap spectrum. Soviet Astronomy Letters, 3, 147–
149.
Hackman, T., Mantere, M. J., Lindborg, M., Ilyin, I., Kochukhov, O., Piskunov, N., et al. (2012,
February). Doppler images of II Pegasi for 2004–2010. Astronomy and Astrophysics, 538,
A126.
Hatzes, A. P. (1991, February). Doppler images of abundance features on Theta Aurigae. Monthly
Notices of the Royal Astronomical Society, 248, 487–493.
Hatzes, A. P. (1997, June). Doppler imaging of the silicon distribution on CUVir: Evidence for a
decentred magnetic dipole? Monthly Notices of the Royal Astronomical Society, 288, 153–160.
Hubrig, S., González, J. F., Savanov, I., Schöller, M., Ageorges, N., Cowley, C. R., et al. (2006,
October). Inhomogeneous surface distribution of chemical elements in the eclipsing binary
ARAur: A new challenge for our understanding of HgMn stars. Monthly Notices of the Royal
Astronomical Society, 371, 1953–1958.
Johns-Krull, C. M., Valenti, J. A., & Koresko, C. (1999, May). Measuring the magnetic field on
the classical T Tauri star BP Tauri. Astrophysical Journal, 516, 900–915.
Khokhlova, V. L., & Pavlova, V. M. (1984). Maps of the iron-group elements on the magnetic
Ap-star alpha-2-canum-venaticorum. Soviet Astronomy Letters, 10, 158–163.
Khokhlova, V. L., Rice, J. B., & Wehlau, W. H. (1986, August). Distribution of chemical elements
over the surface of the magnetic AP star Theta Aurigae. Astrophysical Journal, 307, 768–776.
Kochukhov, O. (2004a, August). Doppler imaging of stellar non-radial pulsations. I. Techniques
and numerical experiments. Astronomy and Astrophysics, 423, 613–628.
9 Doppler and Zeeman Doppler Imaging of Stars 201

Kochukhov, O. (2004b, November). Indirect imaging of nonradial pulsations in a rapidly

oscillating Ap star. Astrophysical Journal, 615, L149–L152.
Kochukhov, O. (2006, July). Remarkable non-dipolar magnetic field of the Bp star HD 137509.
Astronomy and Astrophysics, 454, 321–325.
Kochukhov, O., Adelman, S. J., Gulliver, A. F., & Piskunov, N. (2007, August). Weather in stellar
atmosphere revealed by the dynamics of mercury clouds in ˛ Andromedae. Nature Physics, 3,
526–529.
Kochukhov, O., Bagnulo, S., Wade, G. A., Sangalli, L., Piskunov, N., Landstreet, J. D., et al.
(2004, February). Magnetic doppler imaging of 53 camelopardalis in all four stokes parameters.
Astronomy and Astrophysics, 414, 613–632.
Kochukhov, O., Drake, N. A., Piskunov, N., & de la Reza, R. (2004, September) Multi-element
abundance Doppler imaging of the rapidly oscillating Ap star HR 3831. Astronomy and
Astrophysics, 424, 935–950.
Kochukhov, O., Lüftinger, T., Neiner, C., Alecian, E., & MiMeS Collaboration. (2014, May).
Magnetic field topology of the unique chemically peculiar star CU Virginis. Astronomy and
Astrophysics, 565, A83.
Kochukhov, O., Lundin, A., Romanyuk, I., & Kudryavtsev, D. (2011, January). The extraordinary
complex magnetic field of the helium-strong star HD 37776. Astrophysical Journal, 726, 24.
Kochukhov, O., Makaganiuk, V., & Piskunov, N. (2010, December). Least-squares deconvolution
of the stellar intensity and polarization spectra. Astronomy and Astrophysics, 524, A5.
Kochukhov, O., Makaganiuk, V., Piskunov, N., Jeffers, S. V., Johns-Krull, C. M., Keller, C.
U., et al. (2013, June). Are there tangled magnetic fields on HgMn stars? Astronomy and
Astrophysics, 554, A61.
Kochukhov, O., Makaganiuk, V., Piskunov, N., Snik, F., Jeffers, S. V., Johns-Krull, C. M., et al.
(2011, May). First detection of linear polarization in the line profiles of active cool stars.
Astrophysical Journal, 732, L19.
Kochukhov, O., Mantere, M. J., Hackman, T., & Ilyin, I. (2013, February). Magnetic field topology
of the RS CVn star II Pegasi. Astronomy and Astrophysics, 550, A84.
Kochukhov, O., & Piskunov, N. (2002, June). Doppler imaging of stellar magnetic fields. II.
Numerical experiments. Astronomy and Astrophysics, 388, 868–888.
Kochukhov, O., Piskunov, N., Ilyin, I., Ilyina, S., & Tuominen, I. (2002, July). Doppler Imaging
of stellar magnetic fields. III. Abundance distribution and magnetic field geometry of ˛ 2 CVn.
Astronomy and Astrophysics, 389, 420–438.
Kochukhov, O., Piskunov, N., Sachkov, M., & Kudryavtsev, D. (2005, September) Inhomoge-
neous distribution of mercury on the surfaces of rapidly rotating HgMn stars. Astronomy and
Astrophysics, 439, 1093–1098.
Kochukhov, O., & Wade, G. A. (2010, April). Magnetic Doppler imaging of ˛ 2 Canum
Venaticorum in all four Stokes parameters. Unveiling the hidden complexity of stellar magnetic
fields. Astronomy and Astrophysics, 513, A13.
Korhonen, H., Berdyugina, S. V., Hackman, T., Ilyin, I. V., Strassmeier, K. G., & Tuominen, I.
(2007, December). Study of FK Comae Berenices. V. Spot evolution and detection of surface
differential rotation. Astronomy and Astrophysics, 476, 881–891.
Korhonen, H., González, J. F., Briquet, M., Flores Soriano, M., Hubrig, S., Savanov, I., et al. (2013,
May). Chemical surface inhomogeneities in late B-type stars with Hg and Mn peculiarity. I.
Spot evolution in HD 11753 on short and long time scales. Astronomy and Astrophysics, 553,
A27.
Kovári, Z., Strassmeier, K. G., Granzer, T., Weber, M., Oláh, K., & Rice, J. B. (2004, April).
Doppler imaging of stellar surface structure. XXII. Time-series mapping of the young rapid
rotator LQ Hydrae. Astronomy and Astrophysics, 417, 1047–1054.
Landi Degl’Innocenti, E., & Landolfi, M. (2004). Polarization in Spectral Lines (Vol. 307).
Dordrecht: Kluwer Academic Publishers.
Landstreet, J. D., & Mathys, G. (2000, July). Magnetic models of slowly rotating magnetic Ap
stars: aligned magnetic and rotation axes. Astronomy and Astrophysics, 359, 213–226.
202 O. Kochukhov

Lee, U., & Saio, H. (1990, February). Line profile variations caused by low-frequency nonradial
pulsations of rapidly rotating stars. Astrophysical Journal, 349, 570–579.
Lüftinger, T., Kochukhov, O., Ryabchikova, T., Piskunov, N., Weiss, W. W., & Ilyin, I. (2010,
January). Magnetic Doppler imaging of the roAp star HD 24712. Astronomy and Astrophysics,
509(26), A71.
Makaganiuk, V., Kochukhov, O., Piskunov, N., Jeffers, S. V., Johns-Krull, C. M., Keller, C. U.,
et al. (2011, May). Chemical spots in the absence of magnetic field in the binary HgMn star 66
Eridani. Astronomy and Astrophysics, 529, A160.
Makaganiuk, V., Kochukhov, O., Piskunov, N., Jeffers, S. V., Johns-Krull, C. M., Keller, C.
U., et al. (2012, March). Magnetism, chemical spots, and stratification in the HgMn star '
Phoenicis. Astronomy and Astrophysics, 539, A142.
Marsden, S. C., Jardine, M. M., Ramírez Vélez, J. C., Alecian, E., Brown, C. J., Carter, B. D.,
et al. (2011, May). Magnetic fields and differential rotation on the pre-main sequence - I. The
early-G star HD 141943 - brightness and magnetic topologies. Monthly Notices of the Royal
Astronomical Society, 413, 1922–1938.
Mathys, G., Hubrig, S., Landstreet, J. D., Lanz, T., & Manfroid, J. (1997, June). The mean
magnetic field modulus of AP stars. Astronomy and AstrophysicsS, 123, 353–402.
Mégessier, C. (1975, March). The Ap star 108 Aqr. II - The oblique rotator model. Astronomy and
Astrophysics, 39, 263–273.
Michaud, G., Charland, Y., & Megessier, C. (1981, November). Diffusion models for magnetic
Ap-Bp stars. Astronomy and Astrophysics, 103, 244–262.
Mihalas, D. (1973, September) On the helium-spectrum variations of 56 Arietis and a Centauri.
Astrophysical Journal, 184, 851–871.
Morgenthaler, A., Petit, P., Saar, S., Solanki, S. K., Morin, J., Marsden, S. C., et al. (2012,
April). Long-term magnetic field monitoring of the Sun-like star Bootis A. Astronomy and
Astrophysics, 540, A138.
Morin, J., Donati, J.-F., Petit, P., Delfosse, X., Forveille, T., Albert, L., et al. (2008, September)
Large-scale magnetic topologies of mid M dwarfs. Monthly Notices of the Royal Astronomical
Society, 390, 567–581.
Nesvacil, N., Lüftinger, T., Shulyak, D., Obbrugger, M., Weiss, W., Drake, N. A., et al.
(2012, January). Multi-element Doppler imaging of the CP2 star HD 3980. Astronomy and
Astrophysics, 537, A151.
Petit, P., Dintrans, B., Solanki, S. K., Donati, J.- F., Aurière, M., Lignières, F., et al. (2008, July).
Toroidal versus poloidal magnetic fields in Sun-like stars: a rotation threshold. Monthly Notices
of the Royal Astronomical Society, 388, 80–88.
Piskunov, N., & Kochukhov, O. (2002, January). Doppler Imaging of stellar magnetic fields. I.
Techniques. Astronomy and Astrophysics, 381, 736–756.
Piskunov, N. E., Tuominen, I., & Vilhu, O. (1990, April). Surface imaging of late-type stars.
Astronomy and Astrophysics, 230, 363–370.
Polosukhina, N., Kurtz, D., Hack, M., North, P., Ilyin, I., Zverko, J., et al. (1999, November).
Lithium on the surface of cool magnetic CP stars I. Summary of spectroscopic observations
with three telescopes. Astronomy and Astrophysics, 351, 283–291.
Reiners, A., & Basri, G. (2007, February). The first direct measurements of surface magnetic fields
on very low mass stars. Astrophysical Journal, 656, 1121–1135.
Reiners, A., & Basri, G. (2009, March). On the magnetic topology of partially and fully convective
stars. Astronomy and Astrophysics, 496, 787–790.
Rice, J. B., Holmgren, D. E., & Bohlender, D. A. (2004, September) The distribution of oxygen
on the surface of the Ap star Aur. An abundance Doppler image to compare with UMa.
Astronomy and Astrophysics, 424, 237–244.
Rice, J. B., Strassmeier, K. G., & Kopf, M. (2011, February). The surface of V410 Tauri.
Astrophysical Journal, 728, 69.
Rice, J. B., & Wehlau, W. H. (1991, June). The range of abundances of iron, chromium, and
silicon over the surfaces of the CP stars Epsilon Ursae Majoris and Theta Aurigae. Astronomy
and Astrophysics, 246, 195–198.
9 Doppler and Zeeman Doppler Imaging of Stars 203

Rice, J. B., Wehlau, W. H., & Holmgren, D. E. (1997, October). The distribution of oxygen on
the surface of UMa: an abundance distribution Doppler image. Astronomy and Astrophysics,
326, 988–994.
Rice, J. B., Wehlau, W. H., & Khokhlova, V. L. (1989, January). Mapping stellar surfaces by
Doppler imaging - Technique and application. Astronomy and Astrophysics, 208, 179–188.
Rosén, L., & Kochukhov, O. (2012, December). How reliable is Zeeman Doppler imaging without
simultaneous temperature reconstruction? Astronomy and Astrophysics, 548, A8.
Rosén, L., Kochukhov, O., & Wade, G. A. (2013, November). Strong variable linear polarization
in the cool active star II Peg. Monthly Notices of the Royal Astronomical Society, 436, L10–L14.
Rosén, L., Kochukhov, O., & Wade, G. A. (2015, June). First Zeeman Doppler imaging of a cool
star using all four stokes parameters. Astrophysical Journal, 805, 169.
Rusomarov, N., Kochukhov, O., Piskunov, N., Jeffers, S. V., Johns-Krull, C. M., Keller, C. U.,
et al. (2013, October). Three-dimensional magnetic and abundance mapping of the cool Ap
star HD 24712 . I. Spectropolarimetric observations in all four Stokes parameters. Astronomy
and Astrophysics, 558, A8.
Rusomarov, N., Kochukhov, O., Ryabchikova, T., & Piskunov, N. (2015). Three-dimensional
magnetic and abundance mapping of the cool Ap star HD 24712 . II. Magnetic Doppler imaging
in all four Stokes parameters. Astronomy and Astrophysics, 573, A123.
Ryabchikova, T., Piskunov, N., Kochukhov, O., Tsymbal, V., Mittermayer, P., & Weiss, W. W.
(2002, March). Abundance stratification and pulsation in the atmosphere of the roAp star
boldmath gamma Equulei. Astronomy and Astrophysics, 384, 545–553.
Ryabchikova, T. A., Malanushenko, V. P., & Adelman, S. J. (1999, November). Orbital elements
and abundance analyses of the double-lined spectroscopic binary alpha Andromedae. Astron-
omy and Astrophysics, 351, 963–972.
Ryabchikova, T. A., Pavlova, V. M., Davydova, E. S., & Piskunov, N. E. (1996, November).
Surface distribution of chromium on the CP2 star HD 220825 ( Psc). Astronomy Letters, 22,
821–826.
Saio, H., & Gautschy, A. (2004, May). Axisymmetric p-mode pulsations of stars with dipole
magnetic fields. Monthly Notices of the Royal Astronomical Society, 350, 485–505.
Schrijver, C. J., & Title, A. M. (2001, April). On the formation of polar spots in sun-like stars.
Astrophysical Journal, 551, 1099–1106.
Silvester, J., Kochukhov, O., & Wade, G. A. (2014, May). Stokes IQUV magnetic Doppler imaging
of Ap stars - II. Next generation magnetic Doppler imaging of ˛ 2 CVn. Monthly Notices of the
Royal Astronomical Society, 440, 182–192.
Silvester, J., Wade, G. A., Kochukhov, O., Bagnulo, S., Folsom, C. P., & Hanes, D. (2012,
October). Stokes IQUV magnetic Doppler imaging of Ap stars - I. ESPaDOnS and NARVAL
observations. Monthly Notices of the Royal Astronomical Society, 426, 1003–1030.
Skelly, M. B., Donati, J.-F., Bouvier, J., Grankin, K. N., Unruh, Y. C., Artemenko, S. A., et al.
(2010, March). Dynamo processes in the T Tauri star V410 Tau. Monthly Notices of the Royal
Astronomical Society, 403, 159–169.
Stibbs, D. W. N. (1950). A study of the spectrum and magnetic variable star HD 125248. Monthly
Notices of the Royal Astronomical Society, 110, 395–404.
Strassmeier, K. G. (2009, September) Starspots. Astronomy and Astrophysics Review, 17, 251–
308.
Strassmeier, K. G. (2011, August). The zoo of starspots. In D. Prasad Choudhary & K. G.
Strassmeier (Eds.), IAU Symposium (Vol 273, pp. 174–180).
Strassmeier, K. G., Rice, J. B., Wehlau, W. H., Vogt, S. S., Hatzes, A. P., Tuominen, I., et al.
(1991, July). Doppler imaging of high-latitude spot activity on HD 26337. Astronomy and
Astrophysics, 247, 130–147.
Unruh, Y. C., Collier Cameron, A., & Cutispoto, G. (1995, December). The evolution of surface
structures on Ab-Doradus. Monthly Notices of the Royal Astronomical Society, 277, 1145.
Vogt, S. S., Hatzes, A. P., Misch, A. A., & Kürster, M. (1999, April). Doppler imagery of
the spotted RS Canum Venaticorum Star HR 1099 (V711 Tauri) from 1981 to 1992. The
Astrophysical Journal Supplement Series, 121, 547–589.
204 O. Kochukhov

Vogt, S. S., & Penrod, G. D. (1983a, December). Detection of high-order nonradial oscillations
on the rapid rotator Zeta Ophiuchi and their link with Be type outbursts. Astrophysical Journal
Supplement Series, 275, 661–682.
Vogt, S. S., & Penrod, G. D. (1983b, September) Doppler imaging of spotted stars - Application
to the RS Canum Venaticorum star HR 1099. Publications of the Astronomical Society of the
Pacific, 95, 565–576.
Vogt, S. S., Penrod, G. D., & Hatzes, A. P. (1987, October). Doppler images of rotating stars using
maximum entropy image reconstruction. Astrophysical Journal, 321, 496–515.
Wade, G. A., Donati, J.-F., Landstreet, J. D., & Shorlin, S. L. S. (2000, April). Spectropolarimetric
measurements of magnetic Ap and Bp stars in all four Stokes parameters. Monthly Notices of
the Royal Astronomical Society, 313, 823–850.