Context. The Evershed effect, a nearly horizontal outflow of material seen in the penumbrae of sunspots in the photospheric layers, is a common characteristic of well-developed penumbrae, but is still not well understood. Even less is known about photospheric horizontal inflows in the penumbra, also known as counter Evershed flows. Aims. Here we present a rare feature observed in the penumbra of the main sunspot of AR NOAA 10930. This spot displays the normal Evershed outflow in most of the penumbra, but harbors a fast photospheric inflow of material over a large sector of the disk-center penumbra. We investigate the driving forces of both, the normal and the counter Evershed flows. Methods. We invert the spectropolarimetric data from Hinode SOT/SP using the spatially coupled version of the SPINOR inversion code, which allows us to derive height-dependent maps of the relevant physical parameters in the sunspot. These maps show considerable fine structure. Similarities and differences between the normal Evershed outflow and the counter Evershed flow are investigated. Results. In both the normal and the counter Evershed flows, the material flows from regions with field strengths of the order of 1.5-2 kG to regions with stronger fields. The sources and sinks of both penumbral flows display opposite field polarities, with the sinks (tails of filaments) harboring local enhancements in temperature, which are nonetheless colder than their sources (heads of filaments). Conclusions. The anti-correlation of the gradients in the temperature and magnetic pressure between the endpoints of the filaments from the two distinct penumbral regions is compatible with both the convective driver and the siphon flow scenarios. A geometrical scale of the parameters is necessary to determine which is the dominant force driving the flows.

We present observations of transverse oscillations in slender Ca II H fibrils (SCFs) in the lower solar chromosphere. We use a 1 hr long time series of high- (spatial and temporal-) resolution seeing-free observations in a 0.11 nm wide passband covering the line core of Ca II H 396.9 nm from the second flight of the Sunrise balloon-borne solar observatory. The entire field of view, spanning the polarity inversion line of an active region close to the solar disk center, is covered with bright, thin, and very dynamic fine structures. Our analysis reveals the prevalence of transverse waves in SCFs with median amplitudes and periods on the order of 2.4+-0.8 km/s and 83+-29 s, respectively (with standard deviations given as uncertainties). We find that the transverse waves often propagate along (parts of) the SCFs with median phase speeds of 9+-14 km/s. While the propagation is only in one direction along the axis in some of the SCFs, propagating waves in both directions, as well as standing waves are also observed. The transverse oscillations are likely Alfvenic and are thought to be representative of magnetohydrodynamic kink waves. The wave propagation suggests that the rapid high-frequency transverse waves, often produced in the lower photosphere, can penetrate into the chromosphere with an estimated energy flux of ~ 15 kW/m^2. Characteristics of these waves differ from those reported for other fibrillar structures, which, however, were observed mainly in the upper solar chromosphere.

Convective flows are known as the prime means of transporting magnetic fields on the solar surface. Thus, small magnetic structures are good tracers of the turbulent flows. We study the migration and dispersal of magnetic bright features (MBFs) in intergranular areas observed at high spatial resolution with Sunrise/IMaX. We describe the flux dispersal of individual MBFs as a diffusion process whose parameters are computed for various areas in the quiet Sun and the vicinity of active regions from seeing-free data. We find that magnetic concentrations are best described as random walkers close to network areas (diffusion index, gamma=1.0), travelers with constant speeds over a supergranule (gamma=1.9-2.0), and decelerating movers in the vicinity of flux emergence and/or within active regions (gamma=1.4-1.5). The three types of regions host MBFs with mean diffusion coefficients of 130 km^2/s, 80-90 km^2/s, and 25-70 km^2/s, respectively. The MBFs in these three types of regions are found to display a distinct kinematic behavior at a confidence level in excess of 95%.

A dense forest of slender bright fibrils near a small solar active region is seen in high-quality narrowband Ca II H images from the SuFI instrument onboard the Sunrise balloon-borne solar observatory. The orientation of these slender Ca II H fibrils (SCF) overlaps with the magnetic field configuration in the low solar chromosphere derived by magnetostatic extrapolation of the photospheric field observed with Sunrise/IMaX and SDO/HMI. In addition, many observed SCFs are qualitatively aligned with small-scale loops computed from a novel inversion approach based on best-fit numerical MHD simulation. Such loops are organized in canopy-like arches over quiet areas that differ in height depending on the field strength near their roots.

How and where are coronal loops rooted in the solar lower atmosphere? The details of the magnetic environment and its evolution at the footpoints of coronal loops are crucial to understanding the processes of mass and energy supply to the solar corona. To address the above question, we use high-resolution line-of-sight magnetic field data from the Imaging Magnetograph eXperiment instrument on the SUNRISE balloon-borne observatory and coronal observations from the Atmospheric Imaging Assembly onboard the Solar Dynamics Observatory of an emerging active region. We find that the coronal loops are often rooted at the locations with minor small-scale but persistent opposite-polarity magnetic elements very close to the larger dominant polarity. These opposite-polarity small-scale elements continually interact with the dominant polarity underlying the coronal loop through flux cancellation. At these locations we detect small inverse Y-shaped jets in chromospheric Ca II H images obtained from the SUNRISE Filter Imager during the flux cancellation. Our results indicate that magnetic flux cancellation and reconnection at the base of coronal loops due to mixed polarity fields might be a crucial feature for the supply of mass and energy into the corona.

Magneto-static models may overcome some of the issues facing force-free magnetic field extrapolations. So far they have seen limited use and have faced problems when applied to quiet-Sun data. Here we present a first application to an active region. We use solar vector magnetic field measurements gathered by the IMaX polarimeter during the flight of the \sunrise{} balloon-borne solar observatory in June 2013 as boundary condition for a magneto-static model of the higher solar atmosphere above an active region. The IMaX data are embedded in active region vector magnetograms observed with SDO/HMI. This work continues our magneto-static extrapolation approach, which has been applied earlier ({\it Paper I}) to a quiet Sun region observed with \sunrise{} I. In an active region the signal-to-noise-ratio in the measured Stokes parameters is considerably higher than in the quiet Sun and consequently the IMaX measurements of the horizontal photospheric magnetic field allow us to specify the free parameters of the model in a special class of linear magneto-static equilibria. The high spatial resolution of IMaX (110-130 km, pixel size 40 km) enables us to model the non-force-free layer between the photosphere and the mid chromosphere vertically by about 50 grid points. In our approach we can incorporate some aspects of the mixed beta layer of photosphere and chromosphere, e.g., taking a finite Lorentz force into account, which was not possible with lower resolution photospheric measurements in the past. The linear model does not, however, permit to model intrinsic nonlinear structures like strongly localized electric currents.

We report the detection of oscillations in slender Ca II H fibrils (SCFs) from high-resolution observations acquired with the Sunrise balloon-borne solar observatory. The SCFs show obvious oscillations in their intensity, but also their width. The oscillatory behaviors are investigated at several positions along the axes of the SCFs. A large majority of fibrils show signs of oscillations in intensity. Their periods and phase speeds are analyzed using a wavelet analysis. The width and intensity perturbations have overlapping distributions of the wave period. The obtained distributions have median values of the period of $32\pm17$s and $36\pm25$s, respectively. We find that the fluctuations of both parameters propagate in the SCFs with speeds of ${11}^{+49}_{-11}$ km/s and ${15}^{+34}_{-15}$ km/s, respectively. Furthermore, the width and intensity oscillations have a strong tendency to be either in anti-phase, or, to a smaller extent, in phase. This suggests that the oscillations of both parameters are caused by the same wave mode and that the waves are likely propagating. Taking all the evidence together, the most likely wave mode to explain all measurements and criteria is the fast sausage mode.

The SUNRISE balloon-borne solar observatory, consisting of a 1~m aperture telescope that provided a stabilized image to a UV filter imager and an imaging vector polarimeter, carried out its second science flight in June 2013. It provided observations of parts of active regions at high spatial resolution, including the first high-resolution images in the Mg~{\sc ii}~k line. The obtained data are of very high quality, with the best UV images reaching the diffraction limit of the telescope at 3000~\AA\ after Multi-Frame Blind Deconvolution reconstruction accounting for phase-diversity information. Here a brief update is given of the instruments and the data reduction techniques, which includes an inversion of the polarimetric data. Mainly those aspects that evolved compared with the first flight are described. A tabular overview of the observations is given. In addition, an example time series of a part of the emerging active region NOAA AR~11768 observed relatively close to disk centre is described and discussed in some detail. The observations cover the pores in the trailing polarity of the active region, as well as the polarity inversion line where flux emergence was ongoing and a small flare-like brightening occurred in the course of the time series. The pores are found to contain magnetic field strengths ranging up to 2500~G and, while large pores are clearly darker and cooler than the quiet Sun in all layers of the photosphere, the temperature and brightness of small pores approach or even exceed those of the quiet Sun in the upper photosphere.

We use seeing-free high spatial resolution Ca II H data obtained by the SUNRISE observatory to determine properties of slender fibrils in the lower solar chromosphere. In this work we use intensity images taken with the SUFI instrument in the Ca II H line during the second scientific flight of the SUNRISE observatory to identify and track elongated bright structures. After the identification, we analyze theses structures in order to extract their morphological properties. We identify 598 slender Ca II H fibrils (SCFs) with an average width of around 180 km, a length between 500 km and 4000 km, an average lifetime of ~400 s, and an average curvature of 0.002 arcsec^-1. The maximum lifetime of the SCFs within our time series of 57 minutes is ~2000 s. We discuss similarities and differences of the SCFs with other small-scale, chromospheric structures such as spicules of type I and II, or Ca II K fibrils.

We study the dynamics and topology of an emerging magnetic flux concentration using high spatial resolution spectropolarimetric data acquired with the Imaging Magnetograph eXperiment on board the Sunrise balloon-borne solar observatory. We obtain the full vector magnetic field and the line-of-sight (LOS) velocity through inversions of the Fe I line at 525.02 nm with the SPINOR code. The derived vector magnetic field is used to trace magnetic field lines. Two magnetic flux concentrations with different polarity and LOS velocities are found to be connected by a group of arch-shaped magnetic field lines. The positive polarity footpoint is weaker (1100 G) and displays an upflow, while the negative polarity footpoint is stronger (2200 G) and shows a downflow. This configuration is naturally interpreted as a siphon flow along an arched magnetic flux tube.

We present a new method of Stokes inversion of spectropolarimetric data and evaluate it by taking the example of a SUNRISE/IMaX observation. An archive of synthetic Stokes profiles is obtained by the spectral synthesis of state-of-the-art magnetohydrodynamics (MHD) simulations and a realistic degradation to the level of the observed data. The definition of a merit function allows the archive to be searched for the synthetic Stokes profiles that match the observed profiles best. In contrast to traditional Stokes inversion codes, which solve the Unno-Rachkovsky equations for the polarized radiative transfer numerically and fit the Stokes profiles iteratively, the new technique provides the full set of atmospheric parameters. This gives us the ability to start an MHD simulation that takes the inversion result as initial condition. After a relaxation process of half an hour solar time we obtain physically consistent MHD data sets with a target similar to the observation. The new MHD simulation is used to repeat the method in a second iteration, which further improves the match between observation and simulation, resulting in a factor of 2.2 lower mean $\chi^2$ value. One advantage of the new technique is that it provides the physical parameters on a geometrical height scale. It constitutes a first step towards inversions giving results consistent with the MHD equations.

We try to retrieve the power spectra with certainty to the highest spatial frequencies allowed by current instrumentation. For this, we use 2D inversion code that were able to recover information up to the instrumental diffraction limit. The retrieved power spectra have shallow slopes extending further down to much smaller scales than found before. They seem not to show any power law. The observed slopes at subgranular scales agree with those obtained from recent local dynamo simulations. Small differences are found for vertical component of kinetic energy that suggest that observations suffer from an instrumental effect that is not taken into account.

Properties of magnetic field in the internetwork regions are still fairly unknown due to rather weak spectropolarimetric signals. We address the matter by using the 2D inversion code that is able to retrieve the information on smallest spatial scales, up to the diffraction limit, while being less susceptible to noise than most of the previous methods used. Performance of the code and the impact of the various effects on the retrieved field distribution is tested first on the realistic MHD simulations. The best inversion scenario is then applied to the real Hinode/SP data. Tests on simulations show: (1) the best choice of node position ensures a decent retrieval of all parameters, (2) code performs well for different configurations of magnetic field, (3) slightly different noise level or slightly different defocus included in the spatial PSF produces no significant effect on the results and (4) temporal integration shifts the field distribution to the stronger, more horizontally inclined field. Although the contribution of the weak field is slightly overestimated due to noise, the 2D inversions are able to recover well the overall distribution of the magnetic field strength. Application of the 2D inversion code on the Hinode/SP internetwork observations reveals a monotonic field strength distribution. The mean field strength at optical depth unity is $\sim 130$~G. At the higher layers, field strength drops as the field becomes more horizontal. Regarding the distribution of the field inclination, tests show that we cannot directly retrieve it with the observations/tools at hand, however the obtained distributions are consistent with those expected from simulations with a quasi-isotropic field inclination after accounting for observational effects.

The formation of an Ha filament in active region (AR) 10953 is described. Observations from the Solar Optical Telescope (SOT) aboard the Hinode satellite starting on 27th April 2007 until 1st May 2007 were analysed. 20 scans of the 6302A Fe I line pair recorded by SOT/SP were inverted using the SPINOR code. The inversions were analysed together with SOT/BFI G-band and Ca II H and SOT/NFI Ha observations. Following the disappearance of an initial Ha filament aligned along the polarity inversion line (PIL) of the AR, a new Ha filament formed in its place some 20 hours later, which remained stable for at least 1.5 days. The creation of the new Ha filament was driven by the ascent of horizontal magnetic fields from the photosphere into the chromosphere at three separate locations along the PIL. The magnetic fields at two of these locations were situated directly underneath the initial Ha filament and formed orphan penumbrae already aligned along the Ha filament channel. The 700 G orphan penumbrae were stable and trapped in the photosphere until the disappearance of the overlying initial Ha filament, after which they started to ascend into the chromosphere at 10pm5 m/s. Each ascent was associated with a simultaneous magnetic flux reduction of up to 50% in the photosphere. The ascended orphan penumbrae formed dark 'seed' structures in Ha in parallel with the PIL, which elongated and merged to form an Ha filament. The filament channel featured horizontal magnetic fields of on average 260 G at log(tau)=-2 suspended above the nearly field-free lower photosphere. The fields took on an 'inverse' configuration at log(tau)=-2 suggesting a flux rope for the new Ha filament. The orphan penumbral fields ascend into the chromosphere 9-24 hours before the Ha filament is fully formed. The destruction of the initial Ha filament was likely caused by the flux emergence at the third location along the PIL.

The properties of magnetic fields forming an extended plage region in AR 10953 were investigated. Stokes spectra of the Fe I line pair at 6302 \AA recorded by the spectropolarimeter aboard the Hinode satellite were inverted using the SPINOR code. The code performed a 2D spatially coupled inversion on the Stokes spectra, allowing the retrieval of gradients in optical depth within the atmosphere of each pixel, whilst accounting for the effects of the instrument's PSF. Consequently, no magnetic filling factor was needed. The inversion results reveal that plage is composed of magnetic flux concentrations (MFCs) with typical field strengths of 1520 G at log(\tau)=-0.9 and inclinations of 10-15 degrees. The MFCs expand by forming magnetic canopies composed of weaker and more inclined magnetic fields. The expansion and average temperature stratification of isolated MFCs can be approximated well with an empirical plage thin flux-tube model. The highest temperatures of MFCs are located at their edges in all log(\tau) layers. Whilst the plasma inside MFCs is nearly at rest, each is surrounded by a ring of downflows of on average 2.4 km/s at log(\tau)=0 and peak velocities of up to 10 km/s, which are supersonic. The downflow ring of an MFC weakens and shifts outwards with height, tracing the MFC's expansion. Such downflow rings often harbour magnetic patches of opposite polarity to that of the main MFC with typical field strengths below 300 G at log(\tau)=0. These opposite polarity patches are situated beneath the canopy of their main MFC. We found evidence of a strong broadening of the Stokes profiles in MFCs and particularly in the downflow rings surrounding MFCs (expressed by a microturbulence in the inversion). This indicates the presence of strong unresolved velocities. Larger magnetic structures such as sunspots cause the field of nearby MFCs to be more inclined.

High levels of horizontal magnetic flux have been reported in the quiet-Sun internetwork, often based on Stokes profile inversions. Here we introduce a new method for deducing the inclination of magnetic elements and use it to test magnetic field inclinations from inversions. We determine accurate positions of a set of small, bright magnetic elements in high spatial resolution images sampling different photospheric heights obtained by the Sunrise balloon-borne solar observatory. Together with estimates of the formation heights of the employed spectral bands, these provide us with the inclinations of the magnetic features. We also compute the magnetic inclination angle of the same magnetic features from the inversion of simultaneously recorded Stokes parameters. Our new, geometric method returns nearly vertical fields (average inclination of around 14 deg with a relatively narrow distribution having a standard deviation of 6 deg). In strong contrast to this, the traditionally used inversions give almost horizontal fields (average inclination of 75+-8 deg) for the same small magnetic features, whose linearly polarised Stokes profiles are adversely affected by noise. The almost vertical field of bright magnetic features from our geometric method is clearly incompatible with the nearly horizontal magnetic fields obtained from the inversions. This indicates that the amount of magnetic flux in horizontal fields deduced from inversions is overestimated in the presence of weak Stokes signals, in particular if Stokes Q and U are close to or under the noise level. By combining the proposed method with inversions we are not just improving the inclination, but also the field strength. This technique allows us to analyse features that are not reliably treated by inversions, thus greatly extending our capability to study the complete magnetic field of the quiet Sun.

A new Stokes inversion technique that greatly reduces the effect of the spatial point spread function of the telescope is used to constrain the physical properties of umbral dots (UDs). The depth-dependent inversion of the Stokes parameters from a sunspot umbra recorded with Hinode SOT/SP revealed significant temperature enhancements and magnetic field weakenings in the core of the UDs in deep photospheric layers. Additionally, we found upflows of around 960 m/s in peripheral UDs (i.e., UDs close to the penumbra) and $\approx$ 600 m/s in central UDs. For the first time, we also detected systematic downflows for distances larger than 200 km from the UD center that balance the upflowing mass flux. In the upper photosphere, we found almost no difference between the UDs and their diffuse umbral background.

We present new high-resolution spectro-polarimetric Ca II 8542 observations of umbral flashes in sunspots. At nearly 0.18", and spanning about one hour of continuous observation, this is the most detailed dataset published thus far. Our study involves both LTE and non-LTE inversions (but includes also a weak field analysis as a sanity check) to quantify temperatures, mass flows and the full magnetic field vector geometry. We confirm earlier reports that UFs have very fine structure with hot and cool material intermixed at sub-arcsecond scales. The shock front is roughly 1000 K hotter than the surrounding material. We do not observe significant fluctuations of the field in the umbra. In the penumbra, however, the passage of the running penumbral waves alter the magnetic field strength by some 200 G (peak-to-peak amplitude) but it does not change the field orientation (at least not significantly within our sensitivity of a few degrees). We find a trend of decreasing high-frequency modulation power for more inclined magnetic fields for the line-of-sight velocity and magnetic field strength. In the outer penumbra we find an absence of high frequency power while there is increasingly more power at high frequencies towards the umbra.

When inverting solar spectra, image degradation effects that are present in the data are usually approximated or not considered. We develop a data reduction method that takes these issues into account and minimizes the resulting errors. By accounting for the diffraction PSF of the telescope during the inversions, we can produce a self-consistent solution that best fits the observed data, while simultaneously requiring fewer free parameters than conventional approaches. Simulations using realistic MHD data indicate that the method is stable for all resolutions, including those with pixel scales well beyond those that can be resolved with a 0.5m telescope, such as the Hinode SOT. Application of the presented method to reduce full Stokes data from the Hinode spectro-polarimeter results in dramatically increased image contrast and an increase in the resolution of the data to the diffraction limit of the telescope in almost all Stokes and fit parameters. The resulting data allow for detecting and interpreting solar features that have so far only been observed with 1m class ground-based telescopes. The new inversion method allows for accurate fitting of solar spectro-polarimetric imaging data over a large field of view, while simultaneously improving the noise statistics and spatial resolution of the results significantly.

context: The combination of image restoration and a Fabry-P\'{e}rot interferometer (FPI) based instrument in solar observations results in specific calibration issues. FPIs generally show variations over the field-of-view, while in the image restoration process, the 1-to-1 relation between pixel space and image space is lost, thus complicating any correcting for such variations. aims: We develop a data reduction method that takes these issues into account and minimizes the resulting errors. methods: By accounting for the time variations in the telescope's Mueller matrix and using separate calibration data optimized for the wavefront sensing in the MOMFBD image restoration process and for the final deconvolution of the data, we have removed most of the calibration artifacts from the resulting data. results: Using this method to reduce full Stokes data from CRISP at the SST, we find that it drastically reduces the instrumental and image restoration artifacts resulting from cavity errors, reflectivity variations, and the polarization dependence of flatfields. The results allow for useful scientific interpretation. Inversions of restored data from the $\delta$ sunspot AR11029 using the Nicole inversion code, reveal strong (~10 km/s) downflows near the disk center side of the umbra. conclusions: The use of image restoration in combination with an FPI-based instrument leads to complications in the calibrations and intrinsic limitations to the accuracy that can be achieved. We find that for CRISP, the resulting errors can be kept mostly below the polarimetric accuracy of ~10^-3. Similar instruments aiming for higher polarimetric and high spectroscopic accuracy, will, however, need to take these problems into account. keywords: Techniques: image processing, polarimetric, imaging spectroscopy, Sun: surface magnetism, sunspots, activity

In the present work, we investigate the formation of Stokes profiles and spectro-polarimetric diagnostics in an active region plage near the limb. We use 3-D radiation-MHD simulations with unipolar fields of an average strength of 400 G, which is largely concentrated in flux tubes in which the field reaches typical kilo-Gauss values. We generate synthetic Stokes spectra by radiative transfer calculations, then we degrade the simulated Stokes signal to account for observational conditions. The synthetic data treated in this manner are compared with and found to roughly reproduce spectro-polarimetric high-resolution observations at Mu=0.39 obtained by the SOUP instrument with the Swedish 1-m Solar Telescope at the beginning of 2006.

We discuss penumbral fine structure in a small part of a pore, observed with the CRISP imaging spectropolarimeter at the Swedish 1-m Solar Telescope (SST), close to its diffraction limit of 0.16 arcsec. Milne-Eddington inversions applied to these Stokes data reveal large variations of field strength and inclination angle over dark-cored penumbral intrusions and a dark-cored light bridge. The mid-outer part of this penumbra structure shows 0.3 arcsec wide spines, separated by 1.6 arcsec (1200 km) and associated with 30 deg inclination variations. Between these spines, there are no small-scale magnetic structures that easily can be be identified with individual flux tubes. A structure with nearly 10 deg more vertical and weaker magnetic field is seen midways between two spines. This structure is co-spatial with the brightest penumbral filament, possibly indicating the location of a convective upflow from below.

High-resolution time series of sunspots have been obtained with the Swedish 1m Solar Telescope between 2003 and 2006 at different locations on the solar disc. Proper motions in seven different active regions have been studied. The analysis has been done by applying local correlation tracking to every series of sunspots, each of them more than 40 minutes long. The sunspots' shapes include a different variety of penumbral configurations. We report on a systematic behaviour of the large-scale outflows surrounding the sunspots, commonly known as moat flows, that are essentially present only when preceded by a penumbra not tangential but perpendicular to the sunspot border. We present one case for which this rule appears not to be confirmed. We speculate that the magnetic neutral line, which is located in the vicinity of the anomalous region, might be responsible for blocking the outflow. These new results confirm the systematic and strong relation between the moat flows and the existence of penumbrae. A comparative statistical study between moats and standard granulation is also performed.

The formation of jets such as dynamic fibrils, mottles, and spicules in the solar chromosphere is one of the most important, but also most poorly understood, phenomena of the Sun's magnetized outer atmosphere. We use extremely high-resolution observations from the Swedish 1-m Solar Telescope combined with advanced numerical modeling to show that in active regions these jets are a natural consequence of upwardly propagating slow mode magnetoacoustic shocks. These shocks form when waves generated by convective flows and global p-mode oscillations in the lower lying photosphere leak upward into the magnetized chromosphere. We find excellent agreement between observed and simulated jet velocities, decelerations, lifetimes and lengths. Our findings suggest that previous observations of quiet sun spicules and mottles may also be interpreted in light of a shock driven mechanism.

We present high-quality HST/GHRS spectra in the Hydrogen L alpha spectral region of Vega and Sirius-A. Thanks to the signal-to-noise ratio achieved in these observations and to the similarity of the two spectra, we found clear evidence of emission features in the low flux region, 1190-1222 A. These emission lines can be attributed unambiguously to Fe II and Cr II transitions. In this spectral range, silicon lines are observed in absorption. We built a series of non-LTE model atmospheres with different, prescribed temperature stratification in the upper atmosphere and treating Fe II with various degrees of sophistication in non-LTE. Emission lines are produced by the combined effect of the Schuster mechanism and radiative interlocking, and can be explained without the presence of a chromosphere. Silicon absorption lines and the L alpha profile set constraints on the presence of a chromosphere, excluding a strong temperature rise in layers deeper than tau_R ~ 10^{-4}.