Investigating the magnetospheric accretion process in the young pre-transitional disk system DoAr 44 (V2062~Oph). A multiwavelength interferometric, spectropolarimetric, and photometric observing campaign

Young stars interact with their accretion disk through their strong magnetosphere. We investigate the magnetospheric accretion process in the young stellar system DoAr 44. We monitored the system over several rotational cycles, combining high-resolution optical and near-IR spectropolarimetry with long-baseline near-IR interferometry and multicolor photometry. DoAr 44 is a young 1.2 solar mass star, moderately accreting from its disk, and seen at a low inclination. We derive a rotational period of 2.96 d from the system's light curve. Several optical and near-IR line profiles probing the accretion funnel flows and the accretion shock are modulated at the stellar rotation period. The most variable line profile, HeI 1083 nm, exhibits modulated redshifted wings a signature of accretion funnel flows, as well as deep blueshifted absorptions indicative of transient outflows. The Zeeman-Doppler analysis suggests the star hosts a mainly dipolar magnetic field, inclined by about 20 deg. onto the spin axis, with an intensity reaching about 800 G at the photosphere, and up to 2 +/- 0.8 kG close to the accretion shock. The magnetic field appears strong enough to disrupt the inner disk close to the corotation radius, at a distance of about 4.6 stellar radii (0.043 au). This supports the upper limit of 5 stellar radii (0.047 au) we derived for the size of the magnetosphere from long baseline interferometry. DoAr 44 is a pre-transitional disk system, exhibiting a 25-30 au gap in its circumstellar disk, with the inner and outer disks being misaligned. On a scale of 0.1 au or less, our results indicate that the system steadily accretes from its inner disk through its tilted dipolar magnetosphere. We conclude that in spite of a highly structured outer disk, perhaps the signature of ongoing planetary formation, the magnetospheric accretion process proceeds unimpeded at the star-disk interaction level.


Introduction
T Tauri stars are low-mass pre-main sequence stars at an early stage of their evolution. For a few million years, they accrete material from their circumstellar disks, a remnant of the protostellar collapse and the birthplace of planetary systems. The socalled classical T Tauri stars (cTTSs) have specific properties, such as a rich emission-line spectrum, large photometric variability, and a strong continuum excess flux superimposed onto Based on observations obtained at the Canada-France-Hawaii Telescope (CFHT), at the European Organisation for Astronomical Research in the Southern Hemisphere under ESO programme 0103.C-0097, and at the Las Cumbres Observatory global telescope network (LCOGT). a late-type photospheric spectrum (Herbig 1962). The unique properties of this class of objects eventually vanish as the disk dissipates on a timescale of a few million years, and the young stellar object transitions to a weak-line T Tauri star (wTTS, Ménard & Bertout 1999). Indeed, many of the specific properties of cTTSs are thought to derive from the accretion process (Kenyon & Hartmann 1987;Bertout et al. 1988;Basri & Bertout 1989) and, more specifically, to arise from the interaction region between the inner disk and the star.
The currently accepted paradigm that describes this interaction is the magnetospheric accretion scenario where the strong stellar magnetic field controls, at least partly, the accretion flow from the inner disk edge to the stellar surface (Camenzind 1990;Koenigl 1991, see reviews by Bouvier et al. 2007b; Article number, page 1 of 17 arXiv:2010.00887v1 [astro-ph.SR] 2 Oct 2020 A&A proofs: manuscript no. 38892corr et al. 2016). With large-scale surface magnetic fields of a few hundred to a few kG (e.g., Donati 2020;Sokal et al. 2020), the stellar magnetosphere truncates the inner gaseous disk at a distance of a few stellar radii (typically 3-8 R ) above the photosphere. From there, the accreted material is channeled along magnetic funnel flows, and hits the stellar surface almost at freefall velocity to create a localized accretion shock where the kinetic energy of the flow is dissipated. In this scenario, the emission line spectrum of cTTSs is partly formed in the accretion funnel flows (Hartmann et al. 1994), and the accretion shock is responsible for the continuum excess flux from the X-ray range to the optical wavelengths, which is referred to as veiling (Calvet & Gullbring 1998).
The magnetospheric process extends over a scale less than 0.1 au from the star, which corresponds to ≤1 mas angular separation on the sky at the distance of nearby star forming regions. Therefore, most constraints on the physics of magnetospheric accretion come from observational campaigns monitoring the spectral variability of selected cTTSs over successive rotation periods. The rotational modulation over a timescale of weeks of emission line profiles, veiling, and Zeeman signatures allows one to draw magnetic maps of the stellar surface and to reconstruct the structure of the magnetospheric accretion region and investigate its dynamics. Previous monitoring campaigns have been quite successful in interpreting the observed properties and variability of young stellar objects in the framework of the magnetospheric accretion model (e.g., Bouvier et al. 2007a;Alencar et al. 2012Alencar et al. , 2018Donati et al. 2019). These provided new insights into the physics of the interaction region between the inner disk edge and the stellar surface, which impacts both early stellar evolution, and, potentially, planet formation and/or migration at the inner disk edge.
We present here the results of a new campaign targeting the young star DoAr 44 (also known as V2062 Oph, Haro 1-16, ROXs 44, HBC 268) located at a distance of 146±1 pc (Gaia Collaboration et al. 2016Collaboration et al. , 2018 in the L1688 cloud of the Rho Ophiuchi molecular complex. This moderately bright source of spectral type K3 exhibits strong Hα emission with EW(Hα) 50Å (Bouvier & Appenzeller 1992) and accretes at a substantial rate from its circumstellar disk,Ṁ acc = 6.0-9.3 10 −9 M yr −1 (Espaillat et al. 2010;Manara et al. 2014). From the measurement of the Zeeman broadening of near-IR FeI lines, Lavail et al. (2017) derived a mean surface magnetic field strength amounting to 1.8±0.4 kG, and possibly up to 3.6 kG. These authors further derived a projected rotational velocity of v sin i=17.5±1.0 km s −1 , and estimated a mass of 1.25 M for this source.
The dedicated monitoring campaign of this source we present here combines, for the first time, high angular resolution ( 1 mas) long-baseline interferometry, optical and near-IR high-resolution spectropolarimetry, and multicolor photometry. DoAr 44 was selected for this campaign based on its brightness (V=12.6, K=7.6), which is amenable to both long-baseline interferometry and spectropolarimetry; the presence of a Brγ line in emission in its near-IR spectrum, with EW(Brγ) 4 Å measured on an archival ESO/XSHOOTER spectrum; and the recent report of its high surface magnetic field (Lavail et al. 2017). In addition, below we derive a photometric period of 2.96 d from a publicly available ASAS-SN light curve (see Section 3.2), which we ascribe to the stellar rotation period. These properties made it a suitable target for a dedicated large-scale monitoring campaign, with the additional interest of probing the magnetospheric accretion process in a pre-transitional disk system. This system featured a large dust-depleted cavity in its circumstellar disk, extending up to a distance of 30 au from the central star (Andrews et al. 2011). The goal of this campaign was to constrain the structure and dynamics of the magnetospheric accretion region between the inner edge of the compact inner disk and the central star by measuring the magnetic field strength at the stellar surface, reconstructing the geometry of the magnetospheric funnel flows and accretion shocks from the analysis of line profile variability and photometric variations, and, ultimately, attempting to resolve the interaction region over scales of a few stellar radii above the stellar surface using long-baseline interferometry. The VLTI/GRAVITY interferometric results are reported in an accompanying paper (Bouvier et al. 2020, hereafter Paper I). Section 2 describes the observations and data reduction, Section 3 features our results, and Section 4 offers a consistent interpretation of the observed variability of the system in the framework of the magnetospheric accretion model.

Observations
In this section, we describe the acquisition and data-reduction processes of photometric and spectropolarimetric datasets obtained during the large-scale campaign performed on the cTTS DoAr 44, which also included long-baseline near-IR interferometry. A summary plot of the main part of the DoAr 44 observing campaign is provided in Fig. 1.

LCOGT photomery
Photometric observations were obtained at the Las Cumbres Observatory Global Network (LCOGT, Brown et al. 2013) with the 1m Sinistro telescopes from June 4 to June 24, 2019. We used SDSS/PanSTARRS u'g'r'i' filters with exposure times of 60, 30, 10, and 5s, respectively. A total of 219 images centered on DoAr 44 were obtained over 20 days with a cadence of nearly three photometric sequences per 24 hours. Point-spread function (PSF) photometry was performed using IRAF/DAOPHOT on the pipe-line reduced images provided by LCOGT. Over the field of view (FOV) of 27 arcmin (see Figure 2), the PSF photometry includes the target DoAr 44, the nearby W Uma eclipsing binary V2394 Oph, the T Tauri stars ROX 42B, ROX 43, ROX 47B, and   (RA,Dec: 247.8519d,), referred to as C5 in the following. We verified that Oph 587 and C5, located within 10 arcmin of DoAr 44, were stable at a few 0.01 mag level in all filters. While the two TYC stars are brighter than Oph 587 and C5, and also appear to be photometrically stable at similar levels, they are located in a corner of the image, and are therefore more prone to systematics (e.g. flat-field correction).
We therefore used Oph 587 as a reference star and C5 as a check star to derive the differential light curves for DoAr 44. Stars Oph 587 and C5 are within one magnitude of difference from the target's brightness in all filters. The rms error of the photometric measurement in each filter amounts to about 0.02 mag. DoAr 44 is found to be variable in all four filters: the rms dispersion of the u'g'r'i' differential light curves (DoAr 44 -Oph

CFHT spectropolarimetry
Optical spectropolarimetric observations of DoAr 44 were completed during two runs, from March 15 to 22 and from June 8 to 12, 2019, with the ESPaDOnS spectropolarimeter at the Canada-France-Hawaii Telescope (CFHT), at a resolving power of 65,000 covering 370 to 1,000 nm (Donati 2003). During the first run, seven circularly polarized spectra were collected, and five additional spectra were obtained during the second run. Raw frames were reduced with the standard ESPaDOnS reduction package, and a least-squares deconvolution (LSD, Donati et al. 1997) was applied to all spectra, using a line list appropriate to DoAr 44. The journal of observations is presented in Table 2. Near-IR spectropolarimetric observations of DoAr 44 were completed from June 14 to 21, 2019 with the SPIRou spectropolarimeter at the CFHT, covering from 0.95 to 2.50 microns, at a spectral resolution of 70,000 ) at a rate of one we thus reach an angular resolution of λ/2Bmax = 1.7 mas, corresponding to 0.25 au at 146 pc. We used the instrument in its high spectral-resolution mode (R ∼ 4000), thus partly resolving the Brγ line profile. The observations, data reduction, and interferometric results are described in full in Paper I.

Stellar properties
We derived estimates of the system's properties based on the newly acquired high-resolution spectra. We followed the same procedure as we used in Pouilly et al. (2020) to which the reader is referred for details. We selected 16 spectral windows between 600 and 800 nm where the mean ESPaDOnS spectrum of DoAr 44 is fit with a synthetic spectrum generated from the ZEEMAN code (Folsom et al. 2012) based on MARCS stellar atmosphere grids (Gustafsson et al. 2008) and VALD line lists (Ryabchikova et al. 2015). Assuming log g= 4.0, we derived the following parameters: v sin i= 17.0 ± 1.1 km/s, V r = -5.70 ± 0.24 km/s, T eff = 4600 ± 120 K, and V mic = 0.27 ± 0.27 km s −1 . The veiling, that is to say, the ratio of the continuum excess flux to the stellar photospheric flux, is measured as constant over time and amounts to r 0.18±0.05 over the wavelength range 610-670 nm.
Using the optical photometry of Bouvier et al. (1988) and 2MASS J-band, and adopting a K2-K3 spectral type with intrinsic colors listed in Pecaut & Mamajek (2013), we derived color excesses in the (V-I c ) and (V-J) colors that correspond to a visual extinction A V =2.0±0.2 mag. From the Jband magnitude, bolometric corrections from Pecaut & Mamajek (2013) for K2-K3 young stars, and a distance of 146±1 pc Fig. 3. DoAr 44 photometric variations. The g'r'i' light curves (green, blue, and red, respectively) are shown folded in phase with a period of 2.960 days derived from periodogram analysis (see text). A sine curve was eye fit to the light curves and is shown for illustration with a peakto-peak amplitude of 0.08 mag in the g'r'i' bands. (Gaia Collaboration et al. 2016, we thus obtained L =1.6±0.2 L . Combined with T eff =4600 K derived above, this yields R =2.0±0.15 R . From its position in the HR diagram, we derived M =1.23±0.15 M using the CESAM evolutionary models (Marques et al. 2013), and the star appears to have a modest radiative core, with M rad /M =0.22±0.20 and R rad /R =0.35±0.2, where M rad and R rad are the mass and radius of the radiative core, respectively. We note, however, that the latter values are strongly model dependent, as, using Siess et al. (2000) models, we would derive M =1.4 M , M rad /M =0.73 and R rad /R =0.50.
Finally, we used the above estimate of the stellar radius to derive the inclination of the stellar rotational axis on the line of sight (LoS). Using v sin i=17 km s −1 and P rot =2.96 d (see below), and propagating the uncertainties, yields sin i=0.50±0.05, which translates to i=30±5 • , meaning the system is seen at low inclination. Table 4 summarizes the properties of the stellar system with their uncertainties.

Photometric variability
The LCOGT observations yielded 20 day-long differential light curves for DoAr 44 in the u'g'r'i' filters. We searched for a periodicity in the photometric variations, excluding about five measurements per filter taken when the bright moon was under 15 degrees away from the target. We used both the CLEAN periodogram analysis (Roberts et al. 1987) and the string-length method (Dworetsky 1983). The former yields a period of 3.09 d in the r' filter only, while the latter returns a most probable period in all filters in the range from 2.97 d to 3.07 d. In order to confirm the period, we downloaded ASAS-SN photometry (Shappee et al. 2014), which contains 288 single filter measurements over a single season extending from MJD 57780 to MJD 58030 (February-September 2017). Using the same periodsearch algorithms, we found a clear periodicity of 2.960 +/-0.018 d in this much richer dataset, with a peak-to-peak amplitude of about 0.1 mag. The LCOGT g'r'i' light curves folded in phase with this period are shown in Fig. 3 and suggest a lowlevel modulation by surface spots, of the order of a 0.08 mag peak to peak in the g'r'i' filters. While a period is found in the u'-band light curve as well, the phased light curve exhibits a scatter of the same order as the photometric amplitude. We therefore adopted the more accurate 2.960 d period derived from the Fig. 4. From left to right: mean (black) and variance (red) residual profiles for Hα, Hβ, and HeI 587.6 nm. Hα and Hβ appear on overlapping spectral orders, which are both shown here. For HeI, the variance has been multiplied by 10. The increased scatter over the blue wing of the HeI line profile is due to spectral order overlap.  -5.58 39.6 6.1 0.77 † The rms uncertainty on V r amounts to 0.05 km/s. ASAS-SN dataset and ascribed it to the rotational period of the star. We thus define an ephemeris such that the origin of phase corresponds to the epoch of maximum brightness at the time of our observations, so Finally, we note that the uncertainty on the derived period, δP=0.018d, could induce a phase shift between the March and June ESPaDOnS runs up to δΦ=0.16 rms.

Optical line profiles
Deciphering the shape and variability of the emission line profiles seen in the spectra of cTTSs offers a window to the physical processes occurring close to the star, particularly the structure and dynamics of accretion and outflows (e.g., Alencar & Basri 2000;Edwards et al. 2006;Lima et al. 2010;Esau et al. 2014     H&K doublet and near-IR triplet, as well as in the HeI 587.6 nm, 667.8 nm, and 706.6 nm, [OI] 630.0 nm, and OI 777.3 nm lines. Residual emission line profiles were computed by subtracting the broadened photospheric spectrum of the Hyades cluster member Melotte 25-151, a K2-type star with V r = 37.98 km s −1 , v sin i= 4.83 km s −1 , and T eff = 4920 K (Folsom et al. 2018), also observed with ESPaDOnS.

Line profile variability
The spectral series we obtained on DoAr 44 with CFHT/ESPaDOnS allows us to investigate the spectral variability of the source on timescales ranging from days to months. Figure 4 shows the mean and variance profile for the main emission lines. The Balmer line profiles are double peaked with a slightly depressed red peak, corresponding to the Type II(R) profiles in the classification of Reipurth et al. (1996). This profile shape is qualitatively accounted for by dipolar magnetospheric accretion models considering moderate to low inclinations, i 30 − 45 • (Muzerolle et al. 1998(Muzerolle et al. , 2001. Both the Hα and Hβ profiles exhibit significant variance in the blue and red wings up to velocities of a few hundred km s −1 , and comparatively less variability in the line center. As is often observed in T Tauri stars (Beristain et al. 2001), the HeI line profile is composite, consisting of a centrally peaked broad component, extending up to velocities of ±200 km s −1 , and a narrow component located close to the line center with a FWHM of about 40 km s −1 . The profile's variability seems to be predominantly associated with the narrow component. According to Beristain et al. (2001), the properties of this type of HeI profile suggest the narrow component arises in the post-shock region close to the stellar surface, while the broad component is formed in the main accretion column above the accretion shock.  Figure 6 is another representation of the same dataset, ordered by Julian date and by rotational phase. Similar plots are shown for the HeI 587.6 nm line in Fig. 7. We note that the Balmer line profiles appear more symmetric in June than in March, while the HeI line profile has remained the same, and it exhibits a broad component and a narrow component at both epochs. Remarkably, the Hα and Hβ profiles do no exhibit high-velocity redshifted absorptions below the continuum at any rotational phase, contrary to what is often seen in other accreting sources as the funnel flow intercepts the LoS (e.g., Bouvier et al. 2007a;Alencar et al. 2012Alencar et al. , 2018. The lack of inverse P Cygni profile signatures provides further support for a low inclination of the system (Muzerolle et al. 2001), as independently derived for the stellar rotational axis and for the outer disk that both suggest i 20-35 • (see above and Paper I).
Equivalent width measurements for Hα, Hβ, and HeI emission lines are reported in Table 5. As the continuum is well defined around these lines, the equivalent width measurement is accurate to within a few percent. The median equivalent widths over the two epochs amount to 44, 6.8, and 0.8 Å, for Hα, Hβ, and HeI lines, respectively. The lines have slightly larger median equivalent widths in June than in March. Their variation as a function of rotational phase is shown in Fig. 8. There is a hint of rotational modulation in the equivalent width variations of the three line profiles during the June run, with minimum values close to a rotational phase of 0.5, while no such modulation is seen in the March run. We estimate the mass accretion rate onto the star from the flux measured in the Hα and Hβ lines, using relationships between accretion and line luminosity from Alcalá et al. (2014). From the Hα line we thus derive L acc /L =0.06-0.13, taking into account EW(Hα) variations, which yieldsṀ acc =4.0-8.5 10 −9 M yr −1 assuming a magnetospheric truncation radius of 5 R . Similarly, from the Hβ line, we obtain L acc /L =0.05-0.17 andṀ acc =3-10 10 −9 M yr −1 . These estimates are consistent with those reported by Manara et al. (2014), who obtainedṀ acc =6.3 10 −9 M yr −1 from slab models applied to XSHOOTER spectra.
A periodogram analysis of the line profiles is detailed in Appendix A. The analysis does not reveal any significant periodicity in the intensity variations of the velocity channels across the Hα, Hβ, and HeI line profiles when the full dataset is analyzed. This presumably results from the large temporal gap between the March and June ESPaDOnS runs. As described in Appendix A, when each run is analyzed independently, we do find some evidence for the rotational modulation of the Hα and Hβ line profiles at the stellar rotation period. More specifically, the intensity of the red wing (from about +50 to +150 km s −1 ) of the Hα and Hβ profiles appears to be modulated at the stellar rotation period during the March run. Since the veiling is nearly constant at these dates, we ascribe these variations to changes in the intrinsic line flux, as opposed to continuum variations. This result is of limited significance, due to the poor rotational phase sampling and the low number of spectra obtained during each run. Nevertheless, the examination of the shape of the line profiles plotted as a function of rotational phase in Fig. A.1 tends to support rotational modulation at both epochs, with a systematically depressed line flux in the red part of the profile around phase 0.5 for Balmer lines, as expected from accretion funnel flows as they cross the LoS (e.g., Symington et al. 2005).

Line profile decomposition
We further analyzed the line profiles attempting to reproduce their shape as a combination of Gaussian components. Details on the line profile decomposition are given in Appendix B, and a few examples are illustrated in Fig. B.1. The Balmer line profiles are well fit with three Gaussian components, consisting of a broad central emission peak and two absorption components, one blueshifted, and the other redshifted, relative to the line center. The HeI line profile can be reproduced using only two emission components, a broad and a narrow one. The results are given in Table B.1, which lists the equivalent width, peak intensity, radial velocity, and FWHM of each component.
The narrow component of the HeI line profile shows significant radial velocity variations between +4.4 and +11.4 km s −1 in the stellar rest frame, while its FWHM varies over the range 38-47 km s −1 . We find these variations to be modulated by stellar rotation with a strong phase coherence between the March and June ESPaDOnS runs (see Figure 9). Assuming that the HeI line narrow component arises from the accretion post-shock (Beristain et al. 2001), as suggested by its slightly redshifted velocity, and covers a small area on the stellar surface, one can relate the radial velocity amplitude δV r to the hot spot latitude θ, with δV r = 2·v sin i· cos θ (Bouvier et al. 2007a). We deduce a latitude of 78±2 • for the accretion shock at the stellar surface, assuming an uncertainty of 0.5 km s −1 on δV r .
Interestingly, the radial velocity of the wide component of the HeI line profile seems to be rotationally modulated as well, albeit with increased scatter. If this component arises from a hot wind close to the star, the modulation may result from the geometric projection of the wind velocity pattern onto the LoS. However, its slightly redshifted velocity relative to the star instead suggests it arises from the accretion funnel flow, located above the post-shock region probed by the narrow component (Beristain et al. 2001). The FWHM of the narrow and wide components appears to be rotationally modulated as well. They reach a maximum at phase 0.5 of 259 km s −1 for the broad component and 47 km s −1 for the narrow one, and both vary by about 15% along the rotational cycle. This variation could conceivably be related to the projection on the LoS of the velocity gradients of the emitting regions responsible for the narrow and broad HeI components. The vastly different FWHM of the narrow and broad components, and the fact that the radial velocities of both components appear to be modulated but reach a maximum value at different rotational phases (see Fig. 9), suggest the two components probe different regions along the funnel flow.
Considering the two absorption components in the Hα profile, one being blueshifted, the other redshifted, we find that their FWHMs are rotationally modulated and strongly anti-correlated (see Table B.1). The equivalent width and peak absorption intensity of the blueshifted component are the largest around phase 0, while the equivalent width of the redshifted absorption component is the largest around phase 0.5. As the former is probably related to an outflow, while the latter probes accretion, this suggests that both flows are physically related and occur at opposite rotational phases. We also notice a hint of a correlation between the radial velocity variations of these two absorption components in the Hα line profile, as reported earlier for AA Tau (Bouvier et al. 2003) and LkCa 15 (Alencar et al. 2018).
As illustrated in Fig. B.1, the decomposition of the CaII 854.2 nm emission line reveals the presence of a highly redshifted absorption component during the March run, around V r +120 km s −1 . It is observed at phases 0.83 (JD 8559.15) and 2.83 (JD 8565.07), but, surprisingly enough, not at phase 1.83 (JD 8562.12). Similar high-velocity redshifted absorptions are also seen in the NaI D line profiles at the same phases. This suggests transient accretion episodes onto the stellar surface or, at least, mass accretion rate variations occur within the funnel

Correlation matrices
The observed emission lines of DoAr 44 are highly variable, with emission and absorption components superposed, as can be seen in Fig 5. The different components may come from various circumstellar regions, like the chromosphere, the accretion funnel, and the wind, which can vary over different timescales. To investigate how the variations across a line or between different lines are related, we calculated correlation matrices consisting of 2D plots of the linear correlation coefficients of the line profile intensity variations at different velocity bins. A good correlation between two components indicates a common origin of the components, for example.
The Hα and Hβ lines are similar, and their correlation matrices also show the same characteristics (Fig. 10). There is little correlation between the blue and red part of the profile, which suggests different physical processes at the origin of their variability. The red wing of the profile varies coherently beyond +200 km s −1 and is strongly anti-correlated with the red emission peak around 100 km s −1 . This red emission peak is strongly variable, as can be clearly seen in Figs. 5 and 6. As the velocity channels corresponding to the peak appear to be modulated at the stellar rotation period, we interpret its anti-correlation with the profile's far red wing as the signature of periodic redshifted absorptions due to the funnel flow crossing the LoS.
The HeI line is composed of a broad and a narrow component. The correlation matrix is shown in Fig. 10. It indicates that the BC varies coherently and independently of the NC component. The Hα vs. HeI cross correlation matrix reveals that the HeI BC component is correlated with the high velocity wings of the Balmer profile beyond 200 km s −1 . The red wing of the HeI BC is, however, strongly anti-correlated with the red emission peak A&A proofs: manuscript no. 38892corr seen in the Hα profile around 100 km s −1 . This suggests that redshifted absorption in the Hα profile occurs as the BC of the HeI line profile becomes stronger, which is consistent with both the redshifted Hα absorption and the HeI BC emission originating from an accretion funnel flow crossing the LoS.

Infrared line profiles
The wide spectral range covered by SPIRou includes a number of circumstellar diagnostics, such as HeI 1083.3 nm, Paβ 1282.2 nm, and Brγ 2166.1 nm. These lines are seen in emission in all the spectra acquired during the SPIRou observing run, and their equivalent widths are listed in Table 6. From the equivalent widths of the Paβ and Brγ line profiles measured on the mean spectrum, EW(Paβ) = 8.75 ± 0.15 Å and EW(Brγ) = 3.5 ± 0.2 Å, and using Alcalá et al. (2014) relationship between line and accretion luminosities, we derived a mass accretion rateṀ acc = 6.5 ± 0.5 10 −9 M yr −1 , which is consistent with the estimate obtained above from the luminosity of the optical emission lines. Residual line profiles were computed following the same procedure as described above for optical spectra, using the weakline T Tauri star V819 Tau (SpT K4) as a template, also observed with SPIRou. Line profiles are shown in Figure 11. The HeI profile exhibits a strongly asymmetric shape, with significant blueshifted absorption components reaching down to velocities of -350 km s −1 with a depth that varies on a timescale of days, while the red wing appears in emission and also exhibits significant variability. In contrast, the Paβ and Brγ profiles are much more symmetric, with only low-level variability in intensity over the course of the observations. The profiles are plotted as a function of date and rotational phase in Fig. 12.
We performed a 2D periodogram analysis across the line profiles. The 2D periodogram of the Paβ line profile is shown in Fig. 13. A strong signal appears at the star's rotational period in the redshifted wing of the profile over a velocity range extending from +100 to +200 km s −1 . Radiative transfer models of magnetospheric accretion onto an inclined dipole (e.g., Kurosawa et al. 2008) predict such a rotationally modulated intensity of the high velocity redshifted wing of the Paβ profile, as funnel flows periodically cross the LoS. A strong signal is also seen in the 2D periodogram from the line center up to +100 km s −1 at frequencies around 0.9 d −1 , with its alias at 0.1 d −1 . The former frequency is close to the sampling rate, and the latter is beyond the observation time span. The signal at these frequencies is therefore probably spurious, being induced by the observing window. Nevertheless, the variability seen in this part of the Paβ profile is real, and is also seen in the Brγ line profile over the same velocity channels, where the line intensity seems to first increase then decrease smoothly over the time span of the observations (see Fig. 11).
The 2D periodogram of the HeI line profile displays the strongest signal at a frequency corresponding to the stellar rotation period, extending over redshifted velocity channels from +30 to +100 km s −1 (see Fig. 13). This periodic variability of the red part of the line profile is most likely induced by the rotational modulation of accretion funnel flow as the star rotates (e.g., Fischer et al. 2008). The other parts of the HeI profile are extremely variable, especially around the line center and throughout the blueshifted wing down to velocities of -350 km s −1 . The blue wing exhibits a rapidly changing blueshifted absorption component sometimes reaching below the continuum. Yet, this outflow component shows no sign of modulation. Instead, we observe significant change in HeI line profiles taken at the same rotational phase over successive cycles, for example, at φ=31.14, 32.16, 33.16, and at φ=31.48, 32.49 (see Fig. 12), which suggests intrinsic variations of the outflow component over a timescale of days. The velocity of the peak of the blueshifted absorption component appears to be the most extreme at the beginning of the observations (v blue -200 km s −1 ), shifting toward the line center up to v blue -60 km s −1 within three days, then returning to much bluer channels (v blue -130 km s −1 ) over the same timescale. Its width changes as well, the most blueshifted absorption having the largest FWHM.
Radiative transfer models assign the presence of moderately blueshifted, narrow absorption components in the profile of the HeI 1083.3 nm line to a disk wind (e.g., Kurosawa et al. 2011), while broader absorptions extending to large blueshifted velocities are usually considered as the signature of accretion powered, hot inner winds originating close to the star (e.g., Edwards et al. 2003Edwards et al. , 2006. We are apparently witnessing here a mix of disk and "stellar" winds, with a predominance of the former except for the first spectrum of the series, whose relative strength varies on a timescale of days. Indeed, the gallery of emission line profiles gathered from the quasi-simultaneous optical and near-IR high-resolution spectra obtained for DoAr 44 during this campaign, including HeI 587.6 nm, HeI 1083.3 nm, Hα, Hβ, Paβ, and Brγ, bears strong resemblance with synthetic line profiles derived from the combined magnetospheric accretion and disk wind, low-inclination (i=20 • ) model presented in Fig.7 of Kurosawa et al. (2011). It is therefore likely that most of the emission seen in these profiles comes from magnetospheric funnel flows controlled by a misaligned dipole, with additional wind components best seen in the HeI near-IR line profile. The qualitative agreement between observations and models supports this interpretation. The low inclination of the system is further supported by the maximum edge velocity at which the blueshifted HeI absorption component meets the continuum, which amounts to -350 km s −1 . According to the edge velocity versus the system's inclination relationship reported for this component by Appenzeller & Bertout (2013), here the system's inclination should lie in the range 20-35 • .
The auto-correlation matrices of the three profiles are shown in Fig. 14 and share the same behavior: different parts of the line profile vary independently of each other. In the Paβ line for instance, the blue wing varies independently of both the lowvelocity and high-velocity redshifted channels, and the latter also varies independently. The cross-correlation matrix between Paβ and Brγ reveals the same three components: the blue wing, the low-velocity red channels, and the high-velocity red channels. While each of these components varies independently of     line profiles computed from the eight SPIRou spectra obtained on consecutive nights. The color code represents the periodogram power, and the superimposed white curve is the average line profile. We note the symmetry of the periodogram around the frequency 0.5 d −1 due to the one-day alias resulting from the night-to-night sampling.
the other, they are well correlated between the Paβ and Brγ line profiles. A more complex behavior is seen in the cross-correlation matrices of HeI with Paβ and Brγ. The high-velocity blue wing of the HeI line is clearly correlated with that of the hydrogen lines. More surprisingly perhaps, there is a hint of a correlation, albeit relatively weak, between the high-velocity redshifted wing of HeI and the blueshifted wing of the Paβ profile. Another feature is the apparent anti-correlation between the low-velocity redshifted HeI channels and the high-velocity redshifted hydrogen wings. This occurs over the interval of velocities where both profiles are rotationally modulated, namely around 30-100 km s −1 for HeI and beyond 100 km s −1 for Paβ. If real, this anti-correlation is somewhat surprising as the redshifted absorptions in both profiles, though appearing at different velocities, are expected to occur simultaneously at the time the funnel flow crosses the LoS. This feature is also present in the autocorrelation matrices of HeI and Paβ. In fact, the HeI vs. Paβ matrix closely mimics the structures seen in the HeI autocorrelation matrix. While these features are intriguing, the value of the correlation coefficient remains of the order of 0.5 or less, Article number, page 9 of 17 A&A proofs: manuscript no. 38892corr and would therefore have to be confirmed from a longer time series.

Zeeman-Doppler analysis
The polarimetric capabilities of CFHT/ESPaDOnS allow us to investigate the large-scale structure of the magnetic field at the surface of the star. Figure 15 shows the least-square deconvolution ((LSD) Donati et al. 1997) Stokes I and Stokes V profiles over the two ESPaDOnS runs, computed using a mean Landé factor of 1.2 and a central wavelength of 500 nm. We did not use the spectrum obtained on JD 8559.15 (φ=0.83) for the ZDI analysis as the LSD profiles show contamination by the morning twilight. A clear and relatively stable magnetic signature is present in all Stokes V profiles. The variation of the longitudinal component of the large scale magnetic field, B l , derived from the Stokes I and V signatures (Donati et al. 1997;Wade et al. 2000), are shown in Fig. 17. It varies between -14 and -136 Gauss (see Table 5) and appears to be modulated by stellar rotation. Remarkably, the magnetic field intensity does not exhibit significant change between the March and June runs, nor does the phase of its modulation. This suggests a fairly stable magnetic configuration between the two epochs.
A similar analysis has been performed in the CaII near-IR triplet. The three lines of the CaII triplet exhibit consistent magnetic signatures in their Stokes V profiles as shown in Fig. 16. The longitudinal component of the magnetic field projected onto the LoS is still modulated at the stellar rotation period (see Fig. 17) and now reaches a maximum value of 800 G, which is about four times larger than the intensity deduced from the LSD profiles, as is often the case for young accreting stars (e.g., Donati et al. 2019). It is worth mentioning here that the intensity of B l , as well as its sign, are widely different when using LSD (photospheric) profiles and the CaII IRT (chromospheric) lines. Indeed, these spectral features probe different regions at the stellar surface. The Stokes V signal in the optical HeI line was also analyzed, as it provides the most direct estimate of the magnetic field strength at the base of the funnel flow, close to the accretion shock. Unfortunately, the Stokes V profile in the narrow component of the line is weak (see Fig. 15), which translates into large uncertainties in the derivation of the longitudinal field strength. Figure 17 shows the absence of modulation within error bars, with an average field strength of 2±0.8 kG. The lack of significant modulation may result from the low inclination of the system, where the high latitude accretion shock remains in view during the whole rotational cycle.
We used the Doppler imaging (DI) and Zeeman-Doppler imaging (ZDI) implementation of Folsom et al. (2018) to respectively map the surface brightness and magnetic field of DoAr 44. A full description of the code can be found in Folsom et al. (2018). Briefly, the code uses a time series of Stokes I and V LSD profiles to construct a surface brightness map and the largescale vector magnetic field. The local line models are calculated using a Voigt profile, with a Gaussian width of 1.2 km s −1 and a Lorentzian width of 1.9 km s −1 . The local line models for Stokes V are calculated from the local I line profile, using the weak-field approximation (Landi Degl'Innocenti & Landolfi 2004), and the LSD mean Landé factor and central wavelength. The magnetic field is expressed as a combination of spherical harmonics as in Donati et al. (2006). The integration over the surface of the star is done using a linear limb-darkening coefficient of 0.8 (Claret & Bloemen 2011), adapted to the temperature and gravity of DoAr 44. The fit to the data is performed by both minimizing χ 2 and maximizing the entropy, using the maximum entropy fit-ting routine of Skilling & Bryan (1984). We have adopted the photometric ephemeris defined above, and fixed the inclination of the rotation axis at 35 • , which is consistent with the inclination of the inner disk seen in interferometry on a scale of 0.14 au (see Paper I).
The resulting brightness and magnetic maps are shown in Fig. 18. One large cool spot is observed on the brightness map, located close to the rotational axis and facing the observer around phase 0.83. Two bright spots are present at lower latitudes, facing the observer around phases 0.30 and 0.70. Due to the low inclination of the system, little is known of the southern hemisphere of the star. We used this brightness map to perform ZDI and reconstruct the magnetic map. We fixed the maximum value of the Legendre polynomial degree l to 15 in the spherical harmonics decomposition. The resulting distribution of the magnetic vectors over the stellar surface is shown in Fig. 18. It indicates a dominant dipolar component, encompassing 75% of the total flux, while only 8% and 4% of the flux is contained in the quadrupolar and octupolar components, respectively. The negative pole of the dipole is facing us at the phase 0.5, which is consistent with the maximum absolute value of the longitudinal field. It is located at the latitude of 70 • , and thus inclined by 20 • onto the rotational axis.

Discussion
The variability analysis of the DoAr 44 system is complicated by its nearly integer rotational period of 2.96 d. While we observed the system over several rotational periods, the resulting phase coverage remains limited, with essentially the same three phases being sampled over the observations, at every third of the rotational cycle. This prevents us from having a complete picture of the circumstellar environment of this young stellar object, and also severely limits the longitudinal resolution we have on the stellar surface for image reconstruction. With the data at hand we are nevertheless able to derive the main physical and geometrical properties of the system in the region where the stellar magnetic field interacts with the inner disk.
The combination of the rotational period derived above and the v sin i we measure suggests that the star is seen at a low inclination of 30±5 • . In Paper I, we derived a similar inclination of 34±2 • for the inner edge of the circumstellar disk from long baseline near-IR interferometry. That the inner system seen at low inclination is further supported by the lack of strong photometric modulation. While we do recover a photometric period of 2.96 d, assigned to the stellar rotational period, the photometric amplitude of ≤0.1 mag in the V-band is significantly lower than that of most accreting T Tauri stars (e.g., Grankin et al. 2007). The photospheric radial velocity is modulated over the same 2.96 d period (see Fig. 19), which confirms the presence of a long-lived, large, dark spot at the stellar surface. According to the Doppler image, the high-latitude spot faces the observer at phase 0.83, which yields symmetric LSD profiles (see Fig. 15). Indeed, the photospheric radial velocity lies midway from the extrema of the V r curve at this phase, as expected when the spot is on the LoS (see Fig. 19). These concordant results confirm that the 2.96 d period is the rotational period of the star whose surface, seen at a low inclination, is covered by a high-latitude, large, cold spot. In contrast, the photometric minimum appears to occur at phase 0.5, which suggests a more complex distribution of bright and dark spots at the stellar surface drives the photometric modulation.
In spite of the low inclination, clear modulation signals are also detected in several accretion diagnostics. This is the case of Table 7. Longitudinal magnetic field strength (B l ) expressed in Gauss and computed from photospheric Stokes LSD profiles, from each line of the CaII IR triplet (CaII 1,2,3 ), and from the HeI 587.6 nm line appearing on two successive spectral orders (HeI 1,2   the redshifted wings of emission lines formed in the funnel flow (e.g., Hβ, Hα, Paβ, HeI 1083.3 nm) that smoothly vary along the rotational cycle, even though they do not exhibit inverse P Cygni components, that is, absorptions reaching below the continuum, which are the hallmark of accretion funnel flow passing onto the LoS in high-inclination systems. The radial velocity and the intensity of the narrow component of the HeI 587.6 nm line profile are also modulated at the stellar rotational period and indicate a high-latitude hot spot at the stellar surface, presumably the location of the accretion shock close to the magnetic pole. Indeed, the magnetic surface map reconstructed from the ZDI analysis features a dominant dipolar field whose axis is moderately inclined onto the rotational axis, thus producing an accretion shock at high latitudes. These results suggest that the inner disk is truncated at some distance above the stellar surface and that the accretion onto the star proceeds along magnetospheric funnel flows. If we assume that the disk's truncation radius lies close to the disk's corotation radius, we get r tr r co = (GM ) 1/3 (P rot /2π) 2/3 = 0.043 au = 9.2 R = 4.6 R , and according to Bessolaz et al. (2008), the truncation radius occurs at r tr /R = 2 · B 4/7 Ṁ −2/7 acc M −1/7 R 5/7 , where the stellar field at the equator B is normalized to 140 G, the mass accretion rateṀ acc to 10 −8 M yr −1 , the stellar mass M to 0.8 M , and the stellar radius R to 2 R .
Thus, the condition r tr r co requires a dipolar magnetic field of the order of 537 G at the equator, which translates into 1.1 kG at the pole. The average value of the longitudinal component of the magnetic field we measure in the narrow component of the HeI 587.6 nm line amounts to 2±0.8 kG. Assuming this is the large scale component that interacts with the disk, we would derive a truncation radius that is about twice the corotation radius. However, the large error bars associated with the field strength and derived from the HeI NC component are not inconsistent with the disk being truncated close to the corotation radius. In Paper I, we measured the size of the Brγ line emitting region, thought to reflect the extent of the magnetospheric cavity, from long baseline near-IR interferometry, and we found it to be less than 0.047 au, that is, 5 R . All the results summarized above suggest that DoAr 44 was observed during a phase of stable magnetospheric accretion with a magnetic topology that shares some similarities with other young accreting systems that we previously studied, such as V2129 Oph (Alencar et al. 2012). Some of these systems exhibit dipper light curves at times, with deep periodic eclipses of the central object being produced by circumstellar dust: for example, AA Tau (Bouvier et al. 2003(Bouvier et al. , 2007a and LkCa 15 (Alencar et al. 2018), as well as periodic redshifted absorptions reaching below the continuum appearing in the wings of emission lines formed in the funnel flow. These specific properties are not shared by DoAr 44 in spite of a qualitatively similar magnetospheric structure. We estimate that the dust sublimation radius lies at a distance of R sub = 4.7 R = 0.044 au, as derived from Eq.1 of Monnier & Millan-Gabet (2002) for T sub = 1500 K and Qr=1. Dust might thus to be present at the disk truncation radius, even though the half-flux radius of the inner dusty disk we report in Paper I is of the order of 0.14 au. While a photometric modulation is observed, the light curve is spot dominated, with no evidence for significant dips. Similarly, while the red wing of Balmer line profiles appear modulated at the stellar rotation period, there is no sign of high-velocity redshifted absorptions reaching below the continuum, even at phases where the magnetic pole is facing the observer.
Indeed, dipper characteristics and conspicuous inverse P Cygni profiles are expected to occur primarily in systems seen at a moderate to high inclination (e.g., McGinnis et al. 2015;Sousa et al. 2016;Fonseca et al. 2014). This is clearly not the case in DoAr 44, and the low inclination of this system accounts for the low-amplitude, smooth light curve; the mildly varying red wings of the emission line profiles; and more generally, the global shape of the emission line profiles as suggested by radiative transfer models of magnetospheric accretion onto a misaligned dipole. This clearly highlights the importance of taking into account inclination when investigating the variability of young stellar systems. Systems with otherwise similar properties, notably regarding the magnetospheric accretion process, might exhibit quite different characteristics depending on the viewing angle under which they are seen.
DoAr 44 is a pre-transitional disk system. On the large scale, from mm observations, Andrews et al. (2009Andrews et al. ( , 2011 and van der Marel et al. (2016) reported the existence of a large cavity extending from the central star up to 25-30 au in the circumstellar disk. They showed that cold dust emission arises from an outer, relatively narrow ring (25 to 35 au wide), inclined by 20 to 35 • on the LoS. On much smaller scales, the existence of a compact, inner disk close to the star is supported by the near-infrared excess flux measured in the spectral energy distribution of the system (Espaillat et al. 2010) and significant CO emission arising from the inner disk edge (Salyk et al. 2009(Salyk et al. , 2011. Furthermore, a misalignment between the inner and outer disk was assumed to be responsible for shadows seen in the outer dusty ring (Casassus et al. 2018). In Paper I, we reported the direct detection of the inner disk from long baseline near-IR interferometry. The halfflux radius is 0.14 au, and the compact disk is indeed found to be inclined relative to the outer ring, as predicted by the Casassus et al. (2018) model to account for the outer shadows. The pretransitional status of DoAr 44's disk was a strong motivation to include this specific target in our long-term monitoring program of T Tauri stars. The goal here was to investigate whether the highly structured disk, including a large cavity and inner-outer disk misalignment, both possibly betraying ongoing planetary formation, may impact the accretion process on the central star. All the results obtained here suggest that, at the star-disk interaction scale of less than 0.1 au, this pre-transitional disk system is not qualitatively different from other young stellar objects surrounded by smoother protoplanetary disks.

Conclusion
We monitored the variability of the young stellar object, DoAr 44, over several rotational periods using several complementary observational techniques to investigate the physics of accretion onto the star in a pre-transitional disk system. The specific geometry of pre-transitional disks, including large cavities and possible inner-or outer-disk misalignment, raises issues regarding the accretion process in these systems. How does accretion proceed from the outer disk to the inner scales through the gap? Is the star-disk interaction process fed in the same way or differently . The color code corresponds to successive rotational cycles. We emphasize the phase consistency between the March and June datasets, as well as the apparently constant amplitude of the field between the two epochs. In the top panel, the sine curve has been fit with the period and origin of phase of the photometric ephemeris. It shows that the maximum intensity of the photospheric field occurs at phase 0.5, that is, at the time of minimum brightness (see Fig. 3). In the middle panel, the three measurements at each phase are obtained for each line of the CaII IR triplet. In the lower panel, there are two measurements per spectrum, as the HeI line appears on successive spectral orders. The three plots are not on the same scale. from what is observed in classical T Tauri stars with continuous circumstellar disks? If the pre-transitional nature of disks is the signature of ongoing planet formation in the disk, does this process affect accretion on any scale?
In an attempt to link the outer scales, best studied with mid-IR and radio interferometry, to the inner scales, we focused here on the characterization of the DoAr 44 system at the star-disk interaction level, meaning on a scale extending from the stellar surface up to a few 0.1 au, combining photometric and spectropo- larimetric monitoring with long baseline interferometry. The accretion diagnostics studied here and the direct interferometric measurements we reported in Paper I  suggest the existence of an inner magnetospheric cavity extending from a distance of about 4.6 stellar radii (0.043 au) down to the stellar surface. A large-scale, mainly dipolar stellar magnetic field of a few hundred Gauss over the photosphere, but reaching up to 2±0.8 kG close to the accretion shock, is sufficient to disrupt the inner-disk region at the corotation radius or beyond in this moderately accreting system. The reported line profile variability is consistent with the expected behavior of funnel flows rotating at the stellar period. Yet, due to the low inclination of the inner system, of the order of 30 • , the modulation signatures are not as pronounced as in other, more inclined systems. Importantly, the combination of contemporaneous optical and near-IR high-resolution spectroscopic monitoring provides us with a unique means to investigate diagnostics that simultaneously probe accretion funnel flows and stellar or disk winds.
Overall, over our three-week observation period, DoAr 44 appeared to be in a state of stable magnetospheric accretion, simultaneously driving outflows, in most aspects similar to what we had previously witnessed in other systems we monitored in the same way. This suggests that the physics of accretion from the inner disk to the stellar surface is not significantly altered by the complex large-scale structure of the circumstellar disk specific to pre-transitional systems.
Since the shape of the emisson line profiles has varied between the ESPaDOnS March and June 2019, runs, we present in Fig. A.1   The periodogram analysis of line profiles over the two ES-PaDOnS runs is inconclusive. We therefore proceed to analyzing the March and June 2019 runs independently. Figures A.2 and A.3 show the periodogram analysis of the Hα, Hβ, and HeI line profiles for the two epochs, respectively. During the March run, seven spectra were obtained over eight nights, while during the June run, five spectra were obtained over five nights. Moreover, since the stellar period of 2.96 d is close to an integer, the same rotational phase repeats every third night. The short duration of each run combined with the poor phase sampling makes the periodogram analysis fragile. Nevertheless, we find marginal evidence for the rotational frequency of 0.34 d −1 in the red wing of the Hα and Hβ line profiles around +100 km s −1 during the March run, and in the narrow central component of the HeI line during the June run. While of limited significance, the similar shapes exhibited by the various line profiles at specific rotational phases (see Fig. A.1) lends credit to this result. For instance, the Balmer line profiles appear much more symmetric around phases 0.12-0.16 and 0.83-0.93 than around phases 0.46-0.51, where the red wing is consistently depressed. This suggests additional redshifted absorption occurs at phases around 0.5.

Appendix B: Line profile decomposition
Here, we illustrate examples of line profile decomposition by multiple Gaussian fits for the Hα, Hβ, HeI, and CaII emission line profiles observed on JD 8565.07 and JD8644.89. The Hα Fig. A.2. 2D periodogram of Hα, Hβ, and HeI line profiles from the seven spectra obtained during the ESPaDOnS March run. The color code represents the periodogram power, and the superimposed white curve is the average line profile. We note the symmetry of the periodogram around the frequency 0.5 d −1 due to the 1 d alias resulting from the night-to-night sampling. Fig. A.3. 2D periodogram of Hα, Hβ, and HeI line profiles from the five spectra obtained during the ESPaDOnS June run. The color code represents the periodogram power, and the superimposed white curve is the average line profile. We note the symmetry of the periodogram around the frequency 0.5 d −1 due to the 1 d alias resulting from the night-to-night sampling. and Hβ line profiles are well reproduced with a central Gaussian emission peak onto which a blueshifted and a redshifted absorption components are superimposed. The HeI line is well fit by the combination of two Gaussian components in emission: a broad one and a narrow one, respectively. This is also the case for the CaII line profile except for two occurrences, on JD 8559.15 and JD 8565.07, where an additional redshifted absorption component is required to reproduce the observed line profile.