EDP Sciences
Free Access
Issue
A&A
Volume 600, April 2017
Article Number L8
Number of page(s) 5
Section Letters
DOI https://doi.org/10.1051/0004-6361/201730659
Published online 03 April 2017

© ESO, 2017

1. Introduction

The bright 51−60 A+ CH3OH maser transition at 6668.5 MHz has been recognized for a long time to be a selective signpost for recently formed luminous young stars (Lbol> 5 × 103L). Its excitation requires radiative pumping by 20–30 μm photons through the second torsionally excited state (Sobolev et al. 1997). Intense IR radiation typically permeates the warm dusty environment surrounding massive young stellar objects (YSO). Although the emission of most 6.7 GHz maser sources is stable on timescales of (at least) several years (Sanna et al. 2010; Moscadelli et al. 2011), in some cases, strong flares of up to several 100 Jy have been recorded. In addition, in some sources, selected maser components show a well-defined periodicity (e.g., Goedhart et al. 2004). These variations can be naturally related to an occasional and/or periodic change of the background radiation (amplified by the maser) or the IR pump field.

Recently, we have reported on the first ever detected accretion burst from a massive young star (NIRS 3) in the star-forming region S255 (Caratti O Garatti et al. 2016). This detection was triggered by the serendipitous discovery of a CH3OH maser flare toward S255 NIRS 3 by Fujisawa et al. (2015) in 2015 November. Subsequent subarcsecond near-IR (NIR) observations showed that the K- and H-band fluxes of this source had increased with respect to the pre-burst level by 2.9 and 3.5 mag, respectively, suggesting a relationship with the maser flare (Stecklum et al. 2016). Additional IR observations proved that the integrated luminosity from NIR to millimeter wavelengths had grown from 2.9 × 104  to   1.6 × 105L (Caratti O Garatti et al. 2016).

This letter reports on observations made with the European VLBI Network (EVN) and the Jansky Very Large Array (JVLA) of the 6.7 GHz CH3OH maser emission in S255 NIRS 3  (hereafter NIRS 3) at the time of the outburst. We compare the new observations with previous interferometric and single-dish data, and discuss the change in the maser spatial distribution, structure, and flux.

2. Observations and data analysis

2.1. EVN 6.7 GHz CH3OH maser

We observed the 6.7 GHz CH3OH maser emission toward NIRS 3 with the EVN1 as a Target of Opportunity program on 2016 April 12 (code: RS002). We also reduced archival EVN observations obtained before the flare on 2004 November 6 (code: EL032). Both observations were conducted in phase-referencing mode by fast switching between the maser target (at a Doppler velocity of 5 km s-1) and a strong (C-band flux >0.1 Jy) reference position calibrator, J0613+1708 (for exp. EL032) and J0603+1742 (for exp. RS002). Left and right circular polarizations were observed with two (EL032) and eight (RS002) baseband converters (BBC), each BBC being 2 MHz wide. The EL032 and RS002 experiments were processed at the correlator of the Joint Institute for VLBI in Europe (JIVE) using an averaging time of 0.5 s and 2 s, respectively. Data were analyzed with the NRAO2 Astronomical Image Processing System (AIPS) following the VLBI spectral line procedures. Absolute positions of the CH3OH maser spots are registered with an accuracy of about ± 1 mas at each epoch. Additional information on the EVN observations is summarized in Table 1.

Table 1

EVN and JVLA 6.7 GHz CH3OH maser observations.

2.2. JVLA continuum and 6.7 GHz CH3OH maser

We observed the C-band (4–8 GHz) continuum and 6.7 GHz CH3OH maser emission in NIRS 3 with the JVLA of the NRAO using both the B- and A-Array configurations (code: 16A-424). Table 1 reports the main observational parameters. For high-sensitivity continuum observations, we employed the three-bit samplers, observing dual polarization across the maximum receiver bandwidth of 4 GHz. A narrow spectral unit of 4 MHz was centered at the rest frequency of the 6.7 GHz maser transition and correlated with 1664 channels. 3C48 was observed as flux density, bandpass, and delay calibrator. As a complex gain calibrator we used J0534+1927, which has a correlated flux of 1 Jy and an angular separation of about 7°  from the target maser. Data were reduced with the Common Astronomy Software Applications package (CASA v 4.3.1) making use of the JVLA pipeline. Mainly because of radio frequency interferences, which severely affect the C band, about 30% of the original data was flagged. We performed self-calibration (both in phase and amplitude) on the strongest maser channel at an LSR velocity of 6.4 km s-1 and applied these solutions to all maser channels before imaging. Based on a cross match between the JVLA and EVN maser maps during the outburst phase, we estimate an astrometric accuracy of ±50 mas for the JVLA dataset.

thumbnail Fig. 1

Comparison of 6.7 GHz CH3OH maser spectra obtained toward NIRS 3. Two single-dish pre-burst spectra obtained with the Green Bank (GB–43 m, black) and Effelsberg (EF, magenta) antennas are compared with the emission detected during the outburst phase at three different baselines: an EF total-power (filled histogram), a cross-power (red) of the EF and Westerbork (WB) baseline (~250 km), and a synthesized VLA spectrum (blue) obtained in the B configuration (maximum baseline of 11 km). The legend on the left side reports the observing dates. The dotted vertical line indicates the systemic velocity (Vsys) of the source S255-SMA1 from Zinchenko et al. (2015).

Open with DEXTER

thumbnail Fig. 2

Distribution of the 6.7 GHz CH3OH masers toward NIRS 3. Circles and triangles represent maser spots before and after the outburst, respectively. Relative maser positions between the two epochs are accurate within a few mas. Note that the apparent motion of the 6.7 GHz CH3OH masers in NIRS 3 between the two EVN epochs is negligible (Rygl et al. 2010). The symbol size varies logarithmically with the maser brightness, and colors indicate the maser VLSR according to the right-hand scale. Letters “A” and “P” are used to label prominent maser clusters. The gray-scale image shows the velocity-integrated emission of the 6.7 GHz masers observed with the JVLA A-Array on 2016 October 15. The corresponding intensity scale is given at the top. The JVLA 5 GHz continuum emission is drawn with red contours. The rms noise of the image is 12 μJy beam-1. Contour levels are 0.11, 0.15, and from 0.22 to 2 mJy beam-1 in steps of 0.22 mJy beam-1. The synthesized beam of the continuum map is 0.′′25 × 0.′′24 at PA = 51°.

Open with DEXTER

3. Results

Figure 1 compares 6.7 GHz maser spectra observed both before and during the NIR burst, and obtained with both single-dishes and interferometers. The Effelsberg spectra clearly show that new intense spectral features appeared during the burst at VLSR 6 km s-1  (hereafter, the flare emission), redshifted with respect to the pre-burst main spectral lines. On the other hand, pre-burst spectra show no significant variations since the first detection of this maser by Menten (1991). Figure 1 shows that the flare emission is heavily resolved on even the shortest EVN baselines between Effelsberg and Westerbork, which recover only 10% of the peak flux. About 3.5 months after the EVN run, JVLA B-Array observations recover the entire Effelsberg flux within a beam of about 1′′, indicating a substantial time stability in the outbursting maser spectrum. The only notable change occurs in the secondary peak at VLSR  of 5.9 km s-1, which shows a further increase by about 50%.

Figure 2 shows the spatial and VLSR distribution of individual 6.7 GHz CH3OH maser centers observed with the EVN before (triangles) and after (circles) the burst. The gray-scale image represents the velocity-integrated emission of the 6.7 GHz masers recovered with the JVLA A-Array. The 5 GHz continuum emission in the region (red contours) pinpoints the position of the massive YSO NIRS 3. Pre-burst maser emission originates within a radius of 0.3′′  from the massive YSO, and lies above the 5 GHz continuum emission. By comparing Figs. 1 and 2, we infer that the flare emission arises from a region twice as large as the pre-burst emission, with the brightest spectral feature (VLSR 6.5 km s-1) located at the position of cluster A. Cluster A coincides in position and VLSR with the redshifted side of the rotating envelope observed by Zinchenko et al. (2015, their Fig. 5) with the SMA in 2011. These pre-burst observations reveal a rich reservoir of CH3OH molecules in the gas phase with a relative (to H2) abundance of 10-6 at a temperature of 170 K, suitable for maser excitation. The richest pre-burst cluster (“P” in Fig. 2), located NW of the radio continuum peak, disappears during the outburst phase. The close correspondence between the maser emission in the EVN and JVLA A-Array maps suggests that no significant changes in maser positions occurred between 2016 April and October.

For the brightest maser channel at a VLSR  of 6.43 km s-1, Fig. 3 presents a plot of the visibility amplitude as a function of the JVLA baseline length, after combining the A- and B-Array datasets. This plot shows that the flare emission at its spectral peak is slightly resolved by the JVLA observations. While on the shortest baselines the B-Array configuration recovers the entire Effelsberg flux (1000 Jy, see Fig. 1), about 10–15% of the single-dish flux is already missing on the longest baselines (<235 Kλ). This behavior is consistent with a source angular size of 0.′′24. At the resolution of the A-Array configuration (for baselines >235 Kλ) the maser flux progressively decreases with increasing baseline length, following a profile that can be fitted with a more compact spatial structure of size 0.′′1.

4. Discussion

In the following, we show that the radiation field triggered by the accretion burst excites a plateau of maser emission across a large portion of the redshifted envelope surrounding NIRS3 (between radii of 500–1000 AU). Throughout the wavelength range 20–30 μm that contains the pumping transitions of the 6.7 GHz maser line, the outburst flux from NIRS 3 exceeds the pre-burst level by a factor from 3 (at 20 μm) to 10 (at 30 μm; Caratti O Garatti et al. 2016; see Fig. 3). The 6.7 GHz maser burst was reported less than five months (November 2015) after the estimated beginning of the IR burst (June 2015; Caratti O Garatti et al. 2016). On this basis, it is reasonable to assume that the maser burst has been triggered by the IR burst. At a trigonometric distance of 1.8 kpc (Rygl et al. 2010; Burns et al. 2016), cluster A is projected more than 500 AU from the location of the high-mass YSO (identified by the radio continuum peak in Fig. 2). In order to produce a change in maser excitation conditions at position A, the triggering event has to propagate at a speed 0.02 c from the origin of the burst. This proves that the maser flare cannot be excited by a mechanical perturbation of the circumstellar medium, but must be triggered by the enhanced radiation field that is due to the accretion event.

Masers can operate in two different regimes. We indicate with Tm and τ the brightness temperature and the optical depth of the maser radiation. Weak masers are normally unsaturated, in which case Tm = T0e| τ |, where T0 is the brightness temperature of the background radiation amplified by the maser. At increasing intensity masers start to saturate, and in this regime, the brightness temperature grows only linearly with the optical depth, that is, Tmτ. In general, | τ | ∝ lm, meaning that the maser optical depth is proportional to the total pump rate P to the maser levels, the pumping efficiency, η, and the amplification path, lm. If the masers are radiatively pumped, P is proportional to the flux Fp of pumping photons incident on the masing gas.

In our case, from the ratio between the outburst on pre-burst luminosities, we find that Fp has increased by a factor 5. In 2016 April, during the burst, our EVN observations derived maser intensities in the range 0.1–100 Jy beam-1, corresponding to brightness temperatures of 108–1011 K. Since pre-burst EVN 6.7 GHz observations did not detect emission at the same location of the flaring masers, pre-burst maser intensities should be < 50 mJy or, equivalently, Tm< 5 × 107 K. Comparing the relatively small change in Fp (<10) with the large variation in maser intensity (more than 3–4 orders of magnitude), we can argue that most of the growth in maser intensity has occurred in a regime of non-saturation. In Appendix A, we estimate the brightness temperatures of both the maser saturation level, Ts = 1010 K, and the background radiation, T0 ≤ 50 K. Based on these estimates, the maser gain, G = e| τ |, has to be in the range 107–109 to account for the outburst brightness of the unsaturated masers (108–1010 K). If Fp and | τ | increase by a factor 5 during the burst, the outburst maser gain should be approximately the fifth power of the pre-burst gain. Thus G ≤ 102 before the burst, corresponding to | τ | ≤ 5 and pre-burst brightness temperatures 5000 K. This is about four orders of magnitude below the EVN detection threshold, consistent with the non-detection of emission throughout the maser flare area by VLBI observations before the burst.

Figure 1 shows that the large majority (>90%) of the flux of the strongest peak of the Effelsberg 6.7 GHz maser spectrum, emerging from cluster A (see Fig. 2), is resolved out on the Effelsberg-Westerbork baseline. VLBI observations of the 6.7 GHz masers have presented many cases of spectral features that were heavily resolved even on the shortest baselines of the VLBI array (Minier et al. 2002), suggesting that 6.7 GHz CH3OH masers have a core/halo structure, with core and halo diameters ranging in the intervals 2–20 AU and 10–300 AU, respectively. However, larger flux losses are generally observed for weaker spectral components, whereas most of the flux of the strongest peak is recovered. Instead, in NIRS 3  the reverse occurs, and Fig. 3 indeed proves that the spatial structure of the 6.7 GHz maser spectral peak is extended, with a size of up to , or 430 AU, which is larger than the typical maser halos measured by Minier et al. (2002). Our EVN observations, filtering out spatial scales 70 AU, resolve cluster A into a tight cluster of strong maser spots. These compact spots are just local peaks emerging from a continuous plateau of weaker 6.7 GHz maser emission. From Fig. 3, we estimate that the brightness temperature of this maser plateau is 5 × 108 K.

So far, our discussion has focused on the bursting maser emission. Now, we turn our attention to the main pre-burst cluster (“P”, see Fig. 2). Since the 6.7 GHz maser spectrum toward NIRS 3 is approximately stable over a timescale of 13 yr before the burst (see Fig. 1), we can reasonably assume that the disappearance of cluster P is due to the accretion burst, and we discuss two possible ways of maser suppression: 1) UV photodissociation of CH3OH, and 2) perturbation of the velocity coherent path. Figure 2 shows that cluster P is located at an average distance RP ≈ 0.′′2, or 340 AU, from the YSO and has a size DP ≈ 0.′′14, or 240 AU. Using the value of H2 (number) density nH2 ~ 108 cm-3, as determined by Zinchenko et al. (2015) for the molecular envelope around NIRS 3, we estimate for cluster P an H2 column density NH2 ≈ 4 × 1023 cm-2, corresponding to a visual extinction AV ≈ 400 mag. In Appendix B, we find that the threshold in visual extinction needed to shield CH3OH from photodissociation in cluster P is AV ≳ 7 mag. With the caveat of our simplistic assumptions of homogeneous density distribution for the molecular envelope and isotropic radiation from the YSO, we therefore consider it unlikely that UV photodissociation induced by the accretion burst is the main cause for the disappearance of this cluster of 6.7 GHz masers.

A line-of-sight velocity coherence over a long path is needed for maser amplification. The change in velocity, ΔV, that is due to radiation pressure on the maser cluster P can be estimated with the expression , where c is the speed of light, L is the outburst luminosity of NIRS 3, t is the time over which the mechanism is at work (i.e., the duration of the burst), MP is the mass of the molecular clump harboring the maser cluster P, and ΩP is the solid angle subtended by cluster P at the YSO position. Using RP ≈ 340 AU and DP ≈ 240 AU for the radial distance and the size of cluster P, respectively, and a density nH2 ~ 108 cm-3, we derive ΩP ≈ 0.4 sterad and MP ≈ 4 × 10-3M. With L = 1.6 × 105L and t ≈ 1 yr (approximately the time elapsed from the onset of the burst in June 2015 and the EVN observations in April 2016), we estimate ΔV ≈ 0.03 km s-1. This value, calculated using the whole clump mass, is probably a lower limit, since the whole radiative flux could be intercepted only by the portion of the clump closer to the YSO. Thus, we consider it plausible that radiative pressure, over a timescale of 1 yr, could have produced a velocity change comparable with the 6.7 GHz maser line width of 0.10.2 km s-1 along a sizable fraction of the maser path length. For unsaturated operation, this effect could have drastically reduced the maser intensities in cluster P. Repeating the same calculation for cluster A with a distance RA ≈ 0.̋45 (800 AU) and a size DA ≈ 0.̋240 (430 AU), we find ΔV ≈ 0.002 km s-1. This is an order of magnitude lower than the value estimated for cluster P, which might explain why radiation pressure does not destroy the velocity coherence of the 6.7 GHz masers in cluster A.

thumbnail Fig. 3

Visibility amplitude vs. JVLA baseline length for the brightest maser channel (at VLSR = 6.43 km s-1). The blue dotted and red dashed lines show the Gaussian fits to the amplitude profile using only data with baseline lengths lower and higher than 235 Kλ, respectively, which is the maximum baseline sampled with the B-Array. The amplitude and FWHM size of the fitted Gaussians are reported with consistent colors.

Open with DEXTER


1

The European VLBI Network is a joint facility of European, Chinese, South African and other radio astronomy institutes funded by their national research councils.

2

The National Radio Astronomy Observatory (NRAO) is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

Acknowledgments

A.S. gratefully acknowledges financial support by the Deutsche Forschungsgemeinschaft (DFG) Priority Program 1573. M.C.W. and A.C.oG. were supported by Science Foundation Ireland, grant 13/ERC/I2907.

References

Appendix A: Determination of saturation and background brightness temperatures

In order to evaluate the maser optical depth, we need to estimate the brightness temperatures of saturation and background radiation. To estimate the saturation level we employ the method by Moscadelli et al. (2003), based on plots of the maser brightness temperature, Tm, versus the emission area, Am. Since the flux of saturated masers, expressed by the product TmAm, approaches but cannot exceed the flux of the pumping photons, all saturated masers lie close to a curve TmAm = K, with K being a constant. Figure A.1 shows that all the spots with a sufficiently high value of TmAm (300 × 109 K AU2) lie close to an upper envelope, well fit with the hyperbola TmAm = 480 × 109 K AU2. From the same plot we can evaluate that while all the masers with Tm ≥ 1010 K are saturated, the large majority of those with Tm ≤ 1010 K are still unsaturated. Therefore we can take as a threshold for saturation Ts = 1010 K. Maser theory predicts that the line width of the spectral profile of unsaturated masers is subthermal and that the maser line width relaxes to the thermal value only after the threshold of saturation is crossed. As a further check of the saturation condition, we also determined the FWHM line width of the masers by fitting a Gaussian profile to the spectra of individual maser spots. We find that it varies in the small range 0.18 ± 0.05 km s-1. This value is significantly lower than the thermal line width 0.43 km s-1 expected for a kinetic temperature 150 K (as derived by Zinchenko et al. 2015, in the molecular envelope surrounding NIRS 3), and it supports our conclusion that most of the outbursting masers are still unsaturated.

thumbnail Fig. A.1

Brightness temperature of the spots (in units of 109 K) vs. the deconvolved sky-projected spot area (in AU2). The spot area is defined as Am = (π/ 4) ab, a and b being the deconvolved FWHM along the major and minor axes, respectively. Circles with crosses indicate the spots with TmAm ≥ 300 × 109 K AU2. The red dashed line gives the best hyperbolic fit to these points, Tm ∝ 1/Am. The inset shows an enlargement of the region with Tm ≤ 1010 K.

Open with DEXTER

We identify three possible contributions to the maser background radiation T0: 1) spontaneous emission, 2) dust emission, and 3) ionized gas emission. For unsaturated masers, the brightness temperature of spontaneous emission is given by (Elitzur 1992, Eqs. (2.3.28) and (4.2.11)), where h and k are the Planck and Boltzmann constants, respectively, and ν0 is the maser line rest frequency. For values of η in the range 0.11, spontaneous emission always contributes 2 K. Since dust emission is optically thin at centimeter wavelengths, its brightness temperature is Tdτd, where Td and τd are the kinetic temperature and the optical depth of the dust, respectively. Using Td ≤ 200 K (Zinchenko et al. 2015) and τd ≪ 0.1, we estimate a dust contribution 20 K. Our C-band VLA A-Array observations do not detect free-free emission toward cluster A, where the most intense 6.7 GHz masers are found. The 3σ continuum detection threshold of 36 μJy beam-1 corresponds to a brightness temperature of 16 K. Adding the contributions of the spontaneous, dust and free-free emissions, we derive T0 ≤ 50 K.

Appendix B: Photodissociation of the CH3OH molecule in cluster P

In the following, we wish to evaluate the threshold in visual extinction to shield CH3OH molecules inside cluster P from the UV flux produced by the accretion burst in NIRS 3. The UV energy density, EUV, inside cluster P can be estimated with the expression (B.1)where c is the speed of light, RP = 340 AU is the distance of cluster P from NIRS 3, L = 1.6 × 105L is the outburst luminosity of the YSO, and fUV is the fraction of that luminosity in photons with wavelength 2800 Å, which is the photodissociation threshold of CH3OH (Heays et al. 2017). We find EUV = fUV8 × 10-4 erg cm-3. A probable range for fUV is 0.1 ≤ fUV ≤ 1, which implies 8 10-5EUV ≤ 8 × 10-4 erg cm-3. This value of the energy density is a factor 1.6 × 109−1.6 × 1010 higher than the average field in the interstellar medium (ISM) of 5 × 10-14 erg cm-3 (Draine 2011). The photodissociation rate of CH3OH within a molecular clump exposed to the average ISM field is estimated to be 1.4 × 10-9 exp(−2.76 AV) molecule-1 s-1 (Heays et al. 2017), where AV is the visual extinction of the clump. By scaling this value proportionally with the energy density of the radiation field, we can estimate that CH3OH inside cluster P is photodissociated at a rate of ~10 exp(−2.76 AV) s-1 molecule-1 s-1. Based on this rate, we derive that a value of AV ≥ 7 mag can be enough to efficiently shield all the CH3OH molecules within cluster P over the timescale of 1 yr between the onset of the burst (June 2015) and the EVN observations (April 2016).

All Tables

Table 1

EVN and JVLA 6.7 GHz CH3OH maser observations.

All Figures

thumbnail Fig. 1

Comparison of 6.7 GHz CH3OH maser spectra obtained toward NIRS 3. Two single-dish pre-burst spectra obtained with the Green Bank (GB–43 m, black) and Effelsberg (EF, magenta) antennas are compared with the emission detected during the outburst phase at three different baselines: an EF total-power (filled histogram), a cross-power (red) of the EF and Westerbork (WB) baseline (~250 km), and a synthesized VLA spectrum (blue) obtained in the B configuration (maximum baseline of 11 km). The legend on the left side reports the observing dates. The dotted vertical line indicates the systemic velocity (Vsys) of the source S255-SMA1 from Zinchenko et al. (2015).

Open with DEXTER
In the text
thumbnail Fig. 2

Distribution of the 6.7 GHz CH3OH masers toward NIRS 3. Circles and triangles represent maser spots before and after the outburst, respectively. Relative maser positions between the two epochs are accurate within a few mas. Note that the apparent motion of the 6.7 GHz CH3OH masers in NIRS 3 between the two EVN epochs is negligible (Rygl et al. 2010). The symbol size varies logarithmically with the maser brightness, and colors indicate the maser VLSR according to the right-hand scale. Letters “A” and “P” are used to label prominent maser clusters. The gray-scale image shows the velocity-integrated emission of the 6.7 GHz masers observed with the JVLA A-Array on 2016 October 15. The corresponding intensity scale is given at the top. The JVLA 5 GHz continuum emission is drawn with red contours. The rms noise of the image is 12 μJy beam-1. Contour levels are 0.11, 0.15, and from 0.22 to 2 mJy beam-1 in steps of 0.22 mJy beam-1. The synthesized beam of the continuum map is 0.′′25 × 0.′′24 at PA = 51°.

Open with DEXTER
In the text
thumbnail Fig. 3

Visibility amplitude vs. JVLA baseline length for the brightest maser channel (at VLSR = 6.43 km s-1). The blue dotted and red dashed lines show the Gaussian fits to the amplitude profile using only data with baseline lengths lower and higher than 235 Kλ, respectively, which is the maximum baseline sampled with the B-Array. The amplitude and FWHM size of the fitted Gaussians are reported with consistent colors.

Open with DEXTER
In the text
thumbnail Fig. A.1

Brightness temperature of the spots (in units of 109 K) vs. the deconvolved sky-projected spot area (in AU2). The spot area is defined as Am = (π/ 4) ab, a and b being the deconvolved FWHM along the major and minor axes, respectively. Circles with crosses indicate the spots with TmAm ≥ 300 × 109 K AU2. The red dashed line gives the best hyperbolic fit to these points, Tm ∝ 1/Am. The inset shows an enlargement of the region with Tm ≤ 1010 K.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.