4a5
> \usepackage{color}
15c16
< \author{Priya.~\surname{Desai}$^1$\sep
---
> \author{Priya~\surname{Desai}$^1$\sep
28c29,34
< We investigate flare related changes in the photospheric absorption line (Fe I  617.3 nm) profile in a large sample of solar flares of varying X-ray  and H-alpha class, observed with the Helioseismic and Magnetic Imager (HMI) instrument aboard the Solar Dynamic Observatory (SDO).  A distinct continuum enhancement, along with a corresponding decrease in line-depth is observed in over 40\% of the flares. This continuum brightening can be observed in flares down to GOES X-ray class C 5.0. In most cases,  the corresponding brightening in line-core is significantly larger than the  observed continuum brightening, suggesting that the  line formation is more sensitive to flare effects than the continuum. In powerful flares  (example: GOES X6.9(SOL2011-08-09T08:05) and GOES X1.9 (SOL2011-09-24T09:40)) we observe the absorption line profile temporarily reverses direction and goes into emission during the flare. In this paper we present observations from three flares: SOL2011-09-24T09:40, an X1.9 flare where the absorption line briefly goes into emission; SOL2011-09-07T23:38, a X1.8 flare which shows noticeable reduction in line-depth and continuum brightening  but does not go into emission; and SOL2010-11-04T23:58:00, a M1.6 flare which shows line-core brightening but no significant continuum enhancement. 
---
> We investigate flare-related changes in a photospheric absorption line (Fe~{\sc i}  617.3~nm) profile in a large sample of solar flares, of varying X-ray  and H-$\alpha$ classes, observed with the Helioseismic and Magnetic Imager (HMI) instrument aboard the Solar Dynamic Observatory (SDO).  
> A distinct continuum enhancement, along with a corresponding decrease in line depth, is observed in over 40\% of the flares. 
> This continuum brightening can be observed in flares down to GOES X-ray class C~5.0. 
> In most cases,  the corresponding brightening in the line core is significantly larger than the observed continuum brightening, suggesting that the line formation is more sensitive to flare effects than the continuum. 
> In powerful flares  (examples: SOL2011-08-09T08:05, GOES X6.9, and SOL2011-09-24T09:40, GOES X1.9) we observe the absorption line profile to reverse direction temporarily and go into emission in small regions during the flare. 
> In this paper we present observations from three flares: SOL2011-09-24T09:40 (X1.9) where the absorption line briefly goes into emission; SOL2011-09-07T23:38 (X1.8)  which shows noticeable reduction in line-depth and continuum brightening but does not go into emission; and SOL2010-11-04T23:58:00 (M1.6)  which shows line-core brightening but no significant continuum enhancement. 
37c43,44
<  A solar flare is an explosive release of energy stored in twisted magnetic fields, usually above or near sunspots. The term white-light Flare's ( WLF's)  has traditionally (Carrington, 1859) been used to designate solar flares observable as transient brightenings in the visible continuum. WLF's are important in flare research because they represent a large fraction of the radiated energy of a flare (eg. Neidig, 1989) and may challenge our current understanding of flare energy transport.
---
>  A solar flare is an explosive release of energy stored in twisted magnetic fields, usually above or near sunspots. The term white-light flare (WLF)  has traditionally  been used to designate solar flares observable as transient brightenings in the visible continuum (Carrington, 1859). 
>  WLFs are important in flare research because the continuum represents a large fraction of the radiated energy of a flare (e.g., Neidig, 1989), and their properties challenge our current understanding of flare energy transport (e.g., Krucker et al. 2011).
39c46,62
<       It is currently believed that the visible continuum is enhanced  in all flares, but that the weaker ones are not detected as the signal is lost in the normal spatial and temporal fluctuations  of the photosphere. Furthermore, even in strong flares, of small spatial extent (\~ 5 Mm), and of short duration (\~ 250s), a systematic observation of the white-light component of flares has been historically challenging as it is difficult to serendipitously capture a flare at the right time and location for a spectroscopic observation, and most measurements (including space-based ones) do not have the temporal resolution to adequately observe the WL continuum near sunspots with low signal and contrast. In fact, prior to 1993 (Neidig {\sl et al.} 1991), only \~86 WLF's had been reported and white-light emissions were only detected in flares above GOES magnitude X2. A study of white-light flares observed by Hinode (Wang, H, 2009) found the detectability threshold to be around M1 flares. Hudson, {\sl et.al}. (2006) used white-light observations from the Transitional Region and Coronal Explorer (TRACE) with a spatial resolution of 1'' and detected white-light emission for events as weak as GOES C2.0. Jess et al (2008) detected intense white-light emission in the blue continuum for a C1.6 flare using the diffraction-limited observations of the Swedish 1-meter Solar Telescope.
---
> It is currently believed that the visible continuum is enhanced  in all flares, but that the weaker ones are not detected as the signal is lost in the normal spatial and temporal fluctuations  of the photosphere. 
> Furthermore, even in strong flares, of small spatial extent ($\sim$5~Mm), and of short duration ($\sim$250~s), a systematic observation of the white-light component of flares has been historically challenging.
> It is difficult to capture a flare serendipitously at the right time and location for a spectroscopic observation, and most measurements (including space-based ones) do not have the spatial and temporal resolution to  observe the WL continuum  near sunspots, which have large image contrast. 
> In fact, prior to 1993 (Neidig {\sl et al.} 1991), only $\sim$86 WLFs had been reported, and white-light emissions were only detected in flares above GOES magnitude X2. 
> Surveys of white-light flares observed by Yohkoh (Matthews et al., 2003) and by Hinode (Wang, 2009) found the detectability threshold to be around M1 flare. 
> Hudson {\sl et.al}. (2006) used white-light observations from the Transition Region and Coronal Explorer (TRACE) with a spatial resolution of 1$''$ and detected white-light emission for events as weak as GOES C2.0. 
> Jess et al (2008) detected intense white-light emission in the blue continuum for a C1.6 flare using the diffraction-limited observations of the Swedish 1-meter Solar Telescope.
> 
> Motivated by this renewed interest in white-light flares, which we define as flares exhibiting enhanced emission in parts of the spectrum originating at or near the height of the the visible continuum photosphere, we set out to determine if  we could detect photospheric flare signatures using the Helioseismic and Magnetic Imager (HMI) instrument on the Solar Dynamics Observatory (SDO), launched on February 11, 2010. 
> The HMI instrument provides images in a photospheric line, Fe~{\sc  i} 617.3~nm, at sufficient spatial and temporal resolution to resolve flare signatures and provide continuous images during the flare.  
> Measurements of the line strength and nearby continuum brightness along with Doppler and polarization information, and covering the whole solar disc, are available at a spatial scale of 0.5''  once every 45~sec. 
> Full-disc filtergrams in different polarizations and wavelengths across the line are available every 1.875~sec, offering the potential of high time resolution of the photospheric signatures of solar flares, provided that they can be detected and adequately measured. 
> Thus, we have  an opportunity to study the profiles of a large sample of flares of varying magnitudes using the same instrument, unlike previous ground based observations, which typically were with slit spectrographs providing limited coverage (e.g. Neidig, 1989) and variable observing conditions.
> 
> {\color {red} The HMI capability provides a first systematic opportunity, from space, to study the behavior of the line profile.
> Specifically we would like to understand emission in the line itself, along with the nearby continuum, as a more complete guide to the photospheric participation in the flare.
> Previous observations have on occasion reported line-core emission (e.g., Babin \& Koval, 2007) in some flares, and we use the HMI filtergrams to determine the association between this phenomenon and the occurrence of true continuum.}
41d63
<  Motivated by this renewed interest in white-light flares, which we define as flares exhibiting enhanced emission in parts of the spectrum originating at or near the height of the the visible continuum photosphere, we set out to determine if  we could detect photospheric flare signatures using the Helioseismic and Magnetic Imager (HMI) instrument on the Solar Dynamics Observatory(SDO), launched on February 11, 2010. The HMI instrument provides images in a photospheric line, Fe I 617.3 nm, at sufficient spatial and temporal resolution to resolve flare signatures and provide continuous images during the flare.  Measurements of the line strength and nearby continuum brightness along with Doppler shift and Zeeman splitting and covering the whole solar disc, are available at a spatial scale of 0.5"  once every 45 sec. Full-disc filtergrams in different polarizations and wavelengths across the line are available every 1.875 sec, offering the potential of high time resolution of the photospheric signatures of solar flares, provided that they can be detected and adequately measured. Thus, we have  an opportunity to study the profiles of a large sample of flares of varying magnitudes using the same instrument, unlike previous ground based observations, which typically were with slit spectrographs providing fairly limited coverage (e.g. Neidig, 1989; Babin \& Koval, 2007).
45c67,79
<   Since a strong correlation is known to exist between X-ray emission and observed continuum brightening (Neidig, 1989), we chose to analyze a large sample of  candidate flares belonging to GOES X-ray class X6.9 to class C1.5. In addition we also looked for flare-related  photospheric effects in H-alpha flares that did not have an established detected X-ray counterpart. We primarily examined the HMI observables during the flares  in order to establish the quantities most likely to be affected over the course of a flare, and their typical behavior. For this paper however, we have present observations from three representative flares that seem to best demonstrate the different observed photospheric effects: SOL2011-09-24T09:40, an X1.9 flare where the absorption line briefly goes into emission; SOL2011-09-07T23:38, a X1.8 flare which shows noticeable reduction in line-depth and continuum brightening  but does not go into emission; and SOL2010-11-04T23:58:00, a M1.6 flare which shows line-core brightening but no significant continuum enhancement (See Table 1).    
---
> {\color {red} 
> We include the equivalent width $W$ of the HMI line as one of the observables that we discuss.
> This offers a convenient way to relate the strength of the line (represented by $I_\lambda$ in a small spectral range) to its neighboring continuum $I_c$, defined as $W = \int{I_\lambda d\lambda} / I_c$ and expressed dimensionally.
> In the case of a white-light flare, though, the concept is a bit fuzzy: we could choose the preflare continuum, the enhanced continuum, or the total as a reference.
> For the purposes of this paper we use the first choice simply as a matter of convenience, without any theoretical justification,
> and note that a line reversal changes the sign of $W$.
> We also adopt the convention that the normal absorption line has positive equivalent width.
> }
> 
> Since a strong correlation is known to exist between X-ray emission and observed continuum brightening (Neidig, 1989), we chose to analyze a large sample of  candidate flares belonging to GOES X-ray class X6.9 to class C1.5. In addition we also looked for flare-related  photospheric effects in H$\alpha$ flares that did not have detected X-ray counterparts. 
> We primarily examined the HMI observables during the flares  in order to establish the quantities most likely to be affected over the course of a flare, and their typical behavior. 
> For this paper however, we only present observations from three representative flares that seem to best demonstrate the different observed photospheric effects: SOL2011-09-24T09:40, an X1.9 flare where the absorption line briefly goes into emission; SOL2011-09-07T23:38, a X1.8 flare which shows noticeable reduction in line-depth and continuum brightening  but  for which the line does not go into emission; and SOL2010-11-04T23:58:00, a M1.6 flare which shows line-core brightening but no significant continuum enhancement (See Table~1).    
> 
50c84
< \begin{tabular*}{0.95\textwidth}{l p{1.5cm} l l l p{3cm}}
---
> \begin{tabular*}{\textwidth}{l l l l l l }
52c86
< Date &GOES Peak Time(UT) &AR Number &Location& Flare Class &Comments\\
---
> Date & Peak (UT) &AR  &Location&  Class &Comments\\
54,58c88,91
< 
<  2010.11.04 & 23:58 & 11121 & S20 E76 & M1.6 & line-core brightening but no continuum brightening.\\
<  2011.09.07 & 22:38 & 11283 & N14 W30 & X1.8 & line-core brightening and continuum brightening.\\
<  2011.09.24 & 09:40 & 11302 & N13 E61 & X1.9 & line-core brightening, continuum brightening and line goes into emission.\\
<  2011.08.09 & 08:05 & 11263 & N14 W69 & X6.9 & line-core brightening, continuum brightening and line goes into emission.\\
---
>  2010.11.04 & 23:58 & 11121 & S20 E76 & M1.6 & core, no continuum\\
>  2011.09.07 & 22:38 & 11283 & N14 W30 & X1.8 & core, continuum \\
>  2011.09.24 & 09:40 & 11302 & N13 E61 & X1.9 & core in emission, continuum\\
>  2011.08.09 & 08:05 & 11263 & N14 W69 & X6.9 & core in emission, continuum\\
63c96,105
<         Five HMI observables, continuum intensity (I$_c$), line depth (L$_d$), line-width (L$_w$), dopplergrams (V), and magnetograms (M) are available every 45 sec. In order to produce these quantities however, HMI produces a nearly continuous stream of full-disc images of the sun in a set of six narrow wavelength bands (FWHM 76 m\AA) clustered around the central wavelength of an Fe I photosheric absorption line as 6173\AA\ (See Schou {\sl et al.} 2012). The filtergrams are made at a cadence of one every 1.875 second. They are taken in six combinations each of circular and linear polarization states in order to provide a full set of Stokes parameters for determination of the physical observables. The full sequence of all possible filtergrams repeats at a period of 135 s; the repetition period of filtergrams necessary to produce the intensity and Doppler observables is one-third that, so they are available at a cadence of 45s, the basic cadence Image stabilization provides spatial stability of 0.03 arc-sec throughout the complete sequences of filtergrams required for the observables. The per-pixel statistical noise for filtergrams is approximately 0.3\%. Exposure time knowledge is considerably better, about 0.002\%, and for the continuum intensity measurement, combining all filtergrams over a 45-sec frame sequence, the per-pixel noise is about 0.1\%.
---
> 
> Five HMI observables, continuum intensity (I$_c$), line depth (L$_d$), line-width (L$_w$), dopplergrams (V), and magnetograms (M) are available every 45 sec. 
> In order to generate these quantities, HMI produces a nearly continuous stream of full-disc images of the sun in a set of six narrow wavelength bands (FWHM 76 m\AA) clustered around the central wavelength of the Fe~{\sc i} photospheric absorption line at 6173\AA\ (see Schou {\sl et al.} 2012). The filtergrams are made at a cadence of one every 1.875~s. They are taken in six combinations each of circular and linear polarization states in order to provide a full set of Stokes parameters. 
> The full sequence of all possible filtergrams repeats at a period of 135 s; the repetition period of filtergrams necessary to produce the intensity and Doppler observables is one-third of that, so they are available at a cadence of 45~s, the basic cadence for
> the observations reported here.
> Image stabilization provides spatial stability of 0.03$''$ throughout the complete sequences of filtergrams required for the observables. 
> Typical per-pixel statistical noise for one filtergram is approximately 0.3\%{\color {red}, as determined from time-series fluctuations
> away from the flare time}.
> Exposure-time knowledge is about 0.002\% and does not contribute to the error appreciably.
> For the continuum intensity measurement, combining all filtergrams over a 45~s frame sequence, the per-pixel noise is about 0.1\%.
68,76c110,118
< \epsfig{file=imgs/09_07/im_33.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_07/im_35.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_07/im_37.ps,width=0.3\linewidth,clip=} \\
< \epsfig{file=imgs/09_07/im_39.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_07/im_41.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_07/im_43.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_07/im_45.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_07/im_47.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_07/im_49.ps,width=0.3\linewidth,clip=}\\
---
> \epsfig{file=imgs/09_07/im_33.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_07/im_35.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_07/im_37.ps,width=0.29\linewidth} \\
> \epsfig{file=imgs/09_07/im_39.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_07/im_41.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_07/im_43.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_07/im_45.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_07/im_47.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_07/im_49.ps,width=0.29\linewidth}\\
78c120,124
< \caption{Brightening observed in the line-core images for SOL2011-09-07T22:38:00 for selected times before and during the flare.The region shown is a $6^\circ$ x $6^\circ$0 square, suitably tracked and mapped to a grid centered at latitude 15$^o$.64 N and Carrington longitude 226$^o$.09, the nominal location of the flare. Each image is 90s apart.}
---
> \caption{Brightening observed in the line-core images for SOL2011-09-07T22:38:00 for selected times before and during the flare.
> The region shown is a $6^\circ$ x $6^\circ$ square, suitably tracked and mapped to a grid centered at latitude 15$^o$.64 N and Carrington longitude 226$^o$.09, the nominal location of the flare. 
> The image times are 90~s apart {\color {red} starting at HH:MM:SS~UT}.
> {\color {blue} I suggest replacing the last image with the straight intensity image for spatial reference with respect to the sunspots - HH.}
> }
83,84c129,135
< 
<        We use tracked, 45s cadence, continuum (I$_c$) and  and line-depth (L$_d$) data for a period one hour or ninety minutes (30/45 minutes before and after the flare) to determine the ``line-core'' profile behavior during the flare. We define the ``line-core'' value at a pixel as the difference between the continuum intensity and the line-depth values at that pixel. Instead of using the standard observables, we used observables calculated using the algorithm normally reserved for the Near-Real Time (NRT) data, since the standard HMI observables  are produced using a weighting function extending over 135 seconds before and after the "recorded-time" and has negative side lobes in the range +/- 45-90 seconds. While providing more accurate results under normal circumstances, the standard interpolation scheme may result in spurious artifacts in the case of transient events (see Martinez-Oliveros {\sl et.al.} 2011). The NRT algorithm, on the other hand, involves simple 2-point (linear) interpolation with positive-definite  weightings among filtergrams within 45 seconds of the represented time. We deem the observed and so-produced data to be more a accurate representation of the line profile during an impulsive phenomenon. In the following analysis, we only use the  NRT algorithm data as well as the individual HMI filtergrams which have also been interpolated in time (using the same target time).
---
> We use tracked, 45s cadence, continuum (I$_c$) and  and line-depth (L$_d$) data for a period 1--1.5~h (30--45~min before and after the flare) to determine the line-core profile behavior during the flare. 
> We define the line core value at a pixel as the difference between the continuum intensity and the line-depth values at that pixel {\color {red} i.e., the measured intensity at the base of the profile}.
> Instead of using the standard observables, we used observables calculated using the algorithm normally reserved for HMI's Near-Real Time (NRT) data, since the standard HMI observables  are produced using a weighting function extending over 135 seconds before and after the recorded time, and has negative sidelobes in the range $\pm$45--90~s. 
> While providing more accurate results under normal circumstances, the standard interpolation scheme may result in spurious artifacts in the case of transient events (see Mart{\' i}nez-Oliveros {\sl et al.} 2011). 
> The NRT algorithm, on the other hand, involves simple 2-point (linear) interpolation with positive-definite  weightings among filtergrams within 45 seconds of the represented time. 
> We deem the observed and so-produced data to be more a accurate representation of the line profile during an impulsive phenomenon. 
> In the following analysis, we only use the  NRT algorithm data as well as the individual HMI filtergrams which have also been interpolated in time (using the same target time).
86,87d136
< 
< 
89,90c138,145
< \centerline{\includegraphics[width=0.75\textheight,clip=] {fig1b.ps}}
< \caption{ Mosaic of the time variation plots of the line-core intensity (I$_c$ - L$_d$ using NRT values)for a section of the active region that produced the flare SOL2011-09-07T22:38:00. Each plot corresponds to a single pixel, and light-curves are plotted  for every fifth pixel in a region covering a 20 x 20 pixels.The central pixel labeled (284,271) is located at $15^\circ$.53 N and $30^\circ$.6 W( Carrington longitude 226$^o$.5) . The horizontal(time) axis spans 60 minutes. The vertical axis spans minimum and maximum of the line-core intensity in that period. The central dotted line is  the mean line-core value for that pixel over the 60 minutes duration, and top and bottom dotted horizontal lines correspond to +/- 6 $\sigma$ from the mean line-core value.}    
---
> \centerline{\includegraphics[width=\linewidth] {fig1b.ps}}
> \caption{ Mosaic of the time variation plots of the line-core intensity (I$_c$ - L$_d$ using NRT values) for a section of the active region that produced the flare SOL2011-09-07T22:38:00. Each plot corresponds to a single pixel, and light-curves are plotted  for every fifth pixel in a region covering a 20$\times$20 pixels.
> The central pixel, labeled (284,271,) is located at $15^\circ$.53 N and $30^\circ$.6 W (Carrington longitude 226$^\circ$.5). 
> The horizontal (time) axis spans 60~min. 
> The vertical axis spans minimum and maximum of the line-core intensity in that period. 
> The central dotted line is  the mean line-core value for that pixel over the 60-min duration, and top and bottom dotted horizontal lines correspond to $\pm 6 \sigma$ from the mean line-core value.
> {\color {blue} It would be very helpful if the pixels display could be marked on one of the frames of Figure 1 - hh}
> }    
95,96d149
< 
< 
98,99c151,153
< \centerline{\includegraphics[width=0.75\textheight,clip=] {fig1c.ps}}
< \caption{ Mosaic of the time variation plots of the continuum intensity (NRT I$_c$) for the same section of the active region of flare SOL2011-09-07T22:38:00 as shown in Figure 2. Again, the horizontal axis spans 60 minutes but the vertical axis range includes the minimum and maximum continuum intensity values within that time period. The central dotted line is  the mean (NRT) continuum  value for that pixel over the 60 minutes duration, and top and bottom dotted horizontal lines correspond to +/- 6 $\sigma$ from the mean continuum value.}    
---
> \centerline{\includegraphics[width=\linewidth] {fig1c.ps}}
> \caption{ Mosaic of the time-variation plots of the continuum intensity (NRT I$_c$) for the same section of the active region of flare SOL2011-09-07T22:38:00 as shown in Figure~2. Again, the horizontal axis spans 60~min and the vertical axis range includes the minimum and maximum continuum intensity values within that time period. 
> The central dotted line is  the mean (NRT) continuum  value for that pixel over the 60-min duration, and top and bottom dotted horizontal lines correspond to $\pm 6\sigma$ from the mean continuum value.}    
106c160,163
< \caption{ Spectral intensity across the HMI line for selected times before and  during the flare for the central pixel (284 x 271)(i.e.at $15^\circ$.53 N and 30$^\circ$.6 W) of the above mosaic in Figure 2 and 3. Intensities in both left-circularly-polarized( LCP) and right-circularly polarized(RCP) light are shown.} 
---
> \caption{ Spectral intensity across the HMI line for selected times before and  during the flare for the central pixel (284 x 271) (i.e., at $15^\circ$.53 N and 30$^\circ$.6 W) of the above mosaic in Figures~2 and~3. Intensities in both left-circularly-polarized (LCP) and right-circularly-polarized (RCP) light are shown.
> {\color {blue} Better to give the actual times, rather than just say "selected." Also, the Y-axis labeling is odd. I suggest making the ticks at regular places. The error bars look too regular to be statistical; they are also about 10\% or so not consistent with the
> estimate in the text - HH}
> } 
110,111d166
< 
< 
113,125c168,190
<       The images and plots in Figures 1-4 are from flare SOL2011-09-07T22:38:00, of GOES class X1.8.  Figure 1 is a typical time sequence of tracked and mapped images of the line-core during the flare. The images are taken 90 seconds apart. Fig 2 illustrates the temporal variation in the line-core over a period of one hour (including during the flare) of twenty-five pixels chosen from a grid of 20 x 20 pixels. Each plot corresponds to a single pixel, and light-curves are plotted for every fifth pixel in the region covering the 20 x 20 pixels. The central pixel labeled (284, 271) is located at $15^\circ$.53 N and $30^\circ$.6 W. The horizontal axis spans 60 minutes. The central dotted line is  the mean line-core value for that pixel over the 60 minutes duration, and top and bottom dotted horizontal lines correspond to +/- 6 $\sigma $ from the mean line-core value. Note that these plots are made using tracked ``NRT'' data. As can be seen, a large number of the pixels show greater than a 6 $\sigma$ increase in the line core suggesting a strong enhancement.  
< 
<    Figure 3 shows a mosaic of plots of continuum intensity vs time for all the same pixels as Figure 2. However, fewer number of pixels show a 6$\sigma$ increase in the continuum brightening as compared to the line core. Figure 4 is a plot of the six spectral intensities observed across the line profile before and during the flare at the central pixel (labeled 284, 271) of the mosaic in Figure 2 and 3. The line profile shows a significant continuum enhancement and line core enhancement. Examination of all the pixels where the line core brightening was observed indicates that the line strength never exceeds the continuum and the line always stays in absorption. 
< 
<    Figures 5-7 are from flare SOL2011-09-24T22:38:00 and include the line-core, continuum intensity and line-depth images of a 9$^o$ x 8$^o$ region centered at 12$^o$.82 N and 66$^o$.4 E. Figure 8 shows a plot of the six spectral intensities (filtergrams) before and during the flare at two consecutive pixels located at latitude 12$^o$.56 N (left) and 60$^o$.38 E and 12$^o$.56 N 60$^o$.42 E (right). Temporal plots of the filtergrams show that at some pixels, that absorption line profile gets shallower as the flare progresses, and line core temporarily exceeds the continuum i.e the line goes into emission. Profiles showing the line-core exceeding the continuum were seen in more that one (time) sample at the same pixel.  Figures 8 shows the ``pre-flare'' and ``flare onset'' profiles of two such pixels where the line profile is in absorption before the onset of the flare and then goes into emission during the flare. Plots of the six-point spectra along with the known Doppler velocity in the region of the flare were used to determine the ``true continuum''. Past observations of flares (Babin \& Koval, 2007, Lozitsky, 2007) suggest that while central emission features have been observed in the line-core before, observations of  the entire line going into emission may be rare. However,we have observed similar line emission profiles in the flare SOL2011-08-09T08:05:00 (X6.9)suggesting they may be more common than believed in the past.   
<    
<      Figures 9-12 refer to the flare SOL2010-11-04T23:58:00, GOES  M1.6. Figures 9-11 are a mosaic of the temporal variations in line-core, continuum and equivalent width in a period of one hour including the duration of the flare, from twenty-five pixels chosen from a grid of 20 x 20 pixels centered at. As can be seen, from Figures 9 and 10, the line-core enhancement is significantly more pronounced than the continuum. The equivalent depth is calculated as:
<       E$_w$= L$_d$*L$_w$/I$_c$  * constant. 
<   
<       Temporal variations in the equivalent depth are plotted in Figure 11. (Note, we did not plot equivalent depths for the pixels for flares  SOL2011-09-07T22:38 and  SOL2011-09-24T22:38:00 as the line-width (L$_w$) and line-depth (L$_d$) could not be accurately calculated for some of the the pixels). Examination of the filtergrams for flare  SOL2010-11-04T23:58:00 reveals that the line stays in absorption over the entire duration of the flare. Figure 12 is a typical plot of the the filtergram spectral intensities and show no significant increase in the continuum. 
<    
<    Analysis of all X class flares observed by HMI (nine so far), reveals that an  increase in the continuum is observed is in all the flares, suggesting that all the X-class flares could be categorized at white-light flares. The combination of an overall increase in brightness, the lower base level of line-core intensity, and the reduced absorption in the line implies that the line-core intensity is a more sensitive indicator of the flare than continuum intensity.  
<                                     
---
> The images and plots in Figures~1-4 are from  SOL2011-09-07T22:38:00 (X1.8).  
> Figure~1 is a typical time sequence of tracked and mapped images of the line core during the flare. The images are taken 90~s apart. 
> Figure~2 illustrates the temporal variation in the line-core over a period of one hour (including during the flare) of twenty-five pixels chosen from a grid of 20$\times$20 pixels. 
> Each plot corresponds to a single pixel, and light-curves are plotted for every fifth pixel in the region. 
> The central pixel, labeled (284, 271), is located at $15^\circ$.53 N and $30^\circ$.6 W. 
> The horizontal axis spans 60~min. 
> The central dotted line is  the mean line-core value for that pixel over the 60-min duration, and top and bottom dotted horizontal lines correspond to$\pm 6 \sigma $ from the mean line-core value. Note that these plots are made using tracked ``NRT'' data. As can be seen, a large number of the pixels show greater than a 6-$\sigma$ increase in the line core, indicating a strong enhancement.  
> 
> Figure~3 shows a mosaic of plots of continuum intensity vs time for all the same pixels as Figure~2. 
> However, a smaller number of pixels show a 6-$\sigma$ increase in the continuum brightening as compared to the line core. Figure~4 is a plot of the six spectral intensities observed across the line profile before and during the flare at the central pixel (labeled 284, 271) of the mosaic in Figures~2 and~3. 
> The line profile clearly shows both a significant continuum enhancement, and an extra line-core enhancement. 
> Examination of all the pixels where the line core brightening was observed indicates that the line strength never exceeds the continuum and the line always stays in absorption for this flare. 
> 
> Figures~5-7 are from flare SOL2011-09-24T22:38:00 and include the line-core, continuum intensity and line-depth images of a 9$^\circ$ x 8$^\circ$ region centered at 12$^\circ$.82 N and 66$^\circ$.4 E. 
> The Figures for this flare parallel those of the previous example.
> Figure~8 shows a plot of the six spectral intensities (filtergrams) before and during the flare at two consecutive pixels located at latitude 12$^\circ$.56 N (left) and 60$^\circ$.38 E and 12$^\circ$.56 N 60$^\circ$.42 E (right). 
> Temporal plots of the filtergrams show that, at some pixels, the absorption line profile gets shallower as the flare progresses, and that the line core temporarily exceeds the continuum: the line reverses and actually goes into emission. 
> Profiles showing the line-core exceeding the continuum were seen in more that one (time) sample at the same pixel.  
> Figure~8 shows the ``pre-flare'' and ``flare onset'' profiles of two such pixels where the line profile is in absorption before the onset of the flare, and then goes into emission during the flare. 
> Plots of the six-point spectra along with the known Doppler velocity {\color {blue} need to clarify why the Doppler information is necessary for this - why should it be? -HH} in the region of the flare were used to determine the ``true continuum''. 
> Past observations of flares (Babin \& Koval, 2007; Lozitsky, 2007) suggest that while central emission features have been observed in the line-core before, observations of  the entire line going into emission may be rare. 
> However, we have observed similar line emission profiles in the flare SOL2011-08-09T08:05:00 (X6.9) suggesting they may be more common than believed in the past.
> {\color {red} The flare SOL2010-06-12, as reported by Mart{\' i}nez Oliveros et al. (2011), did not show this behavior.}   
130,141c195,206
< \epsfig{file=imgs/09_24/Lc_58.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_60.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_62.ps,width=0.3\linewidth,clip=} \\
< \epsfig{file=imgs/09_24/Lc_64.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_66.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_68.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_24/Lc_70.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_72.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_74.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_24/Lc_76.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_78.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Lc_80.ps,width=0.3\linewidth,clip=}\\
---
> \epsfig{file=imgs/09_24/Lc_58.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_60.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_62.ps,width=0.29\linewidth} \\
> \epsfig{file=imgs/09_24/Lc_64.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_66.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_68.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_24/Lc_70.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_72.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_74.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_24/Lc_76.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_78.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Lc_80.ps,width=0.29\linewidth}\\
145c210,213
< Brightening observed in the line-core images for SOL2011-09-24T22:38:00.for selected times before and  during the flare. The region shown is a $9^\circ$ x $8^\circ$ rectangle, suitably tracked and mapped to a grid centered at latitude 12$^o$.82 N 66$^o$.4 E. Each image is taken 1 min 15sec apart.}
---
> Brightening observed in the line-core images for SOL2011-09-24T22:38:00 for times before and  during the flare. The region shown is a $9^\circ$ x $8^\circ$ rectangle, suitably tracked and mapped to a grid centered at latitude 12$^\circ$.82 N 66$^\circ$.4 E. 
> The image spacing is 75~s, {\color {red} and the first is at HHMMSS~UT}.
> {\color {blue} again, I suggest making one of these images (the last) a straight intensity plot so that we can see the spots - HH}
> }
151d218
< 
155,166c222,233
< \epsfig{file=imgs/09_24/Ic_58.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_60.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_62.ps,width=0.3\linewidth,clip=} \\
< \epsfig{file=imgs/09_24/Ic_64.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_66.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_68.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_24/Ic_70.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_72.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_74.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_24/Ic_76.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_78.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ic_80.ps,width=0.3\linewidth,clip=}\\
---
> \epsfig{file=imgs/09_24/Ic_58.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_60.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_62.ps,width=0.29\linewidth} \\
> \epsfig{file=imgs/09_24/Ic_64.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_66.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_68.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_24/Ic_70.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_72.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_74.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_24/Ic_76.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_78.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ic_80.ps,width=0.29\linewidth}\\
170c237,238
<  Brightening observed in the continuum (I$_c$ observable for SOL2011-09-24T22:38:00.for selected times before and  during the flare. The region shown is a $9^\circ$ x $8^\circ$ rectangle, suitably tracked and mapped to a grid centered at latitude 12$^o$.82 N 66$^o$.4 E. Each image is taken 1 min 15sec apart }
---
> Brightening observed in the continuum (I$_c$ observable for SOL2011-09-24T22:38:00 for times before and  during the flare. The region shown is a $9^\circ  \times 8^\circ$ rectangle, suitably tracked and mapped to a grid centered at latitude 12$^\circ$.82 N 66$^\circ$.4 E. 
> The image spacing is 75~s, {\color {red} and the first is at HHMMSS~UT}.}
176,178d243
< 
< 
< 
182,193c247,258
< \epsfig{file=imgs/09_24/Ld_58.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_60.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_62.ps,width=0.3\linewidth,clip=} \\
< \epsfig{file=imgs/09_24/Ld_64.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_66.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_68.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_24/Ld_70.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_72.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_74.ps,width=0.3\linewidth,clip=}\\
< \epsfig{file=imgs/09_24/Ld_76.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_78.ps,width=0.3\linewidth,clip=} &
< \epsfig{file=imgs/09_24/Ld_80.ps,width=0.3\linewidth,clip=}\\
---
> \epsfig{file=imgs/09_24/Ld_58.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_60.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_62.ps,width=0.29\linewidth} \\
> \epsfig{file=imgs/09_24/Ld_64.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_66.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_68.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_24/Ld_70.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_72.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_74.ps,width=0.29\linewidth}\\
> \epsfig{file=imgs/09_24/Ld_76.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_78.ps,width=0.29\linewidth} &
> \epsfig{file=imgs/09_24/Ld_80.ps,width=0.29\linewidth}\\
202,208d266
< 
< 
< 
< 
< 
< 
< 
212,213c270,271
<   \epsfig{file=2011.09.24_09:02:15_TAIfiltergrams_for_paper.ps,width=0.40\textheight,clip=}&
<   \epsfig{file=2011.09.24_09:02:15_TAIfiltergrams_for_paper_ii.ps,width=0.40\textheight,clip=}\\
---
>   \epsfig{file=2011.09.24_09:02:15_TAIfiltergrams_for_paper.ps,width=0.46\linewidth}&
>   \epsfig{file=2011.09.24_09:02:15_TAIfiltergrams_for_paper_ii.ps,width=0.46\linewidth}\\
220,226c278,295
< 
< 
< 
<  In the case of the M-class flares, a similar  line-core enhancement was observed in approximately 45 out of the 100 flares investigated, but only 21 of those 45 showed a detectable increase in the continuum. Figure 3 shows the normalized line-core, continuum and equivalent width of a flaring pixel from the M 9.0 flare on July 30th, 2011. There is a strong temporal correlation between the X-ray flux and photospheric effects. The line-core intensity enhancement is again more pronounced that the continuum enhancement. 
< 
< 
< 
---
> In the case of the M-class flares, a similar line-core enhancement was observed in approximately 45 out of the 100 flares investigated, but only 21 of those 45 showed a detectable increase in the continuum. 
> Figure~3 {\color {blue} ? I suggest using the \begin{verbatim}\ref{}, \label{}\end{verbatim} scheme pointing to your figure labels to avoid this sort of confusion - HH} shows the normalized line-core, continuum and equivalent width of a flaring pixel from SOL2011-07-30T02:08:00 (M 9.3).
> {\color {blue} Have I got these details right? There was only one M-class flare that day, it seems, but M9.3 - HH} 
> There is a strong temporal correlation between the X-ray flux {\color {blue} which X-rays? -HH} and the photospheric effects. 
> The line-core intensity enhancement is again more pronounced that the continuum enhancement. 
> 
> Figures~9-12 similarly refer to the flare SOL2010-11-04T23:58:00 (M1.6). Figures 9-11 present mosaics of the temporal variations in line-core, continuum and equivalent width in a period of one hour including the duration of the flare, from twenty-five pixels chosen from a grid of 20 $\times$20 pixels centered at {\color {blue} ?}. As can be seen, from Figures~9 and~10, the line-core enhancement is significantly more pronounced than the continuum. 
> {\color {green} The equivalent depth is calculated as:
>       E$_w$= L$_d$*L$_w$/I$_c$  * constant. }
>  {\color {blue} I suggest redefining it in a simpler way, as given at the top of this section. 
> I think it is the same if $l_c$ refers to the unperturbed continuum, but that is not clear here - HH}.
>   
> Temporal variations in the equivalent width are plotted in Figure~11. (Note, we did not plot equivalent widths for the pixels for flares  SOL2011-09-07T22:38 and  SOL2011-09-24T22:38:00 as the line-width (L$_w$) and line-depth (L$_d$) could not be accurately calculated for some of the the pixels). {\color {blue} I think that any inaccuracies should be described by the error bars - HH}.
> Examination of the filtergrams for flare  SOL2010-11-04T23:58:00 reveals that the line stays in absorption over the entire duration of the flare. 
> Figure~12 is a typical plot of the the filtergram spectral intensities and shows no significant increase in the continuum. 
>    
> Analysis of all X class flares observed by HMI (nine so far), reveals that an  increase in the continuum is observed is in all the flares, suggesting that all the X-class flares could be categorized at white-light flares. 
> The combination of an overall increase in brightness, the lower base level of line-core intensity, and the reduced absorption in the line implies that the line-core intensity is a more sensitive indicator of the flare than continuum intensity. 
254a324
> X-class flares  reveal  that an  increase in the continuum is observed is in all the flares, suggesting that all the X-class flares could be categorized at white-light flares. The combination of an overall increase in brightness, the lower base level of line-core intensity, and the reduced absorption in the line means that the line-core intensity may be a more sensitive indicator of the flare than continuum intensity.  
255a326
> In the case of other M-class flares, a similar line-core enhancement is observed in approximately 45 out of the 100 flares investigated, but only 21 of those 45 showed a detectable increase in the continuum. All the flares show a strong temporal correlation between the X-ray flux and photospheric effects. However, the line-core intensity enhancement is again more pronounced that the continuum enhancement. 
257,261c328,330
<    X-class flares  reveal  that an  increase in the continuum is observed is in all the flares, suggesting that all the X-class flares could be categorized at white-light flares. The combination of an overall increase in brightness, the lower base level of line-core intensity, and the reduced absorption in the line means that the line-core intensity may be a more sensitive indicator of the flare than continuum intensity.  
< 
<    In the case of other M-class flares, a similar  line-core enhancement is observed in approximately 45 out of the 100 flares investigated, but only 21 of those 45 showed a detectable increase in the continuum. All the flares show a strong temporal correlation between the X-ray flux and photospheric effects. However, the line-core intensity enhancement is again more pronounced that the continuum enhancement. 
< 
<     Similar brightenings in the line-core are detected in flares with GOES class greater than C 5.0 and  possibly in a few of the H alpha flares. In many cases, continuum enhancement was not detectable in all the cases where a line-core increase was measurable, but for  all cases with continuum enhancement, the line-core strengthening (and line-depth reduction) was equivalent or significantly larger, suggesting that in general, the line-core enhancement is a far more robust marker for photospheric effects  than the continuum brightening associated with genuine white-light flares. The 45s cadence ''NRT'' HMI observables along with the individual filtergrams provide sufficient spatial and temporal resolution to observe the evolution of the flare.
---
> Similar brightenings in the line core are detected in flares with GOES class greater than C~5.0 and  possibly in a few of the H$\alpha$ flares. {\color {blue} surely they are all H-alpha flares? - HH}
> In many cases, continuum enhancement was not detectable in all the cases where a line-core increase was measurable, but for  all cases with continuum enhancement, the line-core strengthening (and line-depth reduction) was equivalent or significantly larger, suggesting that in general, the line-core enhancement is a far more robust marker for photospheric effects  than the continuum brightening associated with genuine white-light flares{\color {red}, at least for this narrow spectral slice of the continuum.}
> The 45s cadence ``NRT'' HMI observables along with the individual filtergrams provide sufficient spatial and temporal resolution to observe the evolution of the flare.
267a337
> {\color {blue} The location of this pixel should be shown graphically in one of the confusograms, I think. Also, the error bars are overestimated, it appears from their scatter and also because we've estimated 1\% early in the text - HH}
274,279c344,358
< We have detected a photospheric continuum brightening  in a large number of flares. Many of these would traditionally have classified as WLF's. However, the relative enhancement in the line-core profile is significantly stronger than that observed in the continuum and is sometimes detectable even when there is no significant continuum enhancement. The line-core intensity sometimes temporarily exceeds the continuum emission, thus temporarily reversing the absorption line into an emission line. The observed emission may last for over a minute before the line profile returns to absorption.
< 
<    The line-core intensity calculated with the NRT algorithm are more useful that the standard 45s-HMI observables. The flares so observed are spatially and temporally well correlated to the GOES X-ray events, indicating that HMI can be used to systematically study and track the photospheric effect of flares.  
< 
< 
< 
---
> We have detected a photospheric continuum brightening  in a large number of flares. 
> Many of these would traditionally have been classified as WLFs. 
> However, the relative enhancement in the line-core profile is significantly stronger than that observed in the continuum and is sometimes detectable even when there is no significant continuum enhancement. 
> The line-core intensity sometimes temporarily exceeds the continuum emission, thus temporarily reversing the absorption line into an emission line. 
> The observed emission may last for over a minute before the line profile returns to absorption.
> 
> The line-core intensities calculated with the NRT algorithm are more useful that the standard 45s-HMI observables. 
> The flares so observed are spatially and temporally well correlated to the GOES X-ray events, indicating that HMI can be used to systematically study and track the photospheric effect of flares.  
> {\color {blue} Generally WLFs do not correlate as well with GOES soft X-rays as they do with (say) RHESSI hard X-rays.
> This is an important consideration as regards mechanisms (e.g., Hudson 1972).
> It is another statement regarding the identification of the continuum with the impulsive phase of a flare.
> Now that we can look systematically at the line core, and even see that it can reverse, is this true of the line core as well?
> Or, is there a distinction to be made in the time profiles of the continuum and the line core, in that the latter shows more of the gradual phase?
> The best way to find out is to plot some time series and compare with RHESSI.
> Ideally that would be done in this paper, as a survey, but it might be OK to discuss the issue and defer the analysis to a second paper.}
344d422
<