LOCALIZED PLASMA DENSITY ENHANCEMENTS OBSERVED IN STEREO COR1

When viewing difference movies of coronagraph images one is often left with the impression that there is unresolved material flowing away from the Sun. Tappin et al. (1999) performed a correlation analysis on low-latitude data from the Large Angle and Spectrometric Coronagraph (LASCO) C2 and C3 coronagraphs aboard the Solar and Heliospheric Observatory (SOHO) and found a significant correlation indicating outward radial flow. In addition, structured material has also been directly observed by Sheeley et al. (1997) in the form of plasma blobs flowing away along streamers. The authors of that work treated the density enhancements as passive tracers of solar wind speed through the C2 and C3 fields of view. By measuring the heights of these features as a function of time and combining many such measurements, they were able to plot an approximate curve of solar wind velocity versus height above the Sun. Given this application, it is important to understand the nature of the blobs themselves so that we can determine if and where they are passively carried by the ambient solar wind and where they may be undergoing some acceleration due to other phenomena.

Wang et al. (2000, 1998) have proposed three possible origins for the observed inhomogeneities: streamer evaporation, magnetic reconnection at the tips of distended streamers, and reconnection between closed and open field lines within the streamer body. While streamer evaporation may account for some of the events, it is problematic for several reasons. It requires a steady addition of new loops into the streamer to be consistent with the observed stability of coronal streamers, it leads to an increase in the solar magnetic flux, and it is inconsistent with the shapes of plasma blob observations in the outer corona (Wang et al. 1998). Thus, it seems that streamer detachment or footpoint exchanges are the more likely origins for the observed plasma blobs.

Additional observations at lower altitudes can help us distinguish between these two explanations. A two-dimensional model of a streamer disconnection (the second proposed origin) is presented by Wu et al. (2000), and shows the formation and propagation of several blob-like plasmoids. These plasmoids are not unlike small flux ropes recently observed by in situ solar wind instruments (Feng et al. 2007; Mandrini et al. 2005; Moldwin et al. 2000). However, the magnetic islands produced in the model tend to form above the streamer tops, near or above 2 R_☉. Footpoint exchanges between open and closed field lines, on the other hand, seem more amenable to the formation of blobs over a range of heights in the lower corona.

Understanding the physical mechanism by which the blobs are released into the solar wind is critical for determining their reliability as a tracer of ambient wind speeds. Using the inner coronagraph, COR1 (Thompson et al. 2003), of the twin STEREO spacecraft (Kaiser et al. 2008), we can make height–time coronal images (the so-called J-maps; Sheeley et al. 1999). The COR1 field of view (FOV) extends as low as 1.4 R_☉, so using this instrument we can lower the threshold altitude at which blobs are observed (previously 2 R_☉).

In the next section, we show some examples of the observed density enhancements and discuss the image processing techniques used to help identify them. We also show some general characteristics of these events in the COR1 FOV based on height–time measurements for a large sample. The final section gives our interpretation of these measurements.

Observations of plasma blobs in inner coronagraphs are complicated by multiple effects. First, by observing closer to the solar surface we introduce higher instrumental light scattering. In order to see coronal features we must devise some way of removing the stray light background from our images. In COR1 this is typically accomplished by forming an image from the minimum value of each pixel over the course of a day, and then taking the median value of each pixel over a month of such daily backgrounds to obtain a one-month background.

However, using this method we still face the problem that the lower corona contains relatively stationary features like streamers that are bright enough to hide the fainter blobs we are interested in. To remove those features we use a special running-difference technique, where the background subtracted from each image is the minimum value of each pixel in a small window (approximately 1.5 hr) of images around it. This has the advantage of subtracting features like streamers that are stationary on the timescale of days so that faster-evolving features can be clearly seen, while tending to produce a less noisy image than a regular (single-frame) running-difference technique.

Finally, even using the running-background subtraction it can be difficult to see small features that expand against a large, noisy background (see Figure 1). To simplify this task we remove an angular slice centered on the feature's trajectory, subtending 5° latitude in the plane of the sky, from each frame in our difference movie. We integrate the slice over angle (following Sheeley et al. 1999; dal Lago et al. 2003) and stack the resulting pixel columns sequentially to form a height–time image. The angular integration increases the visibility of the feature as it moves out through the corona. In fact, many features are observed in height–time plots that are very difficult to see in the running-difference movies, although a careful observer can usually predict where they will be seen. Figure 2 shows some examples of COR1 height–time images; they have been enhanced to show feature locations more clearly using the Radon transform, similar to the Hough transform used by the CACTus automated coronal mass ejection (CME) detection program (Robbrecht & Berghmans 2004). While the Radon transform can be used to help us confirm the general location of a faint track, it also smears out areas of strong signal over linear paths; it can give the false impression that the feature does not undergo acceleration, or that it has broken up into diverging tracks. For these reasons all data analysis was done using measurements from the original running-difference, height–time images.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Using running-difference images does introduce some ambiguity in the interpretation of our observations. Because a different background image is subtracted from each frame, it is difficult to distinguish relative brightness changes due to reorganization of the coronal plasma from changes in the subtracted background image. In order to show that the features we are seeing are real and not artifacts of the differencing process, Figure 3 shows an event seen in both COR1 running-difference and COR2 base-difference images. The top image shows sections of several COR2 images, with arrows indicating the feature of interest. The lower left image shows the same feature in a COR1 height–time image, enhanced using the Radon transform. The bottom right shows the COR1 and COR2 height measurements plotted together as a function of time; the measurements below 4 R_☉ are from COR1 and those above are from COR2. Despite the ambiguity introduced by using running-difference images, we have been able to confirm that the track seen in the COR1 height–time image corresponds to a real, traveling density enhancement observed in COR2.

Zoom In Zoom Out Reset image size

In order to discern the general character of these events in the COR1 FOV, we divided 19 days' worth of COR1 data (recorded 2007 March 3–2007 April 2, 2007 September 4–2007 September 8, and 2008 March 27–2008 April 2) into 5° angular bins. We made height–time images of each angular bin and measured the trajectories of as many events as possible. We chose the above dates to contain as few CMEs as possible, as these complex events tend to be seen multiple times (as different features) in many bins, so that they would likely skew the overall statistics. (For example, a CME seen on 2007 March 31 contributed 11 separate "events" out of a total of 27 for that day, all with similar trajectories.) Those CMEs that did occur within the observational periods were carefully removed from the data set. We also combined the height–time measurements for apparently collinear events in adjacent angular bins. These combined events most likely represent features that either move nonradially from one angular bin to another, or have angular widths greater than the 5° bin size.

Occasionally a feature could be seen in the height–time image, but its height could not be reliably measured in more than two time steps. This typically happened because the trajectory was ambiguous due to noise or apparently overlapping features. These events were also excluded from the data set.

After combining related events and excluding CMEs, a total of 453 upward-moving events were found. Their height–time profiles were compared to both a line and a parabola using a least-squares fit, with an assumed uncertainty in every height measurement of seven pixels. Figure 4(a) shows the radial plane-of-sky speed distribution of the events, as determined from the linear fits. The mean speed as measured in this fashion was 240 km s⁻¹.

Zoom In Zoom Out Reset image size

The results of the parabolic fits divide the events into three groups: those with negative acceleration (96 features), those with positive acceleration (132 features), and those whose acceleration was smaller than the fit error (225 features). Figure 4(b) shows the radial POS acceleration distribution for the events with acceleration greater than the fit error. The mean acceleration for this set was 20 m s⁻². Of the events whose acceleration was measurably negative, all trajectories fit parabolas whose maximum height was within the COR1 FOV. This is most likely because of the fairly large error bars, which limit the negative accelerations we can measure to relatively large values.

Few events were visible all the way to the edge of the COR1 FOV and none were observed to reverse direction. We did not observe any downward-moving features during the days examined, consistent with previous observations that inflows are rarely seen during times of low solar activity (Sheeley & Wang 2002, 2001; Wang et al. 1999).

Figure 5 shows the distribution of events among the different angular bins. The angular position each bin is centered on is measured from 0° at the right-hand side of the image, so that solar north is at 90° and the solar equator is at 180° and 360°. The measured distribution is consistent with the emergence of the events from near or within the streamer belt during the time periods studied.

Zoom In Zoom Out Reset image size

In Sheeley et al. (1997), the authors combine speed and height measurements of many observed blobs and fit the collection to two curves, one assuming constant acceleration and the other assuming an acceleration that decays exponentially with altitude. The dependence of the speed on the height in these two situations is given by

$$\begin{equation} v^2=2a(r-r_1) \end{equation} \tag{ 1 }$$

and

$$\begin{equation} v^2=v_a^2(1-\exp (-(r-r_1)/r_a)), \end{equation} \tag{ 2 }$$

respectively, where r₁ is the height at which the speed goes to zero and r_a is a length scale of the decay in the acceleration. Using Equation (1) they obtain r₁ = −0.4 R_☉ and a = 3.4 m s⁻², and using Equation (2) they obtain v_a = 298.3 km s⁻¹, r₁ = 2.8 R_☉, and r_a = 8.1 R_☉.

Figure 6 shows (height and speed) data obtained from our COR1 height–time measurements. The instantaneous speeds in this plot are determined from the parabolic fit parameters, and only events with positive acceleration are included. Because some of our density enhancements (particularly the slow-moving ones) were detected in more images than others, we weighted these data points by the number of (height and speed) pairs contributed by the corresponding event when fitting the data. The solid line represents a least-squares fit of this data to Equation (1), with r₁ = 1.27 R_☉ and a = 20.2 m s⁻². The dashed line is the same equation using the parameters given by Sheeley et al. (1997), and the dash-dotted line shows the fit of their data to Equation (2).

Zoom In Zoom Out Reset image size

Our observations show that many events similar to the blobs seen by Sheeley et al. (1997) are seen in the COR1 coronagraph at POS heights below those seen in the LASCO C2 coronagraph. We observe slight density enhancements with narrow angular extent moving away from the Sun, with trajectories consistent with those seen in C2. A comprehensive search for these features was undertaken on 19 days' worth of COR1 data, resulting in 453 events. The POS height–time profiles obtained from these measurements indicate speed and acceleration distributions which are reasonable for this region of the corona. The events are distributed in the plane of the sky in a manner consistent with an origin in or near the streamer belt.

These new observations provide further insight into the origin of the blobs. Based on fits to their data and the relative scarcity of observations below 3 R_☉, Sheeley et al. concluded that the blobs likely originate above the tops of streamers in the 3–4 R_☉ range. Now, using an instrument with a smaller occulter, we have been able to make many additional observations between 1.5 R_☉ and 3.5 R_☉. The velocity profile of our events indicates a larger acceleration and a lower starting height for the density enhancements. Also, our results are more consistent with a constant acceleration than those of Sheeley et al. (1997), although this is probably due to the fact that we are sampling a much smaller range of heights in our FOV.

Sheeley et al. (1997) observed many more events at low speeds in the 2–3 R_☉ range than we did. Indeed, they detected a wide range of speeds in this altitude range (approximately 0–200 km s⁻¹), while in our data there were fewer events with very low speed at this height. This may be a result of the increased visibility of these features near the occulter's edge. Here, we would expect to see more events outside the POS (whose apparent speeds are lower but whose actual speeds are comparable to those in the POS) and additional genuine low-speed features, which can be difficult to see in difference images. Because the COR1 occulter is smaller than that of C2, the cluster of low-speed events is detected at a lower altitude. Given this difference between the two data sets it is not surprising that the Sheeley et al. (1997) fit to Equation (2) does not match our data very well.

The observation of these blobs in the COR1 FOV provides a valuable insight into their origin, and hence to their suitability as a solar wind tracer. The shift of the low-speed events to lower altitudes with the use of a smaller occulter suggests that they do not originate primarily in the 3–4 R_☉ range as previously thought. This lends some support to the idea that these events may be the result of reconnection between closed and open field lines, and presents an interesting challenge for modelers attempting to explain them as the result of reconnection events near the streamer tops.

We are grateful to O. C. St. Cyr and W. T. Thompson for useful discussions concerning this work.

Facilities: STEREO (COR1)