existence of the forerunner as a purely spatial phenomenon, as

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 92, NO. A7, PAGES 7227-7234, JULY 1, 1987

A Search for Forerunner Activity Associated With Coronal Mass Ejections

JUDITH T. KARPEN AND RUSSELL A. HOWARD

E. O. Hulburt Center for Space Research, Naval Research Laboratory, Washin#ton, D.C.

The possible existence of energetic disturbances in the corona significantly before the associated surface events has profound implications for the location and mechanisms of preevent coronal energy storage. Precursor activity associated with coronal mass ejections (CMEs), if it exists, would reflect the evolution and magnitude of the energy release or failure of magnetic equilibrium characterizing the interval before and during the events. Jackson and Hildner (JH) studied 21 CMEs observed with the Skylab white-light coronagraph, and found a low-density plateau rimming each event for which good coverage was available. They concluded that this "forerunner" material could not be explained by mere translation of the overlying coronal plasma. Jackson further inferred from the Skylab data that the forerunner must start moving significantly before the onset of the associated CME, thus identifying this phenomenon as a form of precursor activity. We have performed a systematic search for forerunners using the white-light coronagraph observations obtained with the Solwind instrument on board the P78-1 satellite. Forty-four bright, well-observed events were selected and analyzed, employing selection criteria and analysis methods similar to those used by JH. Approximately half of these events either do not exhibit low-density plateaus in front or are questionable (e•g., a frontal plateau only appears intermit- tently). Based on our analysis of the remaining CMEs, we conclude that identification of the forerunner as a distinct entity probably is not warranted, and that the low-density plasma is an integral part of the

CME itselfi

1. INTRODUCTION

Coronal mass ejections (CMEs) are transient expulsions of solar plasma into the corona and beyond. They are associated frequently with energetic surface phenomena, particularly eruptive prominences [Munro et al., 1979]. A typical mass ejection travels at 500 km s -• and contains approximately 4 x l0 is g of excess material [Howard et al., 1985]. Although plasma ejections from the sun were postulated decades ago, the first significant samples of these events were obtained by spaceborne white-light coronagraphs in the early 1970s [e.g., Koomen et al., 1974; Hildner, 1977].

Approximately half of the events seen by the High Altitude Observatory coronagraph on Skylab assumed the shape of an outwardly moving loop or blob [MacQueen, 1980]. A more detailed morphological classification of the 77 major CMEs observed during Skylab yields approximately 25% loop-type events and 10% filled-bottle events [Munro and Sime, 1985].

Out of the 65 CMEs observed in 1980 by the SMM corona- graph/polarimeter, approximately 80% were classified as loops or bubbles [Waqner, 1984]. In comparison, a similar morphological study of the nearly 1000 CMEs observed by the Solwind white-light coronagraph during 1979-1981 yields a combined total of 28% loop, double-spike, and curved-front events (see Howard et al. [1985] for definitions of Solwind morphological types). These mass ejections probably are not planar loops but rather resemble three-dimensional bubbles, either hollow or filled with excess mass [Howard et al., 1982;

Waqner, 1984]. In general, the loop and curved-front events are the most clearly defined CMEs in terms of discerning an outer edge in excess density in the white-light images. Thus, this type of mass ejection provides the best opportunity to determine the state of the corona ahead of the expelled materi-

al.

Jackson and Hildner [1978] (JH) investigated the morpholo- gy of 21 loop-type CMEs in the Skylab data set, with particu- This paper is not subject to U.S. copyright. Published in 1987 by the American Geophysical Union.

Paper number 6A8673.

lar emphasis on the coronal plasma bordering the primary events. They found a broad rim of excess density above the 2-a level, which they denoted a "forerunner," around the edges of all 18 CMEs for which good coverage was available. The configuration and speed of each forerunner appear similar to those of the underlying CME, so that a nearly constant offset of approximately 1-2 R o separates the outer boundaries of the

forerunner and the CME.

JH first explored the possibility that the forerunner is simply a translation of overlying coronal material pushed from lower heights ahead of the CME, using a simple model of this translation to predict the excess densities that would result. They found that the predicted densities far exceed the forerunner densities actually observed. Two other expla- nations for the additional density in forerunners were con- sidered: preexisting coronal plasma which is compressed in situ, either by driven shocks [e.g., Wu et al., 1978] or by compression of the ambient magnetic field [Mouschovias and Poland, 1978]; or coronal material driven ahead of an impulse which must originate earlier than the CME itself [cf. Jackson, 1981]. These hypotheses represent fundamentally different physical processes and provide important constraints on models of coronal mass ejections. The forerunner phenome- non, then, might be linked to precursor activity as well as to the coronal response to the energy release (or loss of equilibri- um; cfi Low [1981, 1982, 1984]; Wolfson and Gould, [1985]) initiated by certain surface events.

Since the OSO-7 and Skylab era, white-light forerunners have been reported for a few CMEs at most [e.g., Gary et al., 1984]. Other solar phenomena have been observed to precede CMEs spatially or temporally. For flare-associated mass ejec- tions, there is growing evidence that the CME and the flare are initiated separately, although there might be a connection between these temporally correlated phenomena. Weak soft X-ray enhancements have been detected 15-30 min before the onset of flare activity but roughly coincident with the project- ed starting time of mass ejection from the low corona [Harris- on et al., 1985; Simnett and Harrison, 1985]. MHD shocks are thought to produce type II bursts [Wild et al., 1963; Malitson

7227

7228 KARPEN AND HOWARD' A SEARCH FOR FORERUNNER ACTIVITY

TABLE 1. Properties of the Events Studied

Date

Time of

First Number Morpho-

Image, of Speed, logical

UT Images km s- x Type*

1979

April 2 2226 1 150 CF

May 4 0219 3 870 CF

May 8 1028 4 375 CF

May 24 1635 2 1000 L

June 9 1613 2 590 L

July 3 0038 4 580 CF

July 27 0652 6 460 CF

August 14 1306 1 1200 CF

August 16 2246 5 640 L

August 26 0235 9 445 CF

August 26 0235 9 480 L

October 30 1445 10 670 CF

November 1 0656 8 1420 CF

November 15 2239 2 1200 CMPX

November 17 0311 2 890 CF

1980

April 3 0728 3 1100 CF

May 22 2154 8 300 CF

June 12 1237 3 490 CF

June 18 2221 4 520 CF

September 6 0426 4 280 L

September 7 0000 5 700 CF

November 8 0321 3 755 CF

November 17 1122 2 200 CF

1981

January 26 0303 3 1200 CF

February 25 0245 5 715 CF

March 5 0737 4 425 CF

April 10 1135 3 775 CMPX

May 10 0732 5 830 CF

May 10 1229 2 1420 CF

June 27 0928 4 860 CF

August 6 2317 7 245 CF

August 13 2122 5 450 CF

October 18 0330 5 700 CMPX

October 25 0017 5 950 CMPX

November 9 0536 11 490 CF

November 13 0500 5 480 CMPX

November 17 0619 11 CF

November 18 2106 5 900 CF

December 20 0559 5 305 L

1982

January 9 1416 2 840 CF

January 19 0800 10 50 CMPX

July 22 1720 3 1750 CMPX

1984

July 14 0934 10 330 CF

*CF = curved front' L = loop' CMPX = complex [see Howard et al. 1985].

et al,, 1973], which often accompany fast CMEs [Gosling et al., 1974; Sheeley et al., 1984]. Such shocks have been suggest- ed as an explanation for the excess plasma comprising fore- runners [e.g., Dulk et al., 1978; Gary et al., 1984]. Doppler scintillation observations of CME-associated interplanetary shocks indicate that the shock front might travel faster than the leading edge of the white-light CME, and thus might be responsible for a detectable plasma excess ahead of the ejec- tion itself [Woo et al., 1985]. This is not a likely explanation for the Skylab events studied by JH, since most of those

CMEs travelled at low speeds and, therefore, presumably were not shock associated. Further in-depth study of a much great- er sample of events is required, however, to establish a con- nection (if any) between forerunners and other CME- associated activity.

The purpose of the study presented here is to address the

existence of the forerunner as a purely spatial phenomenon, as

section 4.

2. OBSERVATIONS AND DATA ANALYSIS

We briefly summarize the instrumental and data-set charac- teristics, as detailed descriptions of the Solwind coronagraph are published elsewhere [-Sheeley et al., 1980' Michels et al., 1982]. Routine observations of the corona by the Solwind instrument began on March 28, 1979, and ended on Septem- ber 13, 1985. The field of view extends roughly from 2.5 to 10 R 0, with spatial resolution of 1.25 arcmin. The interval be- tween images is 10 min, for the most part, with occasional periods in which images were obtained every 5 min during the 1-hour daylight portion of the 93-rain orbit. The duty cycle and intermittent gaps of no coverage are described more thor- oughly by Howard et al. [-1985].

We have reproduced the analysis procedures of JH as close- ly as possible, given the differences between the Skylab and Solwind coronagraphs and between the methods used to reduce the initial data from each instrument. In this way, our operational definition of the forerunner phenomenon has been confined to an established standard. JH used the following basic format in their forerunner search. The coronal images obtained by Skylab originally were on film. The images con- taining a CME were digitized, calibrated, and digitally sub- tracted from a base image taken at an earlier time. The excess brightness images first were averaged over 5 x 5 neighboring pixels to smooth out excess noise' the spatial resolution thus achieved was 120 by 120 arcsec. The resultant difference images were converted from units of excess brightness to excess column density (cf. Hildner et al. [1975-] for details), and contour plots were made. The outer boundary of the fore- runner was defined arbitrarily as the 6 x 10 -9 gcm -2 con- tour level of excess column density, which is the 2-a level of noise in the difference images. The outer boundary of the forerunners reported by JH actually extended beyond this level and faded monotonically into the background. In con- trast, the majority of CMEs in their sample were characterized by a sharp outer edge which occurred approximately at the 50 x 10 -9 g cm-2 level. Therefore, JH adopted this contour of excess mass as the working definition of the boundary be- tween the transient and the forerunner. Radial scans through

KARPEN AND HOWARD' A SEARCH FOR FORERUNNER ACTIVITY 7229

-7

i i I • i

4 5 6 7 8 9

Solar l•adius

Fig. 1. The 2-or noise level in terms of column mass density vs.

distance from the center of the sun for Solwind coronagraph images.

the approximate center of each event also were made, to show the run of excess column density with height above 1.6 R o.

In our survey of the Solwind coronagraph data, we used the existing set of difference images on Polaroid film to identify promising events for further scrutiny. For proper registration, the two images used to make a difference image are tempo- rally separated by an integer multiple of the satellite orbital period. We chose the smallest possible multiple for which a good base image could be obtained, to minimize the distrac- ting influence of long-term changes in the background coronal structure. The digital data were reduced on a VAX 11/730 computer, with an IIS image processing and display system used to determine the radial-cut locations and contouring areas. Each pair of images was calibrated in units of coronal

brightness before differencing. As was done by JH, we smoothed the difference images by averaging over 5 x 5 neighboring pixels. The smoothed difference images were dis- played for identification of the radial cut lines and the portion of the image to be contoured. Then, these data were converted to units of excess column density [cf. Poland et al., 1981]. To produce contour plots with minimum noise, the data were filtered before contouring using the algorithm of Lee [ 1981].

Our selection criteria were as follows:

1. The event intensity must be average or bright, accord- ing to the classification scheme outlined by Howard et al.

[1985].

2. The morphological type must be a loop, curved front, or complex (as long as criterion 3 is satisfied), according to the classification scheme outlined by Howard et al. [1985].

3. The event must have a clearly visible front throughout most, if not all, of the observation period. In all but a few cases, the position of the leading edge at the initial time of observations was less than about 4 R o.

4. The outer portions of the event are not obscured by one of the polarizing rings (located at approximately 5 and 8 Ro).

For events with good coverage (i.e., several images), this cri- terion was obeyed by discarding those images in which the region ahead of the CME is in one of the polarizing rings.

Events for which most or all of the images violate this cri-

terion were not used.

Using these criteria, we selected 44 CMEs for in-depth study. The observation intervals and observed characteristics for these events are listed in Table 1. Two plots were produced for each image: a contour plot of the excess mass density in the region containing the CME, and a plot of the excess den- sity in and ahead of the CME along a radial cut from the sun's center to approximately 10 R o. The same radial cut was

JULY 22 '82 N N

' I I

N N

I 137 UT

...

1720

JULY 23

1730 1740

I 1749

$ I 1759 t OO02

$

I OO02U:T

$

Fig. 2. White-light images of the coronal mass ejection of July 22, 1982, observed with the Solwind coronagraph. All images except for the first (top left) and last (bottom right) are difference images. The field of view of the images extends from the outer edge of the occulting disk, at about 2.6 R o, to 8 R o.

-w

--w

7230 KARPEN AND HOWARD' A SEARCH FOR FORERUNNER ACTIVITY

lOO

8o

6o

4o

2o

i i i i

20 40 60

X (•ro,',',i-)

\\

do '

Fig. 3. Isodensity contours of excess column density (in gcm -2) for the July 22, 1982, event at 1720 UT, taken from a portion of the second image shown in Figure 1. The contour levels are drawn at 0, 6, 56, 106, 156, 206, 256, 306, 356, 406, 456, and 506 x 10 -9 gcm -2 Values above 506 x 10 -9 gcm -2 are not shown. A portion of the radial cut used to produce the density-height plot (Figure 3) is shown by the dotted line through the CME.

used for all images of each event, and traverses a dense, well- defined portion of the CME.

Because the forerunners reported by JH are very low den- sity structures, it is extremely important to establish the in- strumental noise level before searching for such features. If the noise level were too high, then the forerunner would be indis- tinguishable from the image background. JH defined the front of the forerunner to be at the 2-• level of noise in the Skylab data, AI/I= 0.01, which corresponds to an excess column

density of 6 x 10 -9 gcm -2. We have determined the instru-

Figure 1 displays the Solwind 2-• noise level in units of column mass density (gcm- 2) as a function of radial distance from the sun. To convert the noise level from intensity to column-density units, we assumed that the noise in the pre- viously created difference image is due entirely to differences in K-coronal brightness. This enables a direct conversion to electron density and then mass density Fcf. Poland et al., 1981;

Howard et al., 1982]. We then computed the noise at each radial position by averaging over all position angles around the occulting disk. By comparing our Figure 1 with Figure 4 of JH, we see that the Solwind and Skylab noise levels are quite comparable at all radial distances.

3. RESULTS

To illustrate the characteristics of the CMEs in our data set, we present in detail the results of our analysis for two of the 44 mass ejections. These CMEs occurred on February 25, 1981, and July 22, 1982.

We first observe the July 22, 1982, event at 1720 UT, when the white-light front is already at 3.5 R o (Figure 2). Figure 3 is a plot of isodensity contours for the July 22, 1982, CME at 1720 UT. There is little evidence for forerunner activity (as defined by JH) ahead of the CME, although more of a plateau is evident around the sides. The spiky nature of the lowest

contour (at 6 x 10 -9 gcm- 2) is due to the presence of coronal

The February 25, 1981, event has a different and more com- plex appearance. The primary feature, located nearest the center of the image (Figure 5), is the CME, while the narrower feature below and to the right is a streamer which was pushed aside by the mass ejection. The isodensity contour plot (Figure 6) and radial-cut plot (Figure 7) both show a density profile similar to those of the events studied by JH: a broad ledge of material at the 6-56 x 10 -9 g cm-2 level extends around the front and sides of the denser ejected material.

Six (,--14%) of the events studied resemble the July 22 CME in the absence of apparent forerunner material in all available images. Another 16 events (,--36% of the total) are marginal in one or more ways. For some of these CMEs, the forerunner plateau appears in some of the available images but not in others. The tendency for CMEs to originate at the base of a helmet streamer also prevents definitive identifi- cation of a forerunner in several cases, since the excess density of the activated streamer is in the same range as that of the forerunner material. Finally, if a low-density plateau does appear, it is usually broader at the sides of the event than at the front, leading to a "marginal" classification for those events which manifest a questionable amount of forerunner material ahead but prominent plateaus to the sides.

To determine why only certain events are associated with forerunners, we looked for specific characteristics which dis- tinguish the events with forerunners from those without.

Those events which we classified as "marginal" were not in-

600 7/22/82 17-20 UT

5OO

400

3OO

200 _

100 _

0 .

--100

0 2 4

Height (Solar radii)

Fig. 4. The profile of excess density vs. distance from the center of the sun for the July 22, 1982, event at 1720 UT, along the radial cut partially shown in Figure 3. The entire radial cut starts at sun center and ends around 10 R o.

KARPEN AND HOWARD: A SEARCH FOR FORERUNNER ACTIVITY 7231

FEB 25, 1981

E •

oI3gUT

I I

:.

02_45-0 I I OUT 02_55-0 I I 9UT

E m

... .. ..:

....

:

..

0,305-0 129UT I 03 I 5-0148UT I 032.4-0148UT I

s s s

Fig. 5. White-light images of the coronal mass ejection of February 25, 1981, observed with the Solwind coronagraph.

All images except for the first (top left) are difference images. The field of view of the images extends from the outer edge of the occulting disk, at about 2.6 R o, to 8 R o.

-w

-w

cluded in the following analyses. We found no apparent dis-

tinction between forerunner and nonforerunner events in

terms of associations with other solar activity (e.g., soft X ray, H-alpha, and radio emissions), according to Solar Geophysical Data Prompt and Comprehensive Reports (1979-1984). JH found that the Skylab forerunners were associated with CMEs of greatly different speeds. Although both our forerunner and nonforerunner events also exhibit a wide range of character- istic speeds, the average speed of the Solwind forerunner events (-• 400 km s-x) is only half that of the nonforerunner events (-•800 km s-X). This difference is due to the relative lack of nonforerunner events with low speeds ( < 400 km s- x).

The hypothesis that forerunners may be shocked plasma obvi- ously is inconsistent with these observations--one would expect more forerunners to be associated with fast events, not less, if this hypothesis were true. Due to the small number of CMEs in each sample, 'however, we cannot draw any further conclusions about the underlying significance of this apparent velocity difference.

4. DISCUSSION

It is important to recognize that we do not find forerunners associated with every event in our sample, in contrast to JH.

For the half which is ostensibly consistent with the CME-plus- forerunner scenario of JH, however, we must consider the fol- lowing questions in order to determine whether or not the forerunner is physically distinct from the CME.

1. Is this low-density plateau statistically significant ? 2. What is the location of the front of the CME, as deter- mined from the difference images, relative to the column den- sity levels shown in the relevant contour plots? By definition, the material comprising the forerunner must lie ahead of the main body of the CME (as we define through visual determi- nation of the front).

3. Can gradual evolutionary changes of the pre-CME corona account for the low-density plateau seen in the differ- ence images ?

4. Is this material associated exclusively with CMEs, or do other coronal phenomena exhibit the same type of density

distribution ?

Although the outer edge of the forerunner is at the noise level, the inner edge is well above this level. In some cases, a break appears in the slope of the density distribution as a function of radial height. For others, a smooth, monotonically decreasing profile is seen. In either case, this low-level density excess is statistically significant for the present set of events (see section 2).

For all but a few of the CMEs investigated in this work, visual inspection of the difference images yields positions for the CME fronts that coincide roughly with the 6 x 10 -9 g cm -2 level of excess density in the contour plots (compare Figures 2 and 3, or 5 and 6). In other words, this level is plainly visible in all cases where the images are of reasonable quality, i.e., the image is free of a high background level or

7232 KARPEN AND HOWARD' A SEARCH FOR FORERUNNER ACTIVITY

• oo

•0 [•

•- 6O

2O

0 0

•: 45 UT

. ß /

/

//

20 40 80 O0

X (arerain)

Fig. 6. Same as Figure 3 for the February 25, 1981, event at 0245 UT, taken from a portion of the second image shown in Figure 4.

data drops. Therefore, the images themselves provide no a priori reason to attribute special characteristics to the CME plasma at densities between 6 and 50 x 10 -9 g cm-2

We also subtracted different base images from the event images discussed earlier (February 25, 1981, 0245 UT, and July 22, 1982, 1720 UT), to see whether the characteristics of the base image can account for the appearance of a forerunner ahead of an event. We expected to see a monotonic increase or decrease in the excess density around the CME, as progres- sively earlier images were subtracted from the event images, due to coronal evolution over a duration up to 1 day before the event. Instead, the results were rather confusing, indicating that the corona does not necessarily evolve in a smooth manner over a 1-day interval.

To answer the last question, we return to the February 25 event. Note that the activated streamer exhibits a density pro- file much like that of the nearby CME, with its own

"forerunner"-like plateau (see Figure 6). In fact, if we compare the CME difference images with the contour plots, we find many examples of coronal streamers that are visible at or above the 6 x 10 -9 g cm -2 excess-density level. The fact that these streamers are observed in the difference images indicates

700

600 _

500

400

2:45 UT

• 300

• -

• 200

• -

• 100

• -

0 ,

--10•

• ' ,

O 4 6 8

Height (Solar radii)

Fig. 7. Same as Figure 4 for the February 25, 1981, event at 0245 UT, for the radial cut partially shown in Figure 6.

that additional mass has been introduced into the preexisting structures or that the streamers themselves have moved (as for the February 25 event). The latter situation is easily dis- tinguishable from the other because a "depletion" will appear in the difference image at the original location of the affected material (see Figure 5). In any case, it is clear that CMEs are not the only coronal structures with a low-density plateau at the outer edge.

To further investigate the effects of nearby coronal stream- ers on the appearance of a CME, we examined the undif- ferenced images taken before each of the forerunner and non- forerunner events (the marginal events were excluded). Nearly all of these events occurred in close proximity, if not cospatial with, one or more streamers. A high percentage, but not all, appear to be "bisected" by a helmet streamer. The three non- forerunner events without streamers are low-latitude events.

The three forerunner events without streamers all are high- latitude events, on the other hand, in which case any nearby streamers might be invisible due to projection effects. We con- clude that coronal streamers are ubiquitous companions to most CMEs, and do not influence the presence or absence of forerunner activity. Coronal streamers also were quite common during the Skylab era, so the differences between the present results and those of JH cannot be attributed to differ- ences in coronal-streamer rates.

It is unrealistic to expect that the coronal mass distribution changes only at the location of the CME. Rather, the observed changes most likely reflect the general evolution of the corona as part of the same energetic process which launches the CME. Closer to the CME, the overlying coronal material must undergo mass motion or compression (or both) to adjust to the passage of the ejecta. For a given low-density level, we cannot distinguish between the excess mass associated with the CME and that which originated in the undisturbed corona. Thus, some of the observed plateau might be due to, e.g., the associated evolution of the coronal magnetic field or to the "plowing up" of the ambient plasma. This interpreta- tion is weakened by the low percentage of forerunner identifi- cations in the Solwind data set, since coronal evolution should accompany all events. Furthermore, although the "snowplow"

•oo 2/25/81 2:45 UT

8O

ß •- 60_

20

0 20 40 60 80 100

X (arcrnin)

Fig. 8. Isodensity contours of excess column density (in g cm -2) for the 1981 February 25 event at 0245 UT taken from a portion of the second image shown in Figure 5. The contour levels are at 0, 6, 18, 54, 162, and 486 x 10 -9 g cm -2 (i.e., 6 x 3"). Values above 486

x 10-9 g cm-2 are not shown.

KARPEN AND HOWARD: A SEARCH FOR FORERUNNER ACTIVITY 7233

effect is most probable for the highest velocity CMEs, we see no correlation between event velocity and the presence of fore- runner material. Of course, we cannot rule out the existence of forerunner activity at a level below 6 x 10 -9 g cm-2, or with a thickness smaller than the Solwind spatial resolution. Com- parison between observations and numerical models of CMEs [e.g., Steinolfson, 1986] should aid in determining the relative contributions of the CME and ambient components to the white-light images.

A low-density ledge appears ahead of half of the events studied, at most, so we consider a possible explanation which does not rely on alteration of the ambient plasma. A contour plot of the density distribution in coronal structures should reflect the form of the distribution; if not, misleading infer- ences can be obtained. For coronal streamers, the radial den- sity profile derived from white-light coronagraph observations appears to be a power law of the form r-", where n •--3.8, along the radial distance r from the solar surface [e.g., Bohlin et al., 1971; Saito, 1972]. Therefore, we examined the conse- quences of assuming a power-law density distribution for the CMEs. For most events studied in the present work, the low- density ledge disappears if the contour plots are drawn with levels spaced according to a power law. To illustrate this con- tention, we have replotted the February 25 event with density levels spaced by powers of 3 (Figure 8). This form was chosen for general consistency with the streamer density profiles derived by Saito [1972], but does not necessarily reflect the true density distribution in the CME. Although fewer con- tours are shown, it is clear that this plot differs significantly from Figure 5: the contours are more evenly spaced (particu- larly in the displaced streamer), and there is no sign of a frontal plateau.

We suggest that the plateaulike morphology of the lowest

Acknowledgments. This research was supported in part by the Air Force Geophysics Laboratory, through MIPR FY71218500006. The authors benefitted from their participation in the Workshop on Solar Events and Their Influence on the Interplanetary Medium, organized by NOAA under the auspices of the Solar Maximum Mission Prin- cipal Investigators. We wish to thank E. Hildner, B. Jackson, and N.

R. Sheeley, Jr., for valuable discussions about the present study. We also thank both referees for instigating substantial improvements in the manuscript.

The Editor thanks B.C. Low and another referee for their assist- ance in evaluating this paper.

REFERENCES

Bohlin, J. D., M. J. Koomen, and R. Tousey, Rocket-coronagraph photometry of the 7 March, 1970 corona from 2 to 8.5 Rs, Sol.

Phys., 21,408-417, 1971.

Dulk, G. A., S. F. Smerd, R. M. MacQueen, J. T. Gosling, A. Magun, R. T. Stewart, K. V. Sheridan, R. D. Robinson, and S. Jacques, White light and radio studies of the coronal transient of 14-15 September 1973, Sol. Phys., 49, 369-394, 1976.

Gary, D. E., G. A. Dulk, L. L. House, R. Illing, C. Sawyer, W. J.

Wagner, D. J. McLean, and E. Hildner, Type II burst, shock waves, and coronal transients: The event of 1980 June 29, 0233 UT, Astron. Astrophys., 134, 222-233, 1984.

Gosling, J. T., E. Hildner, R. M. MacQueen, R. H. Munro, A. I.

Poland, and C. L. Ross, Mass ejections from the sun: A view from SKYLAB, J. Geophys. Res., 79, 4581-4587, 1974.

Harrison, R. A., P. W. Waggett, R. D. Bentley, K. J. H. Phillips, M.

Bruner, M. Dryer, and G. M. Simnett, The X-ray signature of solar coronal mass ejections, Sol. Phys., 97, 387-400, 1985.

Hildner, E., Mass ejections from the solar corona into interplanetary space, in Study of Travelling Interplanetary Phenomena, edited by M. A. Shea, D. F. Smart, and S. T. Wu, pp. 3-21, D. Reidel, Hingham, Mass., 1977.

Hildner, E., J. T. Gosling, R. T. Hansen, and J. D. Bohlin, The sources of material comprising a mass ejection coronal transient, Sol. Phys., 45, 363-376, 1975.

Howard, R. A., D. J. Michels, N. R. Sheeley, Jr., and M. J. Koomen, The observation of a coronal transient directed at Earth, Astrophys.

, J. Lett., 263, L101-104, 1982.

density contour, which JH denoted a forerunner and Jackson Howard, R. A., N. R. Sheeley, Jr., M. J. Koomen, and D. J. Michels, [1981] identified as precursor activity, might be explained in Coronal mass ejections: 1979-1981, J. Geophys. Res., 90, 8173-

distinction between the forerunner and the underlying CME has been exaggerated by the choice of data-display pro- cedures, and does not necessarily reflect a genuine coronal disturbance ahead of the CME. In addition, the overlying coronal gas and ma. gnetic field should evolve as part of the same energetic process which forms and ejects the CME.

Therefore, we conclude that the apparent plateau of low- density material can be explained by a combination of the nonlinear density profile and the existence of background coronal structures which presumably are "activated" in con- junction with the CME. This explanation appears satisfactory for all of the events analyzed in the present work, over twice as many events as were studied by JH.

Searches for forerunners in other coronagraph data sets would be very useful to resolve this discrepancy between the Skylab and Solwind results. At present, the SMM Corona- graph/Polarimeter is the only other spaceborne instrument which has observed a significant number of CMEs during the present solar cycle. We suggest that such a search should be repeated when future experiments have collected a sufficiently large sample of bright, well-defined CMEs of the appropriate morphological types. In addition, because the Solwind data is still being reduced, we intend to continue our investigation as new CMEs are discovered. Events observed by both Solwind and the coronagraph/polarimeter would be especially valuable in eliminating possible biases due to the different detector designs and data-reduction techniques.

8191, 1985.

Jackson, B. V., Forerunners: Early coronal manifestations of solar mass ejection events, Solar Phys., 73, 133-144, 1981.

Jackson, B. V., and E. Hildner, Forerunners: Outer rims of solar coronal transients, Sol. Phys., 60, 155-170, 1978.

Koomen, M. J., R. Howard, R. Hansen, and S. Hansen, The coronal transient of 16 June 1972, Sol. Phys., 34, 447-452, 1974.

Lee, J. S., Refined filtering of image noise using local statistics, Comput. Graphics Image Process., 15, 380-389, 1981.

Low, B.C., Eruptive solar magnetic fields, Astrophys. J., 251, 352-363,

1981.

Low, B.C., Self-similar hydrodynamics, I, The 7 = 4/3 polytrope and the coronal transient, Astrophys. J., 254, 796-805, 1982.

Low, B.C., Self-similar magnetohydrodynamics, IV, The physics of coronal transients, Astrophys. J., 281, 392-412, 1984.

MacQueen, R. M., Coronal transients: A summary (Discussion), Phil.

Trans. R. Soc. London, Ser. A, 297, 605-620, 1980.

Malitson, H. H., J. Fainberg, and R. G. Stone, A density scale for the interplanetary medium from observations of a Type II solar radio burst out to 1 Astronomical Unit, Astrophys. J. Lett., 183, L35-38, 1973.

Michels, D. J., R. A. Howard, M. J. Koomen, and N. R. Sheeley, Jr., Satellite observations of the outer corona near sunspot maximum, in Radio Physics of the Sun, edited by M. R. Kundu and T. E.

Gergely, p. 439, D. Reidel, Hingham, Mass., 1980.

Michels, D. J., N. R. Sheeley, Jr., R. A. Howard, and M. J. Koomen, Observations of a comet on collision course with the Sun, Science, 215, 1097-1102, 1982.

Mouschovias, T. Ch., and A. I. Poland, Expansion and broadening of coronal loop transients: A theoretical explanation, Astrophys. J., 220, 675-682, 1978.

Munro, R. H., and D. Sime, White-light coronal transients observed from SKYLAB May 1973 to February 1974: A classification by apparent morphology, Sol. Phys., 97, 191-201, 1985.

7234 KARPEN AND HOWARD' A SEARCH FOR FORERUNNER ACTIVITY

Munro, R. H., J. T. Gosling, E. Hildner, R. M. Mac•ueen, A. I.

Poland, and C. L. Ross, The association of coronal mass ejection transients with other forms of solar activity, Sol. Phys., 45, 201-215,

1979.

Poland, A. I., R. A. Howard, M. J. Koomen, D. J. Michels, and N. R.

Sheeley, Jr., Coronal transients near sunspot maximum, Sol. Phys., 69, 169-175, 1981.

Saito, K., Photometric and polarimetric analysis of the coronal streamers observed at the March 7, 1970 Mexican eclipse, Ann.

Tokyo Astron. Observ., Second Set., 13, 93-148, 1972.

Sheeley, Jr., N. R., R. A. Howard, D. J. Michels, and M. J. Koomen, Initial observations with the Solwind coronagraph, Astrophys. d.

Lett., 237, L99-101, 1980.

Sheeley, Jr., N. R., R. T. Stewart, R. D. Robinson, R. A. Howard, M.

J. Koomen, and D. J. Michels, Associations between coronal mass

ejections and metric type II bursts, Astrophys. d.i •9, 839, 1984.

Simnett, G. M., and R. A. Harrison, The onset of co•ronal mass ejec-

tions, Sol. Phys., 99, 291-311, 1985.

Steinolfson, R. S., White-light brightness in looplike csronal tran-

sients, Bull. Am. Astron. Soc., 18, 710, 1986.

Wagner, W. J., Coronal mass ejections, Annu. Rev. Astron. Astrophys.,

22, 267, !984. '

Wild, J.P., S. F. Smerd, and A. A. Weiss, Solar bursts, Ann. Rev.

Astron. Astrophys., 1, 291-366, 1963.

Wolfson, R., and S. A. Gould, Th e onset of coronal mass ejections,

Astrophys. •l., 296, 287-293,'1985.

Woo, R., J. W. Armstrong, N. R. Sheeley, .!r., R. A. Howard, M. J.

Koomen, D. J. Michels, •nd. R- $ch•wen•n, Doppler scintillations observation• of interplanetar y shocks within' 0.3 AU, J. Geophys.

Res., 90,154-162, 1985. • '

Wu, S. T., M. Dryer, Y.•,Nalcagawa, and S. M. Han, Magnetohydro-

dynamics of atmospheric transients, •II, Two-dimensional numerical results for a model solar c0rona•, •strophY s. •I., 219, 324-335, 1978.

R. A. H6ward and J. T. K•rpen, E. O. Hulburt Center for Space Research, Naval•Research •ab0ratø•ry, Washington, D.C. 20375.

(Received Augu•st 1, 1986'

revised •anuary 14, 1987'