A type U burst is observed within a group of type III bursts which start near 300 MHz (Fig. 4 (click here)). Some of them occur nearly simultaneously with the type U.
The source configuration at 164 MHz shows type III bursts at several sites scattered between channels 0 and 2. The differential radio spectrum (Fig. 5 (click here)) provides an unambiguous identification of the sources in the iso-brightness contours: the burst at 11:39:17 UT is the ascending type U branch. The type III burst at 11:39:19 UT comes from slightly west of this ascending branch. It coincides in the spectrum with the ascending type U branch - in the differential radio spectrum it is distinctly visible between the top of the type U burst and the low frequency boundary of the spectrogram. At the time of the descending branch (11:39:23 UT) two sources are seen nearly simultaneously: the spectrum shows another type III burst which occurs at 164 MHz slightly after the descending U branch. This timing allows us to identify the weaker western source in the contour plot (channel 6) as the descending U source. The brighter type III emission comes from the site of the earlier type III bursts. The inversion line of the large-scale photospheric magnetic field separates the sources of type III and ascending type U emission from the site of the descending type U branch. The sources are north of small-scale soft X-ray structures, but have no direct counterpart in the Yohkoh image, perhaps because of its relatively low sensitivity.
Several type U bursts are observed during and up to 4 hours after
H
-flare-associated type IV continuum and type III burst groups.
The first U burst is
identified at 10:07:47 UT (about 17 min after the hard X-ray flare onset)
within one minute of the H
and soft X-ray maximum
(Fig. 6 (click here) top after background subtraction, which yields gaps at
the frequencies of terrestrial transmitters). Although the spectrogram and the
contour plot display a complex structure, the timing shows that
the brightest 435 MHz sources during the two branches are the sites of type U
emission (Fig. 6 (click here) bottom). The sources at both 435 and 410 MHz
(not plotted in order to avoid confusion) project to opposite sides of the
inversion line of the large-scale photospheric magnetic field. They are emitted
in a loop which bridges the inversion line in the east-west direction (see also
Sect. 2.7 (click here)).
The U burst at 11:44:15 UT belongs to a group of fast-drift bursts (type III and U/J events) with turnover frequencies in the range 150 to 110 MHz. The interval shown in Fig. 7 (click here) comprises, in order of increasing time, a type J burst, the rising branch of the type U, a type III burst which merges with the descending branch of the type U near 150 MHz. The weaker burst in the contour plot at 11:44:18.6 is the descending branch of the type U. Again the type U burst sources project to opposite sites of the photospheric neutral line, but of a different section than the 10:07:47 event, and with beam injection occurring above a region of opposite magnetic polarity.
The U burst pair at 12:31 UT (Fig. 8 (click here)) is an example of a frequently occurring association with a type III which continues the U below its spectral turnover. At 236 MHz the source at the time of the ascending branch can be separated into the two components plotted as solid crosses in the bottom panel of Fig. 8 (click here). One of these sources is likely the high-frequency part of the type III burst. The descending branch of the first U burst is observed both at 236 MHz (large dashed cross) and at 327 MHz (smaller dashed cross). These sources are located north of the photospheric magnetic inversion line, while the ascending branch and the type III bursts are south of it. The second U burst of the pair has the same source configuration, but its descending branch is much weaker.
At 13:31:45 UT a U burst appears together with type III emission during both
its rising and descending branch (Fig. 9 (click here)). The initial type III
burst, which extends to frequencies as high as 340 MHz, is followed by a sudden
flux density increase below 240 MHz. The high-frequency envelope of the
subsequent emission until 
13:31:49.5 UT has the typical type U behaviour
(we stress, however, that the classification of the different spectral features
is not unambiguous).
The event was observed by the NRH at three frequencies: at 164 MHz, a single burst traces emission near the turning point of the Type U, while at 236 and 327 MHz two sources are distinguished: on the one hand the initial type III burst at 327 MHz and the rising U branch at 236 MHz, on the other hand the descending branch of the type U. A weak signature of the initial type III burst is seen at 236 MHz (immediately before 13:31:45 UT, at a position slightly west of the much stronger emission from the rising type U branch). A further type III burst after the descending branch of the type U is seen at 236 MHz near 13:31:50 UT, at a position close to (but different from) that of the descending type U burst branch.
The centroid positions at different frequencies are plotted by different symbols (one for each frequency) at the bottom of Fig. 9 (click here). The extent of the sources, which is similar to the preceding events, is not shown in order to avoid confusion. The configuration is fully consistent with a simple loop structure bridging the inversion line of the photospheric magnetic field and neighbouring open flux tubes near the site of beam injection: the initial type III burst (south of the inversion line) and the descending branch of the type U at 327 MHz (both plotted as diamonds) enclose the sources at lower frequencies, i.e. the sources of the two branches of the type U at 236 MHz (triangles) and the single source at 164 MHz (square), which is emitted near the spectral turnover frequency.
A sequence of several type III/U bursts was observed in the range 100 and 400 MHz between 11:20 and 11:30 UT. Spatially and spectrally related emission features have been noticed earlier (from 09:47 - 09:49 UT) and spuriously later between 40 and 170 MHz (in the afternoon hours). We present three type U bursts out of the mentioned interval (Figs. 10 (click here) to 12 (click here)). Note that the SXT image contains some overexposed pixels in the active region of interest.
The spectrum (Fig. 10 (click here)) shows an asymmetric type U burst turning near 220 MHz. The radio image at 236 MHz shows two closely neighbouring sources as expected when the imaging frequency is near to the turning point (compare e.g. Fig. 15 (click here)).
Figure 11 (click here) gives a comparable type U burst at higher frequencies (turnover 255 MHz). The NRH observes only the rising branch at 327 MHz. We present this figure because the high starting frequency provides a better tracing of the link with the underlying active region, both since the plasma level is at a lower height in the corona and because the radio propagation is less disturbed by the ionosphere. So we corroborate the association between radio source positions and soft X-ray features made for the other type U bursts of this day which are only observed at lower frequencies.
The spectrum in Fig. 12 (click here) is the unique case we found in our data
where a type U burst seems to occur as a pair of harmonically related emission
features. The top frequency ratio of 
is nearer to the 3:2 than
to the 2:1 ratio. There is no spectral indication of emission at frequencies
below 100 MHz which could be identified as the ``true
fundamental" of this event. From an analysis of the turnover frequency ratio of
a sample of 42 U and J bursts Stewart (1974) derives a range of 1.4 to 2.2 for
the frequency ratio between harmonic and fundamental emission (mean value
). The present case is within this range. Further,
Takakura &
Yousef (1974) give evidence for 3:2 ratios of related type J burst features. We
refer to the low frequency branch as fundamental and to the high frequency
branch as harmonic in the following, but these are to be understood as purely
descriptive terms. As expected from the faint spectral signature, at 236 MHz
the source of the harmonic burst can be located during peaks of emission only.
The source configuration is discussed in detail in Sect. 2.6 (click here)
The U burst (Fig. 13 (click here)) is accompanied by a group of type III bursts starting (together with the ascending U branch) near 280 MHz. The type III bursts are weak below 170 MHz.
Peculiar features of the U burst are a bright patch of 5 MHz bandwidth and 1.7 s duration near spectral turnover and a broadening of the descending branch above 250 MHz. The sources in the bottom of Fig. 13 (click here) are, from east to west, the ascending U burst branch at 236 MHz, then at 164 MHz, and the descending branch at 164 MHz. The emission at 10:52:59 UT actually starts at a site near the ascending branch, but one second later a brighter source appears farther west. The spectrogram allows for the interpretation that the early emission (10:52:59-10:53:00 UT) is a type III burst which occurs together with the descending U branch. This would explain the apparent broadening of the latter above 150 MHz. The U burst is followed by a more slowly drifting type III burst (10:53:06 UT) which comes from the vicinity of the ascending U branch source. The U burst sources at 164 MHz have opposite polarization: 4% left-handed during the ascending branch, 6% right-handed during the descending branch.
Probably due to the low sensitivity of the Yohkoh SXT images available for this
day we find no significant soft X-ray features near the radio source positions.
The relative location of the radio sources and the active regions is compatible
with an interconnecting structure between the two active regions located near
E55 S05 and E17 S10. There is no Stanford magnetogram published in SGD on
January 25, but a large quiescent filament (e.g. Boulder H
images,
SGD 594-I) suggests that the descending U branch comes from the opposite side
of the inversion line of the photospheric field than the ascending branch and
the type III bursts.
The spectrum in Fig. 14 (click here) shows a type U like burst with an emission gap after a strong brightening of the ascending branch immediatly before the turning point. Alternatively, the spectrum can be considered as a succession of a type III (but with an unusual kind of low frequency cut-off in the spectrum) and a reverse drift burst. Both burst branches are equally and about 8% right-handed polarized at 164 MHz.
The ascending and the descending sources are close to each other,
despite the fact that the 164 MHz observing frequency is well above the
turnover frequency of 130 MHz. The observations at 164 MHz reveal a similar
source structure and evolution during the two burst elements. Each source is
embedded in a halo of faint emission (visible e.g. after the peak of the second
burst in the contour plots of Fig. 14 (click here)) which spreads both
north-eastward and south-westward from the site of brightest emission,
eventually covering a range of 
. The sensitivity of the
analogue spectral record does not permit to identify the spectral
characteristics of the spreading emission.
Comparison with the Yohkoh SXT image shows both sources lie above active regions near the limb. Due to projection effects, we cannot decide if the two spectral branches belong to different loop legs. The spreading of the radio emission during the decay phase of the two bursts at 164 MHz occurs transverse to the soft X-ray loops, such as if streams of electrons were injected into a cone of diverging field lines, i.e. a system of east-west oriented loops with a wide range of inclinations to the radial direction. The spectral gap at the turnover frequency suggests that either the radio emission of the beams is cut off near the loop summit or that both branches are due to distinct episodes of beam injection, with the first one producing only an ascending branch, the second one only a descending branch. Type U-like bursts with missing or very faint ascending branches were indeed observed on other occasions (cf. Fig. 25 (click here)).
The type U burst shown in Fig. 15 (click here) is a rather isolated, unpolarized burst. The rising branch is also observed at 435 MHz with the NRH. Since the NRH observing frequency 327 MHz is near the turnover frequency, no separate sources are seen, but a gradual transition from the ascending to the descending branch site (the eastward broadening of the contours starting at the time of maximum brightness). The rising branch comes from above the western limb at 327 MHz (the bigger solid cross) and at 435 MHz (the smaller solid cross). The descending branch at 327 MHz is on the disk, on the opposite side of the large-scale photospheric inversion line. This and the Yohkoh SXT image point to sources in a complex of loops surrounding two active regions near the limb.
A sequence of two type U bursts has been observed within a group of type III bursts in the range 170-400 MHz between 11:34 and 11:39 UT (Figs. 16 (click here) and 17 (click here)).
The two type U bursts have similar spectra: both turn over near 200 MHz, are brighter near the turning point (up to about 1.25 times the turning frequency), and the spectral signature of the rising branch is sharp, while the descending branch appears more diffuse. The 11:38 UT type U burst is followed by a fast-drift burst with a peculiar spectral structure: at frequencies below the end of the preceding descending U branch the drift is negative (type III-like), while no or even a reverse drift is visible at higher frequencies, almost up to the starting frequency of the ascending U branch.
Both type U bursts display a similar spatial pattern. Three sites of emission can be distinguished:
The ascending branch of the U burst (large solid cross at 236 MHz, smaller
solid cross at 327 MHz) comes from a site between ARs NOAA 7434 (near S15 E65)
and 7433 (
N15 E45), while the descending branch as well as the second
rising branch are situated on the opposite side of the large-scale neutral
line, near the following part of AR 7432 (
S15 E20) where the subflare is
reported. The beam injection site is therefore not directly related with the
subflare (cf. Table 1 (click here)).
The comparison of the type U radio source sites with soft X-ray features provides an intriguing result: the radio sources are aligned along a faint soft X-ray structure connecting ARs NOAA 7433 and NOAA 7432. But the source position of the 327 MHz ascending branch, which should be closer to the site of beam injection, rather suggests the south-eastern AR NOAA 7434 as a host for the beam injection site. So do the type III bursts (not shown in the figures) which come from a source east of the rising type U branch and are further displaced from the soft X-ray feature in the direction to AR 7434.
The spectrum in Fig. 18 (click here) shows a well-defined U burst with a strong type V continuum tail after its descending branch (see also Sect. 2.8 (click here)). In the frequency range 145- 170 MHz two type III-like bursts are seen between the rising and the descending branch. The radio emission in the range 100-250 MHz belongs to a complex event (dm and m activity earlier, metric type II activity later) related to a flare behind the western limb.
Three radio sources are distinguished at 164 MHz. During the ascending branch two sources brighten in close succession (channels 20 and 14). Both were already acting during the preceding activity. Since the two sources make a comparable contribution to the 164 MHz flux density during the ascending branch, both are plotted as solid crosses upon the Yohkoh-SXT image (Fig. 18 (click here), bottom). The descending branch and the continuum occur at the same place (dashed cross) at lower projected altitude than the ascending branch. The big source size is due to the superposition of the new emissions on the broad and complex ongoing emission.
Besides the confusion with the ongoing event, the observing conditions at low
elevation in winter are unfavourable to get a clear picture of the source
configuration and its association with soft X-ray features. When projected on
the sky, the two baselines of the NRH form an angle significantly different
from 90
. As a consequence, errors in the one-dimensional positioning
yield bigger uncertainties of the heliographic positions than in the other
events studied. Residual effects of ionospheric gravity waves add to the
uncertainty. As a result, the relative positioning between radio sources and
soft X-ray features is subject to an uncertainty of 0.1-0.2 
. Within
this restriction, the type U sources project upon opposite sides of a
large-scale loop system above the south-western limb.
The spectrum (Fig. 19 (click here)) shows a strong type III burst followed by a
fainter type U burst with lower starting frequency than the type III.
The U burst is followed by a faint, short brightening which gives the aspect
of a type U(N) burst. However, the spectral record does not reveal a clear
finite drift rate. The brightness distribution shows the position of the
initial type III burst until 12:12:42 UT. The subsequent peak is slightly west
of the type III position. We identify this site as the source of the ascending
branch of the type U burst and attribute the westward displacement to its
brightening at a place close to, but different from, the preceding type III
burst (projected distance 0.04 
). The descending branch has its
source west of the rising branch. The following source (12:12:50 UT) is close
to the preceding descending branch. This confirms its identification as a
type U(N) burst (like Fig. 17 (click here)).
Comparison with the Yohkoh image and the magnetic inversion line reveals that the sources are situated in an active region loop overlying a system of soft X-ray loops. The type U burst occurred in conjunction with a soft X-ray jet in the trailing part of the active region, underneath the sites of emission during the type III burst and the rising branch of the type U (Aurass et al. 1994).
The type U burst (Fig. 20 (click here)) has well-defined ascending and descending branches and a bright and diffuse extension around the turnover point and the beginning descending branch. A faint third branch (possibly U(N)) is seen in the low frequency part of the spectrum (110-130 MHz) after 9:03:20 UT. At frequencies below the turning point of the type U the ascending branch is continued as a faint type III burst until less than 40 MHz. The narrow-band bursts seen in the spectrogram are associated with a noise storm which is well separated from the U burst at 164 MHz.
The 164 MHz emission has two well separated sources. The rising branch comes from a source which appears double in itself. The dominant component lies above the south-eastern limb. After a detailed comparison with other bursts on the same day we conclude that the secondary source is a sidelobe due to noise storm emission at a remote site. The descending branch comes from a distinct simple source north-eastward of the rising branch. After its end a new rise is visible at the same site from 09:03:15 to 09:03:20 UT, which might be the high-frequency part of the above mentioned faint third branch of the burst. In this case the event would be of type U(N).
The radio sources lie above AR's 6321 (
S25 E50) and 6322 (
S20 E60)
near the south-eastern limb as seen by superposition on the daily image of the
Meudon spectroheliograph. The relative position of the two
U sources is compatible with radio emission from a large loop structure that
interconnects AR 6321 with AR 6322 or with a quiet region in the photosphere.
The type U burst is observed among a diffuse background of type III bursts (Fig. 21 (click here)) which start above 170 MHz and partly extend to below 100 MHz. The background emission is interrupted e.g. during the U burst at frequencies below the burst trace. Like in the case of 14 October 1990 the turning region is much brighter than both branches, but there is no type V like continuum extension.
Two source sites are distinguished during the onset of the burst (channels -18 and -14). Both were already acting during the preceding type III emission. The brightness maximum of the eastern source (08:50:49 UT) fits in time with the spectral signature of the ascending U branch. The analogue spectral record does not indicate the emission of the western source site due to its relative weakness. The descending type U burst source is most intense and consists of a close succession of two brightenings at channels -13.75 and -13.5 (08:50:57 UT). All radio sources are unpolarized.
The combined plot of type U source positions and the spectroheliogram is inconclusive as to the associated loop structure.
In Fig. 22 (click here) a cluster of type J and U bursts is shown with turnover frequencies in the range 130 MHz to below 100 MHz.
The dominant emission at 164 MHz comes from three successive negatively
drifting bursts (13:13:30- 13:13:42 UT). They are also observed at 236 MHz. The
first of these bursts turns below 110 MHz, the following two merge with type U
bursts which turn slightly above 130 MHz. The position and size of the
brightest burst at 164 MHz (13:13:34 UT) is plotted by the solid cross in the
bottom panel of Fig. 22 (click here) upon the H
spectroheliogram.
The burst at 13:13:40 UT is at the same position, while that at
13:13:37 UT is situated 
0.02 
north-east. The sources have
asymmetric profiles in the east-west scans. This means that at least two
neighbouring sources contribute to each of the bursts or that there is the same
kind of spreading as on 03 June 1993. After the fade-out of the negatively
drifting bursts at 164 MHz, a faint signature of the descending branch of one
of them becomes visible (13:13:45 UT) east of the source of the negatively
drifting bursts (channel -22 in the iso-brighness representation of the
east-west NRH branch; not strong enough for the amplitude curve; dashed cross,
bottom of Fig. 22 (click here)). Its heliographic position is not accurately
determined because the NRH data compression procedure does not detect
significant changes in the north-south scans between 13:13:42 and
13:13:44.5 UT. Weak burst activity continues afterwards near the site of the
previous negatively drifting bursts.
We cannnot decide which of several active regions near and behind the limb are connected by the loop structure hosting the type U source.
Within 3h 40min five similar burst groups have been identified in the range 100-170 MHz. The spectra in the top of Figs. 23 (click here) to 27 (click here) show mostly interlaced type III and type U bursts.
While the spectra are quite different, the NRH data reveal a similar source configuration in all cases. The ascending branches and type III bursts come from an eastern source (channel -23 to -21). These sources have asymmetrical contours that extend toward the later appearing descending branch source near channel -15. The descending branch cannot always be identified in the spectral records, but is the dominant feature of the 9:53:30 UT event. A faint brightening is simultaneously observed with the descending site source at the position of the ascending source.
The bursts are emitted in basically the same large-scale coronal loop structure (see also Sect. 2.7 (click here)). The sources of the negatively drifting bursts are somewhat scattered in position. Most of them are intrinsically complex. The source of the descending branch undergoes an apparent shift away from the source of the rising branch.
Both timing and source geometry suggest that the rising branch source of the
type U bursts is connected with the flaring active region NOAA 6197. Given
the
north-westward displacement of the descending branch source with respect to the
rising branch, the small plage region at N37 E40 visible in the H
spectroheliogram (bottom of Figs. 23 (click here) to 27 (click here)) is a
plausible site for a conjugate footpoint of a large-scale magnetic loop.
No significant polarization is observed during the ascending branch of type U bursts. Only in the 13:32 UT event about 10% left-handed polarization is detected during the descending branch.
Figure 4: Spectrum and source configuration during a type U burst. From top
to bottom:
1) Dynamic spectrum (linear frequency scale).
2) One-dimensional brightness distribution (integrated in the
direction perpendicular to the array). Vertical axis: 0 = disk centre, total
field of view 64 channels. Linear contours, 20% - 100% of maximum.
3) Flux density within 
0.25 channels around the position of maximum
brightness (solid: ascending branch, dashed: descending branch).
4) Yohkoh-SXT image (13:28:29 UT,
AlMg filter, 7.5 s exposure), radio source positions and sizes (cf.
Fig. 2 (click here)) and inversion line of the large-scale photospheric
magnetic field (Solar Geophysical Data)
Figure 5: The time derivative of the 4 November 1994 event spectrum shown in
Fig. 4 (click here), top, clearly reveals a type III burst crossing the
turning point of the type U burst
Figure 6: NRH contour
levels: linear spacing, 10% - 100% of maximum brightness, after subtraction
of a pre-event scan. Yohkoh image: 0953:19 UT, filter AlMg, exposure time
30 s composed with earlier less exposed images. Compare also
Fig. 3 (click here)
Figure 7: NRH (north-south branch) contours: linear, 4.3% - 100%
of maximum brightness
Figure 8: NRH (north-south) contours: NRH linear,
10% - 100% of maximum brightness
Figure 9: NRH (north-south) contours: linear, 10% - 100% of maximum
brightness. Bottom: centroid positions of the radio sources at
164 MHz (square), 236 MHz (triangles), 327 MHz (diamonds; southern source =
initial type III burst, northern source = descending U branch)
Figure 10: NRH contours: linear, 20% - 100% of maximum brightness. Yohkoh:
10:53:54 UT, AlMg, 5.3 s exposure time
Figure 11: NRH contours: linear, 4.3% - 100% of maximum brightness. Yohkoh:
cf. Fig. 10 (click here)
Figure 12: NRH (north-south) contours:
linear, 20% - 100% of peak brightness, after subtraction of a
pre-event scan). Yohkoh: cf. Fig. 10 (click here)
Figure 13: NRH contours: linear, 10% - 100% of peak brightness. Yohkoh:
1103:51 UT, AlMg, 0.7 s exposure
Figure 14: NRH contours: linear, 4.3% - 100% of
the peak value. Yohkoh: 1257:49 UT, AlMg, composed of
a 30 s exposed picture with earlier ones of shorter exposure time
Figure 15: NRH contours: logarithmic, 5% - 100% of maximum brightness.
Yohkoh: 0923:53 UT, AlMg, 0.7 s exposure
Figure 16: NRH contours: linear, 4.3% - 100% of maximum brightness. Yohkoh:
1254:12 UT, AlMg, 30 s exposure, composed with earlier less sensitive
images
Figure 17: See caption of Fig. 16 (click here)
Figure 18: NRH contours: logarithmic, 5% - 100% of maximum brightness.
Yohkoh: 0701:41 UT, AlMg, composed of a
30 s exposure and earlier less sensitive images
Figure 19: NRH contours: logarithmic, 1% - 100% of maximum brightness.
Yohkoh: 1213:12 UT, Al.1, 2.7 s exposure. The image reveals a soft X-ray jet
starting one minute before electron beam injection in the ascending loop leg
Figure 20: NRH contours: linear, 4.3% -
100% of maximum brightness. Meudon H
spectroheliogram: 0858 UT
Figure 21: NRH contours: linear, 4.3% - 100% of maximum brightness.
H
: 0902 UT
Figure 22: NRH contours: logarithmic, 1% - 100% of maximum brightness.
H
: 0611 UT
Figure 23: NRH contours: linear, 4.3% - 100% of maximum brightness.
H
: 0705 UT
Figure 24: See caption Fig. 23 (click here). The two
simultaneous sources during the descending branch are confused in the scans
with the north-south branch of the NRH. Only the position of the ascending
branch is plotted in the bottom panel
Figure 25: See caption Fig. 23 (click here)
Figure 26: See caption Fig. 23 (click here)
Figure 27: See caption Fig. 23 (click here)