R O C K E T O B S E R V A T I O N OF T H E EUV I M A G E S OF A S O L A R

F L A R E AND A C T I V E R E G I O N S

T. H I R A Y A M A , K. T A N A K A , T. W A T A N A B E , K. A K I T A , T. S A K U R A I , and K. N I S H I

Tokyo Astronomical Observatory. University of Tokyo, Mitaka, Tokyo I81, Japan

(Received 20 August; in revised form 10 October, 1984)

Abstract. Images of a flare and active regions were obtained in the extreme ultraviolet emission lines such as C uI 977 A, NevIII 770 A, and H I L/3, and hydrogen Lyman continua with a spatial resolution of less than ten seconds of arc together with one-dimensional scanning at 1650 A. A microchannel plate was used as a detector, and pointing accuracy was, for about half of the observation time, around 0.5 arc sec.

The relationship between the shape of the flare and the structure of the photospheric magnetic field is discussed. A map of the electron temperature distribution derived from the intensity ratio of the Lyman continua at 880 A and 815 ,~ showed a lower temperature in regions of higher activity. A very small geometrical thickness of 50-500 m in the C Ill emitting region of the flare was found. And the layer emitting the continuum in 1650/k is shown to be at a temperature of 5300 K in the flare and 4700 K in active regions.

1. Introduction

We present EUV observation of a solar flare and active regions by a Japanese sounding rocket (S 520-5CN). The instrument consists of a Cassegrain telescope and a stigmatic spectrograph, where a microchannel plate (MCP) and a photomultiplier were used as focal plane detectors. During 10 rain flight, decay phase of a flare (optical class 1B and X-ray class C3) was observed. EUV images of this flare and active regions were obtained in CIII 977 A, Nevm 770 ,~, HI L/31ines, and HI Lyman continua at 880 A. and 815 A, etc. together with one-dimensional scan at 1650 A. While EUV pictures of some flares have previously been reported from the Skylab, much can be learned by studying various types of flares (see a review by Moore et al., 1980). Though the resolution was not as good as the film, technically it has been demonstrated that the MCP we used is very useful in this wavelength region. And the fine pointing control system we developed, attaining at a 0.5" accuracy, can be used for further experiments.

In this paper descriptions and observations are given in Section 2, and the results from one-dimensional scan at 1650 .~ and from the EUV images are presented in Sections 3 and 4, respectively. In Section 4, we discuss, briefly, on a map of electron temperature distribution and on the geometrical thickness of EUV emitting plasmas.

2. Instrument and Observation

(a) Optics (Figure 1 and Table I): The telescope is a classical Cassegrain of an aperture of 10 cm (F/15). The secondary mirror is supported on a zimbal structure which enables fine pointing and raster scanning. The 50 cm stigmatic spectrograph makes spectral foci

Solar Physics 95 (1985) 281-295. 0038-0938/85.15 �9 1985 by D. Reidel Publishing Company

282 T. HIRAYAMA ET AL.

TABLE I

Specifications of the instrument

Telescope:

Spectrograph:

MCP:

Photomultiplier:

Miscellaneous:

Resolution:

10 cm aperture classical Cassegrain. Focal length: 1.5 m. Back focus: 10 cm. Magnifi- cation: 4.6. 1" = 7.3 ~tm. Substrate: Zerodur. Gold-coated. 2/4 at 0.5 gm. Effective collecting area: 66 cm 2 (40 mm light shading circular plate on the secondary mirror). Zimbal support at the secondary mirror.

50 cm length stigmatic spectrograph. Concave grating: 1200 gr m m - 1, 700 A blaze, Pt-coated, and R = 498.1 ram. Incident angle: 10.5 ~ Distance between the entrance slit and ~ating: 489.8 mm. Central wavelength: 860 ,~ (MCP). Reciprocal dispersion: 16/~ m m - ~. Entrance slit: 30 lain x 30 ram.

two-dimensional. Tandem and ion-feedback free ( • 10.3 ~ tilted micro-channels to the main light beam). High voltage: 2100 V. 26 mm J~ (Hamarnatsu Photonix Co.). Resistive anode: 1.4 x 1.4 inch (Surface Science Eng. Co.). Maximum count: 4 x 104 cps. Local saturation level: 200 cps/pixel.

Hamamatsu Photonix R1081. Effective wavelength: 1150-2000 A.

weight of the instrument: 46.5 kg. Electric power: 25 W. Telemetry rate: 54.4 kbps.

spatial: nominal: 5.5"(raster direction) x 5.2"; effective (incl. photon count number): 10" x 10".

Temporal: 1 s (one picture: 100 s). Spectral: 4 A (line) and 16A (continuum).

(horizontal foci) on a group of slits on the Rowland circle and after selecting emission lines and hydrogen Lyman continua it makes spatial foci (vertical foci) on a micro- channel plate and a photomultiplier (16 ,~ m m - 1). The effective width of the entrance slit on a 45 o tilted mirror is 30 ~tm.

Lines selected on the exit slits are NaIx 680.7 A, N e v m 769.6 ,~, CIII 976.6 ,~, SixIII 499.5 ,~ (2nd order), HI Lfl 1025.4 A, and OvI 1031.7 A with a slit width of 4 A and with a slight tilt of 2.1 ~ as shown in Figure 1. This tilt will give a larger tolerance for the relative positioning of slits and emission lines against possible dislocation by shocks at the time of rocket firing. Note that OvI 1031.7 A is recorded in the same channel as H I Lfl. Slit length for Si xn 499.5 A is set short because it was feared to be contaminated from neighboring Lfi and CIII. Hydrogen Lyman continua are selected at 732.8 A, 815.0 ,~, and 880.0 A without tilt because of wide slits of a 16 A width.

Spatial resolution of the present observation is largely limited by the resolution of the detector (< 7" = 50 lam), and partly by defocusing and tilting of the secondary mirror (w_ 2"). And the overall resolution of ~ 10" seems to have been attained in the active region where enough photon counts were received.

(b) Detectors: The main detector is tandem microchannel plates of resistive anode type (MCP, see Table I for specification). The resistive anode counts the number of photons coming in each of 256 pixels in the direction of slit length (5.2" each on the Sun) and in each of 8 channels in different wavelengths. Eighty-four pixels out of 256 were cut off from the Sun's light to see scattered light and noise. Quantum efficiency of the MCP, and reflectivity of a gold-coated mirror and of a platinum-coated concave

~<

350r

am

+ 48

9.8m

m

-

Sun/

s j

I L

ight

Qua

dran

t Fi

ne

Sens

or

Fig.

1.

I' 1

Sens

or

~o,,~

?

~t~

.~,, ,

...... .

.4 ~.,

,~T/~.

~,.,, ~

~.,

~1,,.,,

, M

CP

Exi

t S

lit

~/{/

I II

It II

II /11

1 ~ I

t....

Yaw

(Lau

nche

r) cd

A_kV

LJ

O [J

LI

; )d

~

Sche

mati

c layo

ut of

the i

nstru

men

t. MT~

and M

T2, m

icrom

otors f

or til

ting t

he se

cond

ary m

irror

M 2.

MTs

, raste

r sca

n mot

or. B

elow r

ight

, exit s

lits.

284 T . H I R A Y A M A E T A L .

grating in the first order were measured at our observatory, and they are shown in Figure 2. The absolute calibration of the MCP was performed using a windowless photodiode calibrated at the National Bureau of Standard in the U.S.A. via a photo- multiplier coated with sodium salicylate acting as intermediary. A photomultiplier (PM in Figure 1) was used for the measurement of the integrated intensity over a slit length of 11.5' at 1650 .& with F W H M of 83.7 A. The absolute sensitivity of the PM was measured by using a photodiode calibrated also at N.B.S.

Fig. 2.

I I I I I I I I

10-1 ~ ' ~

/ / / , ' " "hk'- Pt Grating ( 1st ,~ /

�9 ~ MCP ~2

lO-a

400 800 1200 1600

Wave length

Quantum eMciency of the MCP, and reflectivity of the telescope mirror (single mirror measurement) and the grating against wavelength e, incident angle in the measurement.

(c) Fine pointing: While the attitude control of the rocket body was conducted by the launching agency, attaining to + 2 arc rain accuracy, we ourselves were responsible for the fine pointing, the technique of which emerged from our experience on solar balloon telescopes (Hirayama, 1978): see detail in Nakatani et al. (1983). Result of the pointing was within 0.5 arc sec in the E - W direction during the whole observation time of 270 s.

EUV FLARE IMAGING 285

However in the N - S direction a half of the observation time showed quasi-periodic large

excursions of the order of + 10 arc sec. Fortunately since the raster scanning was performed in the N - S direction and since we know the deviation from the exact pointing

through the fine sensor, it is possible to correct these large excursions. Although we did not have any device for fine pointing around the optical axis, it was within one second of arc.

(d) Observationalprocedure: The rocket (a diameter of 52 cm and a length of 8.8 m) was launched on September 6, 1982 at 02h00m00 s UT from the Kagoshima Space Center, reaching a maximum altitude of 237 km at 246 s after firing. At 151.3 s the

observation was started at a fixed position near the Sun center for 30 s. From 180 s (= 02h03m00 s, 218 km) the raster scan was started by changing continuously the

position of the Quadrant Fine Sensor block at a rate of 5.5" s - l for 100 s. The

usable observation contains 2.0 raster scans of 9.2 arc min in length. The direction of the scanning was 2.8 ~ t ired to the east from the solar rotation axis and the scan went

first from south to north. (e) Flare and active regions: An 1B flare was occurring during the observation

(01h52m--02h22 m UT, max. 01h53 m) at N 14 ~ and E 12 ~ . The X-ray class was estimated

from the observation of the GOES satellite as C3. Flare area was estimated to be 4.0 • l0 Is cm 2 from two H:~ prints supplied from the Peking Observatory during the raster scanning time of 02h03m--06m20 s, while maximum area was 1.3 x 1019 cm 2.

p-wave bursts from 1.0 to 9.40 G H z became a maximum around 01h50m--53 m and

ended at 02u00 m, and/~-wave absorption at 2.0 and 3.75 G H z ( - 4 to - 5 solar flux unit) was observed for about an hour from 02h00 m. The flare occurred in the Mt. Wilson Region 23314, and its eastern adjacent region (23313) showed remarkable movement in one of two spots during 2-3 days (see Figure 5).

3. Result of One-Dimensional Scan at 1650 ~,

The results of one-dimensional raster scanning at 1650 A ( F W H M 84 A and averaged value over 11.5' ) is shown in Figure 3. Here only a portion covering the flare and active

regions is presented. Since for about half of the whole observing time the fine pointing showed large fluctuation of + 10" in the direction of raster scanning ( N - S direction),

we corrected this deviation from exact pointing with a use of telemetry data. The quiet level in Figure 3 and the observed value outside of the figure indicate a value of 47 ergs cm - 2 s - 1 sterad - 1 ~ k -- 1. Earlier measurements by Nishi (1975) and Semain

et al. (1975) show values of 250-300 in the same unit. As will be seen in the next section, our measurement with the MCP also shows a large deficit in the far ultraviolet compared with earlier observations and this deficit does not depend on wavelength. We surmise therefore that the cause should not have resulted from the change of reflectivity, but from the change of the effective width of 45 ~ entrance slit due to an accelerating impact during the rocket firing. However real reason is not clear. Note that corrections due to limb darkening (Nishi, 1973) increase the intensity only by 4% of the above value. In

286 T. H I R A Y A M A E T AL.

-I 7-7- I I II 140

Flare.

~ 100120-

Active Region

80 / "' 1 'cq

60

N 4 0 - S 5 • 10~km ~ N

20

I IE [ 1 I 210 230 250 -- 270

350 330 310 t Is) . . . . 290

Fig. 3. The observed absolute intensity from one-dimensional raster scan at 1650 ,~, (PWHM = 84 ,~,) against position along the rotation axis of the Sun (with slight deviation angle, see text), which corresponds

also to the time of the observation from 02h00 m UT.

unit of brightness temperature our observation gives 4100 K, while canonical values are around 4430 K (see also Vernazza et aL, 1981).

Considering the uncertainty in the absolute intensity of our measurement, we discuss hereafter only relative values to the quiet region, where the radiation temperature is assumed to be 4430 K. From Hv, pictures we estimate the area covered by the active region to be 20Yo of the quiet region at t = 255 s and 305 s in Figure 3. We obtain the average intensity of the active region to be 2.5 times of the quiet region by subtracting the quiet component. Also the area of the flaring region is estimated to be 5 ~ and subtracting the active and quiet components, we find that the intensity is 25.5 times that of the quiet region. The result in the brightness temperature is then

T a = 4700 K (active region)

and

Tf = 5300 K (flare region)

at 1650 ,~ continuum. Contributions from emission lines in our 84 A band are negligibly small compared with e.g. the variations inside active regions: A T ~ 40 K for the flare and less than 20 K for active regions (Canfield et al., 1980; Cheng and Kjeldseth Moe, 1978). The Skylab data show the brightness temperature of 4800-5200 K for the flares of August 9, September 5, and September 7, 1973, and 4700 K for active regions (Cheng

E U V F L A R E I M A G I N G 287

and Kjeldseth Moe, 1978; Cook and Brueckner, 1979). These agreements are rather

encouraging�9

We notice that the intensity is decreasing from t = 234 s to 316 s in the flaring region in Figure 3. The e-folding time is found to be 10-15 min in rough agreement with the decrease of H7 area, whereas Cook and Brueckner (1979) found 2.5 min for the SN flare of August 9 and 6.5 min for the 2B flare of September 7.

4. E U V P i c t u r e o f a F l a r e a n d A c t i v e R e g i o n s

The results of two-dimensional pictures are shown in Figure 4 a - d in terms of raw count numbers with some corrections (see Appendix). These are obtained by averaging photon counts over two rasters (02h04m40s + 100 S) and also a 3 • 3 ( = 15.6" x 16.5") running

mean was taken in order to compensate small photon numbers particularly in quiet

regions. The noise level was estimated to be almost zero from those MCP pixels where no light was incident�9 These iso-count levels are therefore directly proportional to the intensity�9 The representation is cyclic in order of increasing photon count numbers begining from dotted line, then thick line and finally thin line: the count level starts from

1.5, and increases from 2.5 to 15 by an increment of 2.5, and then from 15 till 50 by an increment of 5, and from 50 by an increment of 50.

:.. ". o - "::' ........ ,.'" C?"::,. .. (3 ~

:;"?"" '"' 0 :'"';'''f!" , ~o

I

" ~.,Y-"-] I F] !P .-'=::!. , : : " " :~'' .: ~ ~ ~'~176

o ,--, u ~ ,_~ c~ o "" ".~-,. "'! " ' " ; ' I~&k-~;2,5 A <..s.:.>~ ?-,r

Fig. 4(a). Map ofiso-photon count, proportional to the intensity o fH I Lye 880, averaged over two rasters. Here 780" x 550" is shown. Photon count numbers are indicated (see text). Thick and dashed line, magnetic

neutral line from Figure 6.

T, H I R A Y A M A E T A L .

Fig. 4(b).

288

CIII 977/~, covering the same region as Figure 4(a). See 4(a).

�9 , , � 9

r., , , �9

1/'"s":.:

'~ "; ,': "2,

, - - ":

5 . - . . 72 (" " , , _ . . . . . / . "-.,> ....,,., ,:.

�9 . ' . , t . :

i

L . :

Fig. 4(c). NevIII 770/~, covering the same regaon as Figure 4(a). See 4(a).

E U V F L A R E I M A G I N G 289

::i 7 <

o ,: --.~:: ...-., ff-'-~(_..) _ .-.-----, _

'""'"',-;7." .... ::7-: o

r-",

'::: : }. !:_', 0 o o

Fig. 4(d).

i.-....._...,.

..--"" 0

'_ ',:,,....:; . . . . . . .

(.71": .=(. - ?

�9 . ; " : ' 7 "

.:..

HI Lfl 1025 A, covering the same region as Figure 4(a). See 4(a).

Now we enumerate morphological characteristics of our EUV pictures as compared with He photos in Figure 5. We also compare Figure 4 with the magnetic field map shown in Figure 6 which was taken at the Kitt Peak National Observatory and whose lines of force were calculated with an assumption of potential field (Sakurai, 1982).

(1) While the Ha flare is brighter only in the eastern part of the magnetic neutral line, Lyman continua, C III, Ne viii, and Lfl are brighter in the both sides of the neutral line. This might be related to the fact that the magnetic field strength is stronger in the northern part which has a big spot.

(2) The brightest portion of the flare in CIII is located towards north compared with those of Lyman continua, NevlII, and Lfl.

(3) The spots (three in all) are not significantly brighter than other places in the active region.

(4) Dark filaments appear particularly dark in Lyman continuum at 880 ,~. (5) EUV emission lines are intense in strong field regions, as is expected, but one to

one correspondence is not well seen if structures less than 30" a r e examined. (6) The electron temperature derived from the intensity ratio of Lyman continua

815 ,i, to 880 ,~ is lower in the active region and flare, and slightly higher in dark filaments (Figure 7, see the derivation below). However the shape of the flare in the map of the electron temperature is rather obscure.

290 T. HIRAYAMA ET AL.

Fig. 5. The Ha photo taken at 02h07 m in Peking, on the average 2m20 s later than our observation time, covering the same region as Figure 4(a). The lower end of the photo corresponds to t = 180 s and the upper end to t = 280 s, which can be compared with Figure 3. The inset: flaring portion at 02h00 m (the same

enlargement).

In the following we describe quantitative results. Here we exclude the far left and far right 10~o portions of Figures 4 where the effect due to the tilt of exit slits for emission

lines becomes significant (Figure 1). Table II lists the averaged number of raw photon count rate over the quiet region, the intensity in the mean quiet region, intensity ratio of active region to the quiet, and that of flaring region (most intense portion) to the active region. In converting raw photon count rates to the intensity, collecting area of the telescope, reflectivity of the mirrors and the grating, quantum efficiency of the MCP, and non-uniformity of its efficiency (max. 15~) are all taken into account. The intensities of NaIx 680.7 A and NevlII 769.6 A are tabulated after the subtraction of superposed Lyman continuum intensities which were estimated from 22732.8 and 815.0. The line of SixII 499.5 A (2nd order) is contaminated by the leak from HI L/3 and CIH so that it is omitted from the table together with OvI 1031.7 ~. which is too weak. In the sixth column the ratio of the quiet region intensity from the present observation to that from the Skylab data (Vernazza and Reeves, 1978) is shown. Average ratio of our intensity to the Skylab intensity, excluding Na Ix 680.7 .~, is 0.19, which happens to be the same as the ratio of the 1650 A intensity of the present observation to the averaged quiet value by Nishi (1975) and Semaln et aL (1975). Hence the discrepancy may reasonably be attributed to narrowing of the entrance slit discussed in the previous section.

EUV FLARE IMAGING 291

Fig. 6. Longitudinal magnetic field strength (KPNO) and fines of force, covering the same region as Figure 4. Thick line: magnetic field towards us ( + ). Dotted fine: away from us ( - ). Contour levels: _+ 20, +_ 50, _+ 100, + 200, and _+ 500 G. 2 = 0 is the central meridian at 15 h UT, September 6, 1982 which is 7 ~ E

at the observation time.

Because of this discrepancy in the absolute unit, we only discuss relative values to

the quiet region and the quiet value is taken from other data. First we derive average electron temperature, Te, of the Lyman continuum emitting region by taking the intensity ratio of 22880 and 815 which are free from emission lines (Machado etal., 1980):

1(2) = b 1Bz(Te), where Bz(Te) is the Planck function and bl is the ratio of number density of the hydrogen ground state to LTE value. Also the brightness temperature is defined by 1(2) = B~(Tb). The result is shown in Table III. Values in parentheses are taken from Vernazza et aL (1981, Figure 31).

Since the temperature gradient of the Lyman continuum emitting region may be very steep, the Lyman continua 880 A and 815 i should be emitted from different tempera-

ture regions. We assume, however, that they originate in the same region as we are only roughly comparing among various phenomena such as a flare and dark filaments. We notice first that the electron temperature and b 1 decrease from the quiet region, to the active region and to the flare in this order. Particularly in the case of the flare, it may be that due to enhanced ionization the electron density becomes larger, b 1 nearing to thermal equilibrium (b I ~ 1), and that the electron temperature becomes lower because we can see through into the lower chromosphere by the effect of evaporation (Hirayama,

292 T. HIRAYAMA ET AL,

Fig. 7. Map of electron temperature distribution derived from the intensity ratio of Lyc 880 ~ to 830/~, covering the same region as Figure 4(a). Thick line, magnetic neutral line.

TABLE II

Average observed intensity

Wave- Ion log T e Average count Intensity ~ length (K) (quiet) (erg cm - 2 (A) (ph s - l pixel- 1) s 1 sterad- l)

Present obs./Skylab Active/quiet b Flare/quiet b

680.7 N a i x 5.9 0.3 4.7 c 0.83 ~ 2 ~ 8 ~ 732.8 HI Lyc 3.9 0.5 0.64 0.20 3 10 769.6 N e v m 5.6 0.7 11 ~ 0.20 c 3 ~ 11 ~ 815.0 HI Lye 3.9 2.0 3.4 0.29 2.6 8 880.0 HI Lyc 3.9 3.9 7.3 0.17 3.1 11 976.6 CnI 4.7 3.9 140 0.14 3.5 65

1025.4 HI L/~ 4.3 4.5 180 0.24 2.2 11

" Unit for Lyc is A - 1, b Intensity ratio, differs more than a factor of two from place to place. ~ Lye substracted.

1974). N o n - L T E f a c t o r b 1 ( = 0 .4 ) c a n b e l e s s t h a n un i ty . T h i s o c c u r s w h e n t h e i o n i z a t i o n

is g o v e r n e d b y t h e r a d i a t i o n w h i c h o r i g i n a t e s in t h e a d j a c e n t u p p e r r e g i o n o f h i g h e r

t e m p e r a t u r e , a s is e v i d e n c e d b y a n o n - L T E f l a re m o d e l o f D i n h (1980 , T a b l e 9).

EUV FLARE IMAGING

TABLE III

Temperature and non-LTE factors from Lyman continua

293

Region Quiet Active Flare Dark filament

T~(K) (9300) 7500 7000 ~9500 To(K ) (6600) 6900 7300 ~6400 bj 1300 6 0.4 ~4000

Tabulated values for dark filaments are not accurate due to low count rates, but the large

bl value may reflect the smallness of the electron density.

We can also estimate the geometrical thickness, L, of the flare for Ne viii and C tli

emitting regions, using the emission measure, n2L, directly derived from the intensity. The G O E S soft X-ray flux leads that n2L times area is 1047.3 c m - 3 and from H:~ area

n2L = 1028.7 c m - 5 is obtained. Since at the decay phase of flares the pressure balance

will be held in the flaring transition region, the effective thickness, with the electron

density/'/e as a parameter, can be deduced by assuming a temperature of 107 for the soft

X-ray emitting region and by adopting n~L from the ratio in the column 8 of Table II

and from n2L values in the quiet region given by Dupree (1975): see Table IV. Values

in parentheses are assumed ones. Since the electron density of 10142 (case (c)) for the

hydrogen Lyman continuum is probably too large, and the electron density of 109.5 in

the 107 K region (case (a)) is perhaps too small, the physical condition realized in this

flare is considered to be between case (a) and case (b). In any case the effective thickness

of L = 50-500 m in the C m emitting region is quite small.

TABLE IV

Electron density and geometrical thickness of the flare

log T~ logn2L (a) (b) (c) (K) (cm -5)

logn e L logn e L logne L (cm-3) ( k i n ) (cm-3) (km) (cm-3) (km)

Soft X-ray (7.0) 28.7 (9.5) 5 X 104 (10.0) 5000 (11.0) 50 Ne viii 5.85 27.7 10.7 20 11.2 2 12.2 0.02 CIn 4.95 27.8 11.6 0.5 12.l 0.05 13.1 5 • 10 -4 HI Lyc 3.845 - 12.7 - 13.2 - 14.2 -

5. Concluding Remarks

Conclusions are summarized in the abstract. In this paper we presented maps of EUV

lines and continuum by smoothing over 3 pixels because of the smallness of the photon count. However the core of the flaring region showed 100-500 photon counts ( s - 1) and further detailed study can be done without smoothing. Also a study of quantitative

294 T. HIRAYAMA ET AL.

inter-relations among the intensity, the magnetic field strength, and the electron temperature using, say, scatter plots is left over. They will be discussed in a later work.

Acknowledgements

We acknowledge with sincere gratitude the outstanding cooperation and help to Prof. K. Hirao and Dr I. Nakatani of the Institute of Space and Astronautical Science, the institute responsible for the launch, during the present rocket experiment. We wish to thank Dr J. W. Harvey of Kitt Peak National Observatory for sending us the magneto- gram, and Dr W. Jialong of Peking Observatory for Ha pictures. We are indebted to Messrs S. Koga of the Meisei Denki K.K. (electronics), K. Nakamura of Mitaka Koki K.K. (telescope and spectrograph), and M. Kawamura of Mizoziri Kogaku K.K. (optical elements). Thanks are due to Messrs S. Hamana, M. Nakagiri, and A. Yamaguchi of our observatory for help in the experiment. Mr S. Hamana was responsible for designing the electronics of the fine pointing.

Appendix: Image Corrections for the MCP Data

In drawing the iso-photon count maps of Figure 4, the following corrections were made. Firstly due to mal-alignment, the axis of the MCP was not exactly parallel to the entrance slit-grating ruling direction (a deviation of 3.9 ~ was found). The peak position of, say, the flare in the original map was deviated systematically as a function of wavelengths. This deviation, an almost linear function of wavelength, is determined first by summing up all the photon counts in raster direction (N-S or the ordinate in Figure 4) for each wavelength, and by searching for shifts of summed-up curves in E - W direction among different wavelengths. In summing up the counts in different raster positions we excluded those positions passing flaring regions so that, for example, we can later discuss differences in flare knot positions among different emission lines or continua.

Secondly there appeared a systematic excess of the photon counts in particular pixels (such as in No. 128 and No. 192) and deficiency in the adjacent pixels for all wavelengths and at all raster positions. Being much larger than statistical fluctuations, these deviations are considered to be due to rather systematic error in the analogue-digital (AD) conversion of electric current of the resistive anode, and may certainly be related to the accuracy of the AD converter expressed as + ~ least significant bit (1/512). If we had known, prior to flight, photon count rates for each of pixels of the MCP against spatially uniform light source, the correction should have been straightforward. Since we had not known, we created a smoothly varying light source from the observational data themselves: it was derived again by summing up photon counts for each pixels over whole raster positions and over the 30 s observation at a fixed position (Section 2(d)), and also by summing up counts for all wavelengths without taking into account of the above mentioned 'almost linear shift'. This last point happened to be rather advanta- geous, because structures on the Sun became less conspicuous. And a running mean over 21 pixels (256 pixels in all) was taken for the summed up curve, and the ratio of

EUV FLARE IMAGING 295

the running mean to the summed-up curve was adopted as the correction factor for each pixels. This correction factor is to be multiplied to raw photon counts in each pixels. Note that when a running mean over, say, 17 was taken, small peaks around Nos. 128 and 192 pixels which showed largest deviations of 400% did not yet disappear (the average deviation ~ 20%). On the other hand when a running mean over, say, 33 was taken, it was observed that the portion covering the active region including the flare was underestimated and outside region was overestimated by a factor of 17% or so. Hence the figure of 21 is rather meaningfull.

The other factors which affect the photon counts are vignetting due to the concave grating (a ruled area of 50 x 30 mm2), and non-uniform sensitivity of the MCP measured at our observatory. Both are easy to correct, but they are at most 10%, respectively, and significant only at the left and fight ends of Figure 5, where counts are less than three or so, and hence no corrections were performed.

In practice we first derived the correction factors for the error from the AD converter and then performed the wavelength shift. In spite of rather large corrections particularly to the AD errors, we believe that corrections are fairly accurate, probably within 5 ~o, and do not invalidate discussions given in the text.

References

Canfield, R. C. et al.: 1980, in P. A. Sturrock (ed.), Solar Flares, A Monograph from Skylab Solar Workshop H, Colorado Associated Univ. Press, Boulder, p. 451.

Cheng, C. C. and Kjeldseth Moe, O.: 1978, Solar Phys. 59, 361. Cook, J. W. and Brueekner, G. E.: 1979, Astrophys. J. 227, 645. Hirayama, T.: 1974, Solar Phys. 34, 323. Hirayama, T.: 1978, Publ. Astron. Soc. Japan 30, 337. Dinh, V. Q.: 1980, Publ. Astron. Soc. Japan 32, 495. Dupree, A. K.: 1975, Astrophys. J. 178, 527. Moore, R. J. et al. : 1980, in P. A. Sturrock (ed.), Solar Flares, A Monograph from Skylab Solar Workshop H,

Colorado Associated Univ. Press, Boulder, p. 341. Machado, M. E., Avrett, E. H., Vernazza, J. E., and Noyes, R. W.: 1980, Astrophys. J. 242, 336. Nakatani, I., Sato, T., Tange, T., Hamana, S., Hirayama, T., Nishi, K., and Akita, K.: 1983, Inst. Space

Astronautical Sci. Rep. No. 608. Nishi, K.: 1973, Solar Phys. 33, 23. Nishi, K.: 1975, Solar Phys. 42, 37. Sakurai, T.: 1982, Solar Phys. 76, 301. Semain, D., Bonnet, R. M., Gayet, R., and Lizamber, C.: 1975, Astron. Astrophys. 39, 71. Vernazza, J. E. and Reeves, E. M.: 1978, Astrophys. J. Suppl. 37, 485. Vernazza, J. E., Avrett, E. H., and Loeser, R.: 1976, Astrophys. J. SuppL 30, 1. Vernazza, J. E., Avrett, E. H., and Loeser, R.: 1981, Astrophys. J. Suppl. 45, 635.