' a, . / S" : - <. jy- ESSA TR ERL 169-SDL 14 A UNITED STATES DEPARTMENT OF COMMERCE PUBLICATION ESSA Technical Report ERL 169-SDL 14 U.S. DEPARTMENT OF COMMERCE Environmental Science Services Administration Research Laboratories Extreme Ultraviolet Flashes of Solar Flares Observed Via Sudden Frequency Deviations RICHARD F. DONNELLY BOULDER, COLO. SEPTEMBER 1970 ESSA RESEARCH LABORATORIES The mission of the Research Laboratories is to study the oceans, inland waters, the lower and upper atmosphere, the space environment, and the earth, in search of the under- standing needed to provide more useful services in improving man's prospects for survival as influenced by the physical environment. Laboratories contributing to these studies are: Earth Sciences Laboratories: Geomagnetism, seismology, geodesy, and related earth sciences; earthquake processes, internal structure and accurate figure of the Earth, and distribution of the Earth's mass. Atlantic Oceanographic and Meteorological Laboratories: Oceanography, with emphasis on the geology and geophysics of ocean basins, oceanic processes, sea-air interactions, hurricane research, and weather modification (Miami, Florida). Pacific Oceanographic Laboratories: Oceanography; geology and geophysics of the Pacific Basin and margins; oceanic processes and dynamics; tsunami generation, propaga- tion, modification, detection, and monitoring (Seattle, Washington). Atmospheric Physics and Chemistry Laboratory: Cloud physics and precipitation; chem- ical composition and nucleating substances in the lower atmosphere; and laboratory and field experiments toward developing feasible methods of weather modification. Air Resources Laboratories: Diffusion, transport, and dissipation of atmospheric con- taminants; development of methods for prediction and control of atmospheric pollution (Silver Spring, Maryland). Geophysical Fluid Dynamics Laboratory: Dynamics and physics of geophysical fluid systems; development of a theoretical basis, through mathematical modeling and computer simulation, for the behavior and properties of the atmosphere and the oceans (Princeton, New Jersey). National Severe Storms Laboratory: Tornadoes, squall lines, thunderstorms, and other severe local convective phenomena toward achieving improved methods of forecasting, detecting, and providing advance warnings (Norman, Oklahoma). Space Disturbances Laboratory: Nature, behavior, and mechanisms of space disturb- ances; development and use of techniques for continuous monitoring and early detection and reporting of important disturbances. Aeronomy Laboratory: Theoretical, laboratory, rocket, and satellite studies of the physical and chemical processes controlling the ionosphere and exosphere of the earth and other planets. Wave Propagation Laboratory: Development of new methods for remote sensing of the geophysical environment; special emphasis on propagation of sound waves, and electro- magnetic waves at millimeter, infrared, and optical frequencies. Institute for Telecommunication Sciences: Central federal agency for research and services in propagation of radio waves, radio properties of the earth and its atmosphere, nature of radio noise and interference, information transmission and antennas, and meth- ods for the more effective use of the radio spectrum for telecommunications. Research Flight Facility: Outfits and operates aircraft specially instrumented for re- search; and meets needs of ESSA and other groups for environmental measurements for aircraft (Miami, Florida). ENVIRONMENTAL SCIENCE SERVICES ADMINISTRATION BOULDER, COLORADO 80302 .^eN-i-p^ Sc 'Hct Mttf& U.S. DEPARTMENT OF COMMERCE Maurice H. Stans, Secretary ENVIRONMENTAL SCIENCE SERVICES ADMINISTRATION Robert M. White, Administrator RESEARCH LABORATORIES Wilmot N. Hess, Director ESSA TECHNICAL REPORT ERL 169-SDL 14 Extreme Ultraviolet Flashes of Solar Flares Observed Via Sudden Frequency Deviations RICHARD F. DONNELLY n. o 10 keV) 25 5.2 Soft X-Ray Bursts ( X > 1A) „ 26 5.3 Ultraviolet Flashes (1030-3000A) 29 5.4 White Light Flares 29 5.5 Low-Chromosphere Optical Line Emission 30 5.6 Ha Flares 32 5.6.1 Area Dependence 32 5.6.2 Intensity Dependence 32 5.6.3 Sunspot Dependence 33 5.6.4 Association of EUV Bursts with the Ha Explosive Phase 33 5.6.5 Dependence on Location of the Ha Flare 35 5. 6. 5. a Dependence of Occurrence on Central Meridian Distance 35 5.6.5.b Peculiar Long-Duration Events 38 5.6.5.C Limb Flares 4 8 5.6.6 Bright Impulsive Ha Kernels 48 5.7 Relation to Surges, Sprays and Eruptive Prominences 50 5.8 Microwave Bursts 51 6. SFD's FOR FAMOUS FLARES 56 7. CONCLUSIONS, RECOMMENDATIONS, AND SPECULATIONS 56 8. ACKNOWLEDGEMENTS 67 9. REFERENCES 68 APPENDIX A 72 o 1. Procedures for Analyzing SFD Data to Deduce the 10-103QA Flux Enhancement: Brief Descriptions 72 2. Sensitivity of Frequency Deviations to Solar Bursts 76 3. Comments on SFD Observations 81 4. Miscellaneous Unpublished Studies of SFD's 81 APPENDIX B 85 LIST OP FIGURES Page 1. SFD sensitivity to ionizing radiations as a function of radiation wavelength. 2 2. The dependence of SFD's on the impulsiveness of the flare enhancements of ionizing radiation, t = the effective electron-loss rate in the ionosphere . 4 3. Monthly rate of occurrence of SFD's at Boulder from October, 1960 through June, 1970. 10 4. Diurnal dependence of Boulder SFD's. 10 5. Seasonal dependence of Boulder SFD's. 12 6. Intensity distribution of all Boulder SFD's. 13 7. Intensity distribution of selected Boulder SFD's. 14 8. Rise-time distribution of Boulder SFD's. 17 9. Duration distribution of Boulder SFD's. 19 10. An EUV flare with quasi-periodic fine structure . 22 11. Spectrum of fine structure . 23 12. Distribution of fine structure periods. 23 13. Comparison of timing of hard X-ray burst and SFD of 2141 UT March 1, 1969, OSO 5 X-ray observations made by Frost (1969). 27 14. Illustration of components of flare radiation. The impulsive component A occurs concurrently at centimeter wavelengths , hard X-rays, certain EUV wavelengths , 'certain spatial portions of Ha. flares, and occasionally in white light. The gradual or slow component, B occurs with slightly differ- ent time dependence in soft X-rays, certain other RUV wavelengths , most of the Ha flare, and centimeter wavelengths where it is sometimes called a postburst increase . The Type IV Component, C, occurs mainly at radio wave- 28 lengths . o 15. EUV burst, 5324A Fe emission and Ha flare of 2010 UT April 21, 1969. 31 16. The dependence of the association of SFD's with explosive-phase Ha flash on the intensity of the SFD. The number at the bottom of each histogram bar denotes the number of events involved. 34 17. The association of SFD's and Ha explosive-phase flares as a function of Ha flare importance and brightness . The number at the bottom of the histogram bar& denotes the number of events involved. 36 18. Association of an EUV burst with a small bright Ha kernel. 37 19. Dependence of EUV flashes on the central meridian distance of the associated Ha flare for 1964-1968 for all Ha flare classifications . 39 20. Dependence of EUV flashes on the central meridian distance of the associated Ha flare for 1966 for all Ha flare classifications . 40 21. Dependence of EUV flashes on the central meridian distance of the associated Ha flare for 1967 for all Ha flare classifications . 41 22. Dependence of EUV flashes on the central meridian distance of the associated Ha flare for 1968 for all Ha flare classifications . 42 23. Dependence of EUV flashes on the central meridian distance of the associated Ha flare for 1964-1968 for all Ha flare classifications of importance >_ 1. 43 24. An example of an SFD with an unusually long duration of positive frequency deviation. The station-identification break is shown by SB. 46 o 25. Peak 606 MHz emission versus peak 10-1030A flux enhancement. EL denotes East Limb flare location, • Central Meridian Distance (CMD) >_ 70°, 9 20° < CMD < 70°, and 9 CMD <_ 20°. 53 o 26. Peak 1415 MHz emission versus peak 10-1030A flux enhancement. 53 Page 27. The average of peak radio emissions observed at 2695, 2700, and 2800 MHz versus peak 10-1030A flux enhancement. 54 o 28. Peak 4995 MHz emission versus peak 10-1030 A flux enhancement. 54 o 29. Peak 8800 MHz emission versus peak 10-1030A flux enhancement. 55 o 30. Peak 10700 MHz emission versus peak 10-1030A flux enhancement. 55 31. Boulder SFD observations for the flare of 1924 UT May 21, 1967. 58 o 32. Normalized 10-10 30 A flux enhancement as a function of time for the flare of 1924 UT May 21, 19 67. 59 o 33. SFD and 10-1030 A flux enhancement as a function of time for the flare of 1710 UT July 8, 1968. ' 60 34. Boulder SFD observations for Q the flare of 1740 UT March 12, 1969 (a) Oblique- Path Observations A$C 10-10 30A, t) for this event is presented in Mcintosh and Donnelly (1970). (b) Near Vertical-Incidence Observations . 61 35. Boulder SFD observations for the flare of 2009 UT April 21, 1969 h§(10-1030A, t) for this event is discussed in section 5.6.6. 63 36. Models for the impulsive EUV flash. 66 37. Examples of electron density and loss time constant as a function of height. 74 38. An example of the wavelength dependence of SFD s ensitivity to ionizing radiations for f = 2 MHz ■*■ f = 2 MHz, f = 5 MHz ■+ f = 5 MHz. 80 J J y '( ' J V V 39. Examples of consistency in midlatitude SFD observations . 82 LIST OF TABLES Page 1. Radiation detector characteristics of sudden frequency deviations. 6 2. Rf, $f and R$ as a function of wavelength. 7 3. Transmissions used in Boulder SFD observations. 9 o 4. Estimates of peak 10-1030A flux enhancements. 16 5. SFD's with complex time structure. 21 6. Some statistical properties of SFD events. 24 7. Contingency table for SFD's and Ha flares with an explosive-phase. 34 8. Long duration events. 44 9. Association of SFD's with surges and sprays. 50 10. Famous events. 57 11. Boulder SFD events 1968 through June 19 70. 85 vn ABSTRACT The properties of EUV flashes from solar flares measured via a type of ionospheric event, called a sudden frequency deviation (SFD) , are presented. These results are based on ten years of observations at Boulder wherein about 2000 SFD events were detected and scaled. The characteristics of SFD's as a detector of solar radiation are discus- sed. SFD's are sensitive to bursts of solar radiation at wavelengths in the o 1-10 30A range. Unlike most detectors of solar radiations, SFD's are sensitive to the impulsiveness of the solar radiation and are insensitive to the non-flare o radiation. The sensitivity of SFD's to 1-1030A bursts depends upon the specturm o of the 1-1030A radiation, the solar zenith angle, time of day and-season, the pre- flare electron density as a function of height, the upper atmospheric constituent densities, and other ionospheric disturbances. The main assets of SFD observation are 1 sec. time resolution, low cost, ground-based equipment, and essentially con- tinuous daylight coverage. The wavelength spectrum of the radiation responsible for SFD's is revised. The characteristics of He II 303. 8A, OV 629. 7A, HLyy 972. 5A, CIII 977. 0A and o o HLya 1215. 7A have essentially the same time dependence as the 1-1030A flash res- o ponsible for SFD's. Soft X-rays (2-20A) and certain EUV lines that are normally o coronal lines have a much slower time dependence than the 1-10 30A flash and contribute little to SFD's. The rise time, duration, intensity, and fine time-structure of SFD's are statistically studied. Although most SFD's have some fine structure, quasi- periodicity in EUV flashes is quite rare. When it does occur, it usually is o weak relative to the peak 10-10 30A flux enhancement. Previously unpublished SFD data for famous flares are given in detail. A revised list of Boulder SFD observations is given for 19 6 8 through June, 19 70, when routine Boulder SFD observations were terminated. Present knowledge of the close association of EUV flashes with hard X-ray bursts, white-light emission and microwave radio bursts are reviewed. The re- lationship of EUV flashes to small bright impulsive kernels in Ha and other op- tical observations of solar flares are presented and the location of these ker- nels discussed. The intensity of EUV flashes is shown to depend upon the central meridian distance of the flare location; the intensity decreases at the limb. Several models for the impulsive EUV source region are proposed. Key Words: sudden frequency deviations, extreme ultraviolet, X-ray, solar flare, white-light flare, ionosphere. vm EXTREME ULTRAVIOLET FLASHES OF SOLAR FLARES OBSERVED VIA SUDDEN FREQUENCY DEVIATIONS Richard F. Donnelly 1. INTRODUCTION A sudden frequency deviation (SFD) is a type of sudden ionospheric disturbance (SID) caused by bursts of X-rays and extreme ultraviolet (EUV) radiation from solar flares. In effect, SFD observations are detectors of impulsive solar ionizing- o radiation in the 1-10 30A wavelength range. This report presents our knowledge of EUV flashes of solar flares based on about ten years of SFD observations at Boulder, Colorado. An SFD is an event in which the received frequency of a high-frequency radio- wave reflected, usually from the F-region of the ionosphere, increases suddenly, peaks, and then decays to the transmitted frequency. The frequency deviation some- times has several peaks and usually takes on negative values during the decaying portion. The start- to-maximum time is typically about one minute, and the peak frequency deviation is usually less than 1/2 Hz. Those SFD's observed on paths reflected from the bottom of the E layer or from sporadic E are sensitive only to o 1-10A flare radiation and are generally smoother, slower, and much smaller than those observed on paths reflected from the F-region. In this report, the term "SFD" will refer to events observed on paths reflected from the F-region or upper-E- region, unless noted otherwise. These SFD's differ from most types of SID effects in that the frequency deviation is proportional to the time rate of change of electron density primarily in the E and Fl regions produced by flare radiation in the o 10-1030A range, rather than being proportional to the D-region electron-density o enhancement produced by 1-10A flare radiation. In the next section, we first treat SFD observations as detectors of solar-flare ionizing radiation and present the characteristics of this detector, in a way analogous to reports on satellite X-ray measurements. Wavelength dependence of the radiation responsible for SFD's and the statistical characteristics of SFD's are then reviewed, and the characteristics of EUV flashes of solar flares based on SFD observations and their relationships to other solar flare radiations are dis- cussed. 2. RADIATION DETECTOR CHARACTERISTICS Figure 1 shows a typical curve of the sensitivity S(X) of SFD's to radiation incident from above the earth's atmosphere as a function of the wavelength (A) of the ionizing radiation. S(A) is defined by Af(f v ) (1) S{X) R~R A4>(A)dA t x where Af is the frequency deviation in Hz, A4> is the radiation enhancement in ergs -2 -1 °-l cm sec A , R is a dimensionless factor that accounts for the time dependence of A*, R accounts for the solar zenith angle dependence and f is the equivalent vertical-incidence frequency of the SFD probing radio wave. For fast EUV bursts and an overhead sun, R = R = 1; S(A) is derived in Appendix A. him — i — i — i r miii — i — r ii i i — i — r ^ -.nil i c o o s o 4i O s s a a CO £ o +i « . a en fc g S 0) ^ a K O R. •?» O O HJ> -P <3 +i CO r CO I I I I I I jnV* *D ^_D9S 2 _UUD S6J8) ZH (\) S The value of S(X) drops off rapidly with decreasing wavelength below about 1A because radiation at those wavelengths produces ionization mainly low in the dense D-region and negligible ionization in the E and F regions. In the D-region, most of the freed electrons are rapidly lost via attachment to form negative ions. The electrons that remain free have such a high collision frequency with the neutral gas that they do not interact effectively with the probing radio wave. The sensitivity drops o off abruptly with increasing wavelength at about 10 30A because radiation at wave- o lengths higher than 1027A is very ineffective at ionizing any of the major con- stituents of the upper atmosphere. There is a similar dropoff near 800A because o molecular nitrogen ceases to be ionized, and another dropoff above 900A because atomic oxygen ceases to be ionized. The remaining variation in sensitivity o throughout the 1-10 30 A range is a consequence of the variation with wavelength of the absorption cross sections of the major constituents of the upper atmosphere. The sensitivity of frequency deviations to solar flare radiation varies with solar zenith angle, frequency of the probing radiowave, season, and numerous other factors. These effects are discussed in detail in Appendix A. Despite these complex- ities, SFD observations are much like a very broad-band satellite detector. The solar zenith angle dependence of SFD's is analogous to the aspect angle dependence of satellite measurements. Just as satellite X-ray observations are complicated by particle radiation, SFD's are complicated by other types of ionospheric disturbances. Because of the "noise" from small ionospheric fluctuations that are always present, o the smallest 1-10 30 A flux enhancement detectable by midlatitude SFD observations . . . -3 -2 -1 under the best conditions is about 4 x 10 ergs cm sec , and, more typically, -2-2-1 . ° 10 ergs cm sec . The frequency deviation is linearly related to the 1-1030 A -2 -1 flux enhancement for enhancements less than 1 ergs cm sec , above which a small nonlinearity may occur after the EUV flash has been in progress for about 100 sec. o The main difference between SFD detection of 1-10 30 A radiation and satellite broadband radiation detectors is that SFD's are sensitive to the impulsiveness of the o 1-1030 A radiation. Figure 2 illustrates this feature. The solid curve represents the time dependence of a burst of ionizing radiation. The dashed curve illustrates the time dependence of the time-rate-of-change of electron density and, hence, the SFD, for the case where the rise time (t ) and decay time (t,) of the burst are much less than the effective electron- loss time constant in the E and Fl regions of the iono- sphere. In this case, the SFD time dependence is essentially the same as that of the incident radiation enhancement, except late in the decay stage. The dash-dot curve illustrates the case when the rise and decay times of the burst of radiation are comparable to the effective electron- loss time constant. The SFD in this case has a time dependence similar to the radiation burst but the distortion is quite appar- ent. The SFD data can be used to estimate the radiation burst by computing the effects of the electron loss processes. The dotted curve illustrates the case when the burst time constants are much larger than the ionospheric electron- loss time constants. The SFD is relatively small and highly distorted with respect to the time dependence of the radiation burst. Considering the "noise" in SFD data caused by non flare- related ionospheric fluctuations and our lack of precise knowledge of the electron loss rates as a functions of height at the time of a particular SFD, it is usually impractical to reconstruct the radiation enhancement from SFD measurements in this o CD - .£ O 00 O O O CVJ d o C\J O" 1 0) « I ^ S ^O Ca +i ^ O •+i Ca <*-» (» o CO » s o ^^ to Cfl s ca £ ■» II ca s •> o s o CO "^ - +s Cl « Pt, '^ « c» ^^ o ca cj K ca ca Cj5rS! ^ CO o ca *^ ft, ca ca ^Krs; ca co Eh £ ca ca • ca « c\i o ^ K ca « co ?H ^ CO 3 K o cjj ca ^ XDUU 'XDW b y IP NVp case. The electron-loss time constant in the E and Fl region is typically between 15 and 60 sec; hence, SFD observations are relatively insensitive to radiation bursts with smooth rises of five minutes or more. Obviously, one major limitation of SFD o observations is that they give no information on the non-flare 1-10 30 A radiation. The above discussion qualitatively explains the "impulsiveness" dependence of SFD's; Appendix A gives more quantitative information of some of the complexities involved. Table 1 summarizes some of the basic characteristics of SFD's as a detector of o 1-10 30 A flux enhancements. o 3. WAVELENGTH DEPENDENCE OF SOLAR FLARE RADIATION WITHIN THE 1-10 30 A RANGE o Information on the wavelength dependence of the 1-10 30 A flare radiation comes mainly from many satellite experiments, particularly the EUV observations of Hall and Hinteregger (1969) . Donnelly (1968a, b) has examined the relationship between these various satellite observations and SFD's. The results are summarized in table 2 and include some corrections and additions. Most of the corrections involve an improved estimate of the preflare EUV flux based on OSO 3 measurements (Hall, 1969) of abso- lute EUV flux near the time of the EUV flares upon which the results in table 2 are based, and on improved information of the variation of some EUV lines with respect to the 10.7 cm flux based on Hall et al. (1969) and Hinteregger and Hall (1969). Con- sidering the scatter in the ratio of their EUV flux measurements to the radio flux, the preflare EUV flux estimates may easily be in error by a factor of two. In table 2, the symbol Rf indicates how much of an observed SFD is produced by the flare radiation at a particular emission line or wavelength range. The quantity $f is useful for estimating the flare radiation enhancement at particular wave- o lengths from SFD observations. The quantity $f for the entire 1-10 30A range is approx- -2 imately 0.0 8 ergs cm . The intensity of the flare radiation enhancement at a part- o icular wavelength relative to the total enhancement in the 1-10 30 A range is indi- cated by R$. Most of the entries in table 2 are based on only a few events and should therefore be considered as preliminary rough estimates. More narrowband satellite measurements in the 20-1030 A range are needed for more flares and wave- lengths to complete our knowledge of the spectrum of the ionizing radiation that produces SFD's. One interpretation of the results in table 2 is that the EUV recombination continuum and line emission from the more abundant solar constituents H, He, 0', C, and N, are the main cause of SFD's observed on paths reflected from the F region. However, another interpretation includes a bremsstrahlung continuum throughout the o 1-1030 A in addition to the above radiations. McClinton (1968) and Grebenkemper 's O o (1969) 1085-1350 A and 1225-1350 A measurements for the proton flare of August 28, 1966, are consistent with a continuum enhancement and not explainable by an enhance- ment of Lyman a. They suggest that the EUV and white light flash are bremsstrahlung o emission from an optically thick source region. Some of the 1-1030 A burst is un- doubtedly bremsstrahlung emission; but how much is not known. The sum of Rf in table 2 accounts for only half the SFD; part of the remaining half could be some type of continuum emission. Part of the continuum emission Hall and Hinteregger (1969) observed in the hydrogen recombination continuum wavelength range could have been a broader-band continuum emission. No such continuum is known to have been observed. Table 1 Radiation Detector Characteristics of Sudden Frequency Deviations Wavelength Range ~ 1-1030 X. Dynamic Range: -2 -2 -1 Minimum detectable flux enhancement « 10 ergs cm sec -2 -1 Maximum measureable flux enhancement w 10 ergs cm sec The minimum level is a consequence of "noise" in the frequency deviations produced by ionospheric variations that are usually present and unrelated to flares. The maximum level is a consequence of ionospheric absorption of the SFD probing radio waves (induced by the 1-10 X flare radiation) and the lowering of their height of reflection to the bottom of the E layer because of the flare-induced electron density enhancement in the E and Fl regions. Solar Zenith Angle (x) Dependence of Sensitivity: c w cos x Rise Time (tj.) Dependence of Sensitivity: R t « 1 for t r & 10 sec; for t r > 10 sec, R t « c (l-e"^ ' c') for a linear rise, where t is the effective electron-loss time constant in the ionosphere; for more general relations, see Donnelly (1969a). Time Resolution: m 1 sec for Boulder observations with Af structure 2 IHz processed with 1 minute 2 l.k inches of chart record. Higher resolution could be achieved if needed. A resolution of 1 sec means that two impulses of 1-1030 A radiation separated by a 1 sec could be distinguished as more than one pulse. Fine structure with rise times t r 2 1 sec are measureable. Structure with tjj. < 1 sec would still be evident in SFD data, although their rise times would be immeasurable. No SFD has been observed at Boulder that contained significant unresolved structure. Relative Intensity Resolution w 10 ergs cm" sec" or 1% of maximum A*(l-1030 A), whichever is larger, for resolution of intentity of fine structure having t r £ 10 sec. Wavelength Resolution: At best, one can estimate the radiation in the following wavelength groups: 1-10 A, 10-100 A + 910-1030 A, 100-400 X + 800-910 X, and 1+00-800 X. Spatial Resolution: None. Absolute Intensity Accuracy: typically ~ a factor of k; i.e. the real flux should be within the range l/k x's to h x's the given value. This poor accuracy is a consequence of our inadequate knowledge of the spectrum of the 1-1030 A flare radiation, of the ionospheric electron loss rates along the paths of the SFD probing radiowaves, etc. Absolute Timing Accuracy » ± 2 sec, for Boulder observations with Af structure 2 1Hz processed with 1 minute 2 1.4 inches of chart record. Time Coverage « h5%. SFD's are observed only during the daytime. Ionospheric storms and other ionospheric disturbances undoubtedly prohibit the observation of some small SFD's. (see below). Data Storage: Two weeks of three channels of SFD data and one time code channel recorded on one 1800' reel of l/V magnetic tape. Data Processing: Standard audio spectrum analyzers used to obtain records of frequency deviation versus time. Interfering Ionospheric Phenomena for Midlatitude SFD Observations: Sporadic E - Causes anomalous propagation of the probing radio waves making the SFD observations temporailly insensitive to EUV flashes. •Traveling Ionospheric Disturbances - rarely occur at the time of an SFD. Common Accoustic-Sravity Waves - always present, the main source of interference with small SFD's, minimized by using one-hop oblique propagation paths of 500-1500 km ground length. Magnetic and Ionospheric Storms - For a day or more after some large solar flares, the ionosphere is disturbed with small fluctuations because of the ionospheric effects of particle radiation from the flare, or the SFD probing radiowaves may be reflected at an anomalously low height. Short Wave Fadeouts - During some flares the 0.5-8 A x-ray enhancement is so large and consequently the D-region ionization and radiowave absorption are enhanced so much, that the SFD probing radiowave is lost in the radio noise and local interference. Advantages of SFD's as Detectors of 1-1030 8 Impulsive Flare Radiation: Inexpensive, high time resolution and accuracy, good rela- tive intensity resolution, good time coverage, ground-based, easy maintenance, broadband, and convenient high-density data storage . Disadvantages of SFD's as Radiation Detectors: No spatial resolution, poor wavelength resolution, poor absolute intensity accuracy, high value of minimum detectable flux enhancement, insensitive to non- impulsive radiation, tedious process for converting Af to A* , and influenced by unknown time variations of the upper atmosphere and ionosphere . Rf, (l-1030A,t ) *f(l-1030A) where: Af is the frequency deviation (Hz) computed from the observed radiation enhancement A 4> , Af is the c o o observed frequency deviation, f is the equivalent vertical incidence frequency of the SFD probing frequency, t is time and t is the time of the peak of Af , R and R are correction factors to account for the rise p o t X time of the SFD relative to the electron loss time constant and for the solar zenith angle dependence, respec- tively, A, - X, is the wavelength range of A

10 keV) 0.5-3 1-8 8-20 303 .8 335. 3 368.1 465.2 554 584.3 529 .7 760 765.1 790 834 680-911 949.7 972 .5 977.0 990 1025 . 7 1031 .9 1085 1175 1206.5 1215 .7 1080-1225 1238.8 1225-1350 6563 3500-6500 3-10 cm He II Fe XVI Mg IX Ne VII Group IV He I o V Group V N IV Group O IV Group o Usui H Cont H I Ly<5 H I Lyy C III Group N III H I LyS O VI Group N II Group C III Si III HI Lyct NV Ha White Light radio .0£ 2.4 <2.1 5.9 <0 .6 . 4 . 3 4.0 .6 1.2 0.6 0.8 2. 3 2 . 2 8.1 0.2 6.2 1.4 2.3 .1 <0.1 8 7 6 <4 0x10 OxlO - ^ 3x10 j 6xl0~^ -.0 7 <5 001 09 8 8 5 2x10 6 6 2 .8x10 1 .6x10 -3 9 3x10 -4 1 7x10 ■A J 3x10 -3 fa 4x10 - 3 2 0x10 -3 1 3x10 -3 3 2x10 -3 1 1x10 • 7 1 7x10 -? 1 4x10 -? 8 xlO -7 1 4x10 •■1 6 1x10 -3 7 3x10 2.4x10" 0.5 0.4 . 2 3.4 0.5 1 .0 0.5 .6 2 .0 2 .7 12 . .2 .4 8.0 2.5 1.6 4.1 1.4 71. 17. -100 . 17. .8 9 .1 <4.3 -100. 3x10" It It, Is Is Is Is 4t ,1s Is Is Is Is Is Is It, Is 2t,ls Is Is Is Is Is Is 7 ,7t ,1s It Is It 3t 4t 43t A*(1-8A) decays slower than A*Jl-1030A,t) . 'A$(8-70A) rises, peaks, and decays slower than A* ( 1-10 30A , t) . Good time dependence agreement with A*(l-1030A,t) . l,2^|Appears to be slower than 1,2 | A*(1-1030A) . 1,7 1,7 1,7,-jGood time dependence agreement with 1,7 J A*(l-1030A,t) . 5ood time dependence agreement with A*(l-1030A,t) . 3ood time dependence agreement with A*(l-1030A,t) . A \'M 3ood time dependence agreement with A* (1-1030A, t) . ood time dependence agreement with A*(l-1030A,t) . Good time dependence agreement with A«(1-1030A, t) . pood time dependence agreement with A*(1-1030A) to within the time resolution of white light photographs . All results based on only one or two events should be considered as unverified rough estimates only. Rf and *f are based on SFD observations at f 5MHz and should not be applied when f * 5MHz. If Rf were v s r v based on f = 7MHz, then Rf = 100% for 1-10A and = for X > 10A. The reference t = 5MHz was chosen be- v v cause it is about the lowest value of f for Boulder SFD observations for which Af is not highly sensitive v o to the preflare electron concentration height-profile. *t, with high time resolution; s, low time resolution. **In view of results in ref . (4) , there "Lya" observations may have been influenced by a continuum enhancement. References: 1. Donnelly (1969a, b) 7. A« Q (X i ), Hall and Hinteregger (1969); 3. Kane and Donnelly (1970); *' A *o (X l " *2 1, McClinton (1968) and Grebenkemper (1969); 5. A« (Ha), Thomas (1970); 6. Mcintosh and Donnelly (1970). The nonflare EUV "continuum" (other than recombination continua) observed by satellites is not a continuum in the true sense in that it may consist of small unresolved lines plus light scattered in the spectrometer from strong lines at other wavelengths, as well as any true continuum emission. Therefore, even if a small "continuum" enhance- ment were observed, it would be difficult to determine whether it were a true contin- uum. The results in table 2 for wavelengths outside of the 1-1030 A range will be o discussed in Section 5. Since almost all of an SFD is produced by the 10-10 30 A flux o o enhancement and very little by 1-10 A radiation, and since 1-10 A radiation is gener- ally much slower than the radiations mainly responsible for SFD's, we will frequently refer to SFD's in the rest of this report as being caused by, or providing information o about, the impulsive 10-10 30 A flux enhancement of flares. 4. STATISTICAL CHARACTERISTICS OF SUDDEN FREQUENCY DEVIATIONS AND EUV FLASHES SFD observations were made in Boulder under the direction of Dr. K. Davies for about 10 years, which amounts to nearly 4 x 10 hours of daylight observation. About 3 2 x 10 SFD's were detected, scaled and tabulated. Preliminary lists of Boulder SFD's have been published in NOAA's Solar Geophysical Data since 1963. Revised lists have been published by Agy et al. (19 65) for October 19 60 through December 19 62, by Baker et al. (1968) for 1963 through 1967, and in Appendix B of this report for 1968 through June, 19 70, when Boulder SFD observations were terminated. Tabulated informa- tion for all SFD's observed in Boulder is available either in computer listings or punched card form from World Data Center A, Upper Atmosphere Geophysics, NOAA, Boulder, Colorado, 80302. Table 3 documents the main transmission frequencies and pro- pagation paths used in the Boulder SFD observations. Previous studies of the characteristics of SFD's were made by Chan and Villard (1963) , based on about one year of SFD observations, and by Agy et al . (1965) , based on about two years of observations. The results of these studies, include the following: (1) Diurnal Dependence : The number of SFD's observed and the percentage of Ha flares accompanied by SFD's varied with the hour of day because of the solar zenith angle dependence of SFD's. (2) Seasonal Dependence : The percentage of flares during SFD observing periods that were accompanied by SFD's varied from month to month but exhibited no distinct seasonal dependence, (a seasonal dependence is presented in the next section) . (3) Rise Time : The start- to-maximum times of SFD's were most commonly one to two min. and the average value increases slightly with increasing importance of the associated Ha flare. (4) Duration : The start-to-end times for SFD's ranged from 1 to 15 min. with the most common duration being from 2 to 5 min. The SFD duration tends to increase with the importance classification of the Ha flare. 4.1 Time of Occurrence Figure 3 shows the monthly rate of occurrence of SFD's observed at Boulder for the past ten years. The solar minimum in 1964 is quite evident, but probably the most striking features of this figure are the large spikes caused by certain active Table 3 Transmissions Used in Boulder SFD Observations Transmitter Frequency Dates of SFD Observations Call Letters Locati on D (km) MHz From To WWV WWV KKE42 KKE42 KKE42 KKE42 KKE42 KC2xBI WW I WWI WWI WWI WWI WWI Greenbelt, Md Sunset , Colo . Akron , Colo . Havana , 111 . 2420 25 20 15 10 09/01/60 09/08/61 12/21/60 01/17/62 12/15/60 01/11/61 170 5 1 1290 9 .9 it 9 9 ii 8 9 ti 11 1 ii 12 .1 n 13 5 02/09/63 4 07/30/61 5.054 08/04/61 5.1 02/17/67 3.3 06/04/65 2 09/27/67 11/24/65 :S 04/13/66 01/04/67 08/0 3/66 08/08/66 09/19/66 09/19/66 06/02/61 01/17/62 01/10/61 09/19/66 12/20/60 09/19/66 07/15/63 05/23/69 02/17/67 05/23/69 04/15/69 05/23/69 01/05/66 01/04/67 01/08/69 06/08/70 05/12/70 01/08/69 07/01/70 D = Approximate ground range from the transmitter to receivers located near Boulder, Colo, 70 Q CO 50 40 E 30 1884 Events Average m i6SFDs/Month I EJa nrfU y m I I960 I IE 1962 I 1963 I 1964 I 1965 1967 I 1968 I 1969 I 1970 Figure 3. Monthly rate of occurrence of SFD's at Boulder from October, 1960 through June, 1970. 200 LU Q co -Q E October I960 -June 1970 1884 Events 1 i i I I I I I I I I I I I I I I I I I I I I I 6 7 8 9 iO II 12 13 14 15 16 17 18 19 20 21 22 23 I 2 3 4 5 6 Hours UT Figure 4. Diurnal dependence of Boulder SFD's. 10 regions which were prolific producers of Ha flares and EUV bursts, e.g. in March 1966. Figure 4 shows the number of SFD's as a function of hour, which agrees in shape with the results of Chan and Villard (19 6 3) . This diurnal dependence is caused mainly by the diurnal variation of solar zenith angle; however, a precise analytic formula for the solar zenith angle dependence of SFD's has not yet been developed, (see appendix A) . The occurrence of SFD's as a function of month of the year is shown at the top of figure 5. Note the low values in November, December, and January. The dip at winter solstice is probably caused by the large values of solar zenith angle and low number of hours of daylight. The second graph in figure 5 gives the monthly average number of events per hour of observation for solar zenith angles <_ 80° . The winter solstice dip is only partially removed, which shows that the solar zenith angle effect is more complicated then just a dependence on hours of daylight. Note the peaks in occurrence near vernal and autumnal equinoxes. These peaks are probably the side effects of the main unexpected feature of this figure, namely the dip at summer soltice. It appears also in the percentage of Ha flares (during Boulder SFD observations when the solar zenith angle <_ 80°) that were accompanied by SFD's. The original SFD records were searched and it was found that Sporadic E (Es) , a thin but dense layer of ionization that rather sporadically forms at heights near 110 km, severely influenced the SFD observations, particularly in June, and in some years also in May or July. Sporadic E is known to occur at midlatitudes primarily in the daytime in June (see fig. 3.30 of Davies, 1965). The effect on SFD's of the type of Es involved, i.e., blanketing Es , is that probing radio waves are reflected from heights at about 110 km and are cut off from probing the upper-E and F regions. o At such times, SFD observations are sensitive mainly to 1-10 A soft X-rays, which o usually do not produce much frequency deviation; they are insensitive to the 10-1030 A impulsive radiations that normally produce SFD's. Hence, Es acts like a shutter to SFD observations. When the shutter is open, normal SFD's are observed; hence, their average frequency deviation exhibits no peculiarity. When Es occurs or the shutter is closed, SFD events go undetected; therefore, the effective observing period is reduced near the summer solstice. The presence of Es effects is readily evident in frequency deviation measurements processed with spectrum analyzers. Future SFD observations should routinely list periods with Es . Note that the above diurnal and seasonal dependences of SFD's are related to variations in the sensitivity of SFD's to EUV flashes and not to diurnal or seasonal dependences in the solar EUV flashes. 4.2 Intensity Figures 6 and 7 show the distribution of Boulder SFD's with respect to peak frequency deviation. Note that the intensity intervals are uniform on a log scale, i.e., each interval is a constant multiple of the adjacent interval. In figure 6, each interval includes the number at the upper end of its range and excludes that at the lower end while the converse is true for figure 7, but the results appear to be independent of this difference in definition of intervals. Figure 7 shows similar results by Strauss et al. (1969) using just SFD's reported on 11.1 MHz and confined to observing periods when the solar zenith angle was not too large. 11 Ld Q Li- en 200 150 50 0.09 0.08 in 0.07 03 .c > UJ 0.06 Q Ll. CO o Q 0.05 o Z> 0.04 ^_ o o J_ 003 e 0.02 0.01 1961-1969 1684 Events Average per month (nine years cumulative) 1961 - 1969 £ 1.2 I 1.0 < 3 0.8 C (1) > o uj q: to 5 1 t5 Ll. t5 -Q X -S E E 2 IE .2 i2 P o October I960 -June 1970 1884 Events 109 134 225 202 223 128 163 148 Averoge tor all events . =Q86Hz " 227 146 94 85 966-1968 638 SFD Events 3133 na riares i Average tor all months JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Figure 5. Seasonal dependence of Boulder SFD'q 12 i i i I i i r o 01 «! d d vj o uj E ID CO ^ 00 a; to s o Pq o s o +i 3 +s •^> H is stuaA3 qjs jo jaqajnfsi 13 to V) O i_ cn E o X c o o CL «0 to a o -^ o to > D" 'ri 05 Si S|U9A3 ^o jaqiunfsj 14 They found that above 0.2 Hz the distribution was well fit by a straight line of near unity negative slope; hence, considering the logarithmic choice of intervals, they found that the distribution of SFD's by peak frequency deviation exhibited approximately an inverse squared dependence on the peak frequency deviation. The results in figure 6 for all events shows a similar dependence, except for those below about 0.2 Hz, where some SFD's are probably not detected because of back- ground noise from nonflare ionospheric variations. The relationship also appears to break down above 10 Hz, but this may be the consequence of too few events at such large intensities for a 10 year set of data to sufficiently describe the distri- bution there. The data used in figure 6 are much less homogeneous than those used in figure 7 (e.g., see table 3) , so it is somewhat surprising that they are so similar. o Although the peak 10-10 30 A flux enhancement depends on the start- to-maximum time of the SFD as well as" on the peak frequency deviation (see fig. 2) the results in figures 6 and 7 suggest nevertheless, that the number of EUV flashes have a dis- tribution roughly inversely proportional to the square of the intensity of the c -2 -1 10-1030 A flux enhancement (ergs cm sec at 1AU) . Note that in figure 5, the average peak frequency deviation decreases near the winter solstice, believed to be the consequence of the relatively large solar zenith angles. The average peak frequency deviation in June is not unusually low, despite the summer soltice dip in the other data in figure 5, because the Es effect simply reduces the SFD observing period per day without affecting those SFD's that are observed between periods of Es . The intensity of EUV emission from flares depends upon the location of the flare on the sun; evidence supporting this result and the details of the dependence are presented in sections 5.1 and 5.6.5. Only a small percent of Boulder SFD's, mostly o large SFD's, have been analyzed in detail to estimate the 10-1030 A flux enhancement, A$(10-1030 A) . The peak flux estimates for the analyzed events are given in table 4. 4.3 Rise Time Figure 8 shows the distribution of start-to-maximum times of SFD's. No SFD's have been observed at Boulder with start-to-maximum times less than 10 sec . The individual spikes recorded in some SFD's during the rise to maximum frequency de- viation may have rise times less than 10 sec, but in ten years of observations, no spike has been observed with a rise time of less than 2 sec. (In order to preserve a time resolution of 1 sec in processing the SFD data, the spikes referred to above are those with peak frequency deviations of greater than 0.8 Hz). These lower limits for rise times are also real lower limits for flashes of 10-1030 A solar radiation. The maximum in the distribution of start-to^maximum times of SFD's in figure 8 is believed to be a consequence of a maximum in the distribution of start-to-maximum times of the 10-1030A flux enhancement; however, this result needs further confirma- tion by satellite EUV observations. The sensitivity of SFD's to the time-dependence o of the 10-1030A flux enhancement discussed in section 2 partially causes the peak observed near one minute; but, considering the sharpness of almost all SFD peaks and that the filtering effect should produce blunt peaks, we believe that this peak near start- to-maximum times of one-minute also occurs in the start-to-maximum o time distribution of 10-1030 A bursts. Nevertheless, because of the uncertainty of the SFD filtering effect, independent measurements will be necessary for confirmation. 15 Table 4. Estimates of Peak 10-1030A Flux Enhancements (a) Complete List of Large Events Observed at Boulder, A$ (10-1030A) > 2. max — ergs cm -2 sec at 1 AU Time UT *1 Accuracy Date Time UT *1 Date Start Peak Start Peak Accuracy 11/12/60 1324 1327. 5 11 A 08/28/66 1523.4 1527.2 9 B 01/30/61 2002 2004 2 A 02/27/67 1640 .0 1643 . 8 3 B 09/28/61 2212 2217 3 A 05/21/67 1920 .8 1925 .4 3 B 03/01/62 1636 1639 . 4 7 A 05/23/67 1835.5 1841 > 8 B 04/19/62 1935 . 3 1935.9 5 B 05/23/67 1936.4 1942 3 B 04/20/62 1958 2000 . 7 2 B 07/08/68 1707.3 1710 2 B 04/27/62 1410 141.2.5 3 A 08/08/68 1814.9 1816.4 5 B 04/15/63 1614 1617 3 A 03/12/69 1738.6 1741.3 4 B 08/18/63 1757 1758 3 A 04/21/69 2006 .1 2008.7 3 B 09/16/63 1303 1304 6 A 01/28/70 1916.9 1918.7 2 A 09/20/63 2356 2359 4 A 03/01/70 1530 .1 1531 .0 2 A 03/30/66 1248 1252 4 A 03/0 1/70 2001 .8 2005 .4 3 A 07/07/66 0025 . 7 0037- 0045 5 B (b) Partial List of Medium Sized Events Observed at Bo ulder , .5 < A$ (10-1030A) < — max 2.0 ergs cm~ 2 sec "I at 1 . AU. Time UT *1 Accuracy Date Time UT *1 Date Start Peak Start Peak Accuracy 01/31/61 1511 1514 1.5 A 08/01/67 1727.5 1732 1.7 A 06/15/61 1635 1641 1 .1 A 08/18/67 2120.0 2135 .5 .5 B 09/04/61 1430 1432 1. 5 A 08/29/67 1329 .7 1332 1.5 A 09/04/61 1513 1513 .5 .6 A 08/29/67 1941. 8 1944.9 .5 B 09/04/61 1911 1915 . 7 A 01/11/68 1659 .2 1702 .9 A 09/27/61 1952 1954 .8 A 01/29/68 1538.2 1539 .7 B 03/13/62 1448.5 1450 1.4 A 02/01/68 1801. 7 1802 . 7 .7 B 10/13/62 1805 .2 1805 .5 . 5 A 02/01/68 1917.1 1919 .2 .8 A 04/19/63 1754 1758 1.0 A 02/14/68 1534.2 1535 .4 . 7 A 09/15/63 2017 2018. 5 0. 7 A 03/21/68 1913.3 1915 .8 A 09/17/63 1927 1928 . 7 A 03/25/68 1459 .2 1505 .0 1.1 A 09/19/63 2257 2300 1. 3 A 09/29/68 1617 . 1 1620 1.8 B 05/20/65 2320 2321 0.9 A 11/0 1/68 2002 .7 2004.5 1.2 B 06/05/65 1808 1810 . 7 A 01/17/69 1704.1 1704.8 1. 8 A 10/02/65 1413 1414 .5 A 02/09/69 1723 1725 1 .0 A 03/20/66 1759 1802 .6 0.6 B 02/27/69 1403 1408 .8 B 03/31/66 1856 1905 .0 .5 B 03/01/69 2139.8 2148 1.2 A 04/12/66 1718 1719 0.6 B 03/27/69 1323 1341 1.7 A 06/25/66 1527 1534. 4 1 .1 B 05/17/69 1922 .4 1934 .5 A 09/18/66 1453 1456 .6 B 06/06/69 1604. 1 1606 .4 1.5 A 10/13/66 1334 1344 1 . 1 B 08/11/69 1215 .4 1221 1.2 A 10/24/66 1502 1503 . 7 A 10/08/69 1630 . 2 16 30 .8 .9 A 12/09/66 1756 1800 .9 B 10/24/69 2111.8 2114 . 7 A 03/04/67 1715 .6 1716. 2 1.1 B 11/20/69 1619 .6 1620 .9 1.0 A 03/26/67 1604.6 1606 .7 . 7 B 11/22/69 2123 .6 2124 . 5 1.4 A 05/21/67 1534. 3 1539 .8 1.0 B 03/26/70 1726 .6 1728. 3 .6 A 07/24/67 0033.3 0035 .6 B 03/26/70 2004. 3 2007.8 .5 A 07/26/67 2329 . 3 2331 .6 A 05/09/70 1559 . 8 1600 .6 .9 A Accuracy: A - Absolute flux-enhancement accurate to within a factor of 10, usually analyzed by methods 4 or 5 discussed in the Appendix: B - Absolute flux enhancement accurate to with- in a factor of 4, usually analyzed by methods 3 , 2 or 1 in the Appendix. *1 — 2 - 1 A (10-1030A) ergs cm sec at 1 AU. max 3 16 1 1 1 1 1 n. l E o A I C/l 01 E i- r- CT5 E 4 D i E L x o 5 1 i o i — OO *_ o CO c 1 S C o CO II c > L_ 01 UJ E E r^ X CO to sr O CD CO 5 00 i c o >< o i E o 3 o * < 3 c o CO n n ~3 0) a> E | o 01 > T3 • E c 1 1 < 2 2 J 1 r^ L I 1 ■ 1 I 1 1 1 | 1 1 1 I 1 1 I 1 1 1 i i i 1 1 CD o CO to s o o K Q 3 is CO -a S i 5Hz) have a significant fine structure. Recent studies have suggested max — a quasi-periodicity in the impulsive emission of flares (Parks and Winckler, 1969a, b; Janssens and White, 19 70) . Because of the physical significance of such a quasi- periodicity, a detailed study of the periodicity of structure in SFD's was made and the results are reported below. 18 I I I I I o IT) Al O q) a> c -J I O 10 O o O co c c e e .. II c c o o '•►- TJ O <1> Q 01 o> a k- < UJ r J_L i i i (0 Cl Bt! to ir> 5s rO 0) 3 o Pq o ^ r-O l/> o u s 3 o c •^> 5 +i n s LO o ►Q CM •^ o Ss k_ -+-* 3 o to Q tJ U. CO K o o> — oo — r — — <-o LO S|uaA3 qjs jo jaquunfyi 19 Structured SFD's were selected for study using the following criteria: (a) the peak frequency deviation Af >_ 0.5Hz, (b) the number of major peaks >_ 4 and (c) the Af (t) trace was well defined, i.e. free of smearing, fuzziness or spread. Criteria (a) o and (c) were necessary to be able to accurately scale Af(t) and compute A$(10-1030 A, t) . The "major peaks" in criterion (b) means either the main peak (Af ) or a peak with Af , > 1/3 Af and the adjacent minimums in Af being < 2/3 Af . . Boulder SFD's peak — max J 3 — peak from March 19 66 through June 19 69 were systematically searched and then several older events were added to the study. The first criterion (a) eliminated about three-fifths of all SFD's from the study. Criterion (b) eliminated about four-fifths of the remain- ing events. Less than 7% of all SFD's satisfied all the criteria which means that relatively few SFD's exhibit pronounced complexity, much less quasi-periodicity . The list of complex events studied with information on the "periodicity" of o their structure is given in table 5. For these events, the 10-10 30 A flux as a function of time was computed, plotted, and processed by a Fast Fourier Transform subroutine. The spectrum was plotted and examined for quasi-periodic peaks. Ten events devoid of fine structure were similarly spectrum analyzed to examine the noise in Fast Fourier Transform spectra for this type of data. Figure 10 shows the best example of a quasi-periodic SFD. In this case, the o fine structure of the SFD is still quite apparent in the computed A$(10-1030A) flux. The o ruled lines below the A$ (10-10 30 A, t) curve in figure 10 point out sequences of peaks that are quasi-periodic, but not uniformly spaced. Figure 11 shows the spectrum of the structure for the same event. Several of the spectrum peaks corres- pond to the same frequencies for which the rulings are shown in figure 10 , namely frequencies f , , f , f, and f . Note that f - 2f, , f, = 3f, and f, = 2f . Also ^ b'cd e c bd b b a the spectrum exhibits a tendency towards peaks at odd harmonics of f and minimums a at even harmonics of f . a Figure 12 shows the distribution of major periods of the EUV fine-structure. There is a cut-off in structure for periods less than 5 sec. Most events have their main periodicity in the 10-30 sec range. Some of the events in table 5 were not o periodic. For most of the events, the fine-structure in A$(10-1030 A, t) was relat- o ively small compared to A$ (10-10 30 A) , which is denoted by comment "A" in table 5; hence, events with strong quasi-periodicity, like that shown in figure 10, are very rare. Consequently, if there is a physical process in the flare region causing the periodic structure; it probably does not play a major role in most flares . Two classes of processes may be involved in producing the quasi-periodic fine structure: (1) fixed location emission regions that oscillate in intensity as a consequence of either magnetohydrodynamic waves in the flare region or bunches of energetic particles mirroring back and forth along magnetic tubes or (2) several emission cores, spatially separated, that occur at different times during the "flare in a time sequence, where each core produces one peak but the set of peaks occur nearly uniformly spaced in time, either accidently or as a consequence of a spatial structure in the flare region. The statistical properties of SFD's are summarized in table 6. 20 Table 5, SFD's with Complex Time Structure Approximate SFD Peak Time UT Pe riods of Fine Structure sec Comments 11/12/60 09/28/61 03/01/62 04/20/62 05/01/62 03/20/66 03/20/66 03/21/66 03/30/66 03/31/66 04/12/66 07/07/66 08/28/66 09/18/66 09/20/66 02/24/6.7 02/27/67 03/22/67 03/27/67 03/28/67 04/11/67 05/20/67 05/21/67 05/21/67 05/23/67 05/23/67 05/03/68 07/08/68 07/09/68 08/08/68 08/21/68 09/29/68 12/29/68 02/09/69 02/27/69 03/01/69 03/12/69 03/12/69 03/20/69 03/20/69 03/27/69 03/29/69 04/21/69 05/02/69 05/22/69 05/22/69 05/29/69 06/11/69 1328 2217 1639 2001 1917 1858 2002 2027 1250 1902 1718 0027 1527 1456 X712 1905 1644 0031 2112 1738 2110 2004 1539 1924 1838 1938 2128 1710 1812 1816 1841 1619 1922 1725 1407 2141 1740 2008 1632 2149 1330 2001 200 . 1749 1901 1935 1942 1621 37, 118, 67, 46, 22 37 , 21 67, 25 20 50 , 21 30 11 14 32 35 15, 11 36 , 25 28 23 , 35, 70 , 122 85 , 30 17, 20 22 14 , 7 9 , 12 14, 17 31 6 , 11, 39 , 20, 73 43, 15 , 21, 9 17, 27 21, 11,' 43 28, 130, 64 17, 11 64, 29, 120 17 , 8, 32 17 8, 5 14 34, 100 10 4 60 , 21 28 9 , 18 10 , 18, 76 27 15, 33 20 38, 24, 73, A , C , A, C A A A A,B A A A A , B B A A, D A C, D A, D A, D A , B C D B A A A, D A , B A, C, A , C , A , C , A , B 19 C , D , > 10 major peaks . Lettered Comments: A - Fine structure strong in SFD but relatively weak in A ( 10 -10 30A ,t) because of a large slow enhancement in A* ( io- 10 30A) . B - Strong fine structure in SFD but most of it not quas i -per iodi ca 1 ly spaced. C - Nearly harmonic relationship among some of the major periods of fine structure. D - Much quasi -periodic fine time -s tructure in SFD. 21 CD E to Q> O, I to « Ss S (^ 0201-01 ) XDUJ 10 keV) Kane and Donnelly (1970) studied the relationships between SFD's and satellite o measurements of hard X-ray emission (A < 1 A or photon energies > 10 keV) and established the following: o (1) The occurrence of an EUV flash large enough to produce a distinct SFD (A$ (1-10 30A) -2 -2-1 > 10 ergs cm sec at 1AU) is accompanied by the occurrence of a hard X-ray burst. o (2) The intensity of 10-1030 A flashes observed via SFD's is approximately linearly ° -2 proportional to the hard X-ray flux, the ratio of 10-1030 A flux (ergs cm sec at 1AU) to the hard X-ray flux being about 10 (see table 2 of this report) . o (3) The ratio of 10-1030 A flux to the hard X-ray flux varies on the average with the central meridian distance (CMD) of the associated Ha flare, generally de- creasing with increasing CMD. (4) The detailed time structure of the EUV flash deduced from SFD's is very similar to that of the impulsive hard X-ray burst (see also Donnelly, 19 69c) . Kane and Donnelly (19 70) have proposed a model to explain the above results wherein non-thermal electrons with a power law energy distribution (energy decreasing with increasing electron energy >_ 10 keV) produce the hard X-ray burst via bremsstrahlung 12 -3 emission in a region of ambient hydrogen or proton density < 10 cm , and they produce the EUV emission via collisional ionization and excitation with subsequent recombination and line emission at the bottom of this region where the ambient 12 -3 hydrogen density > 10 cm The CMD dependence in (3) above was suggested to be caused by absorption of the EUV emission in the solar atmosphere. Numerous hard X-ray bursts have been observed that were not accompanied by an SFD (Kane and Donnelly, 1970) , but this is probably because these events were too small for the EUV flash to be detected by SFD's. I would expect any impulsive hard -5 -2 -1 X-ray flare (rise time < 1 minute) with a peak flux greater than 10 ergs cm sec at 1AU at photon energies >_ 10 keV that was not a near-limb event to surely be accompanied by an SFD providing good quality SFD observations were being made; so far, no such event has been found that was not accompanied by an SFD. The detailed CMD dependence of the ratio of hard X-ray flux to EUV flux during impulsive bursts is not yet precisely known. Also, for a given small range in CMD, this flux ratio varies appreciably from one flare to another, perhaps because of experimental errors in the EUV estimates (see sec. 2) or perhaps because of varying amounts of EUV absorption depending upon whether or not filaments or other active region structures lie along our line of sight to the impulsive EUV source regions. These items will probably require simultaneous EUV and hard X-ray measurements via satellite that are more sophisticated than those planned for the present solar cycle. The long-duration events, the CMD dependence of occurrence and average size of SFD's, and the limb events, discussed in section 5.6.5, further determine the nature of this CMD dependence. 25 Kane and Donnelly (1970) suggest that the size of the EUV flash is more dependent upon the total energy flux above 10 keV for the impulsive X-ray burst than upon the spectrum of the hard X-ray radiation. A more precise determination of this result for more events seems desireable. Also, although the detailed time dependences of EUV flashes and hard X-ray bursts for impulsive events are very similar, there occassionally appear to be slight impulsive differences (Donnelly, 1969c) that need further confirmation to prove whether they are real or are caused by experimental faults. Futhermore, Kane and Anderson (1970) have found that the decay time of a hard X-ray burst decreases with increasing photon energy, or the spectrum varies with time. Although the time dependence of the EUV flash agrees closely with the hard X-ray burst, e.g. the EUV decay time of the fast spikes in the August 8, 1968 event (Donnelly, 19 69c) appear to be as fast as that of the hard X-ray burst to within a few seconds, there must be a systematic difference in timing with increasing hard X-ray energy, considering Kane and Anderson's result. This systematic difference should be quantitatively determined with better EUV measurements for more events. o Slow differences -between hard X-ray bursts and the 10-1030 A emission deduced frcm SFD's are to be expected, both because of errors in estimating the electron- loss o rates in the ionosphere (see sec. 2) and because SFD's respond to the whole 1-1030 A wavelength range, which includes slow soft X-rays and slow EUV emissions in addition to the impulsive EUV emissions that are related to the hard X-ray burst (see sec. 3) . Figure 13 shows a hard X-ray burst observed by Frost (19 69) and the associated SFD which is an example of good timing agreement in the impulsive bursts but also of a slow difference in time dependence. Note that the SFD remains above its preflare o level until 214 8.3 UT ; the major peak in the 1-10 30 A flux enhancement (not shown) occurs at about 2148 UT, well after the major hard X-ray peaks and the corresponding secondary peaks in the 1-1030 A flux. This is one of the long duration events discussed in section 5.6.5 and is associated with a near-limb flare (89°W) . I interpret this as a case where the impulsive EUV emission is relatively small compared to the slow soft X-ray and slow EUV emissions because the impulsive EUV emission encounters relatively high absorption in the solar atmosphere. The slow emissions are probably associated with rising arches and suffer negligible absorption because of the high solar altitude of their source region. o 5.2 Soft X-Ray Bursts (.A > 1A) Soft X-ray flares generally start earlier, rise more smoothly, peak later, and last longer than the EUV flashes that cause SFD's (Donnelly, 1968d, 1969b) . Soft X-ray bursts contribute less than about 5% to the peak frequency deviation (see o table 2) and less than 14% to the energy flux enhancement in the 1-1030 A range o at the time of the main impulsive peak in the 1-10 30 A flux. Note tha't this latter o result is not the same as the percent ratio of peak soft X-ray flux to peak 1-10 30 A flux, which would be > 14%. Figure 14 illustrates that the soft X-ray flare is a different component of the flare than the impulsive component that radiates the EUV flash that produces SFD's. There are a few cases of broad-band soft X-ray observations that suggest a small impulsive component but for all practical purposes it is buried 26 2140 2145 2150 UT 600 55 to 82 keV 2140 2145 Time 2150 UT Figure 13. Comparison of timing of hard X-ray burst and SFD of 2141 UT March 1, 1969, 0S0 5 X-ray observations made by Frost (1969). * 27 I — max Radio Emission at 2800 MHz or Higher Frequencies 7 $ E u Q T3 O rr o E i — max $> max X < 0.1 A Hard X-Rays °0 Figure 14 impulsi lengths tial po The gra ent tim lengths where i IV Comp 10 20 Time, Minutes 30 Illustr ve oompone 3 hard X-r rtions of dual or si e dependen j most of t is somet onent, C s ation of components of flare radiation . The nt A occurs concurrently at centimeter wave- ays, certain EUV wavelengths , certain spa- Hot, flares, and occasionally in white light, ow component , B occurs with slightly dif fer- ae in soft X-rays, certain other EUV wave- the Ha flare, and centimeter wavelengths imes called a postburst increase . The Type occurs mainly at radio wavelengths . 28 by the slow component (Donnelly 1969b) . Although the slow component (B) in soft X-rays is quite different from the impulsive component (A) observed by SFD's, the two components are related. For example, A normally occurs during the rise of B. Numerous soft X-ray flares observed by Mr. R. Kreplin of the U.S. Naval Research Laboratories with SOLEAD satellites (listed in Solar Geophysical Data ) that were large _o enough to produce D-region types of SID's (SWF, SPA, SCNA, etc., A$(1-8A) >_ 10 -2-1 ergs cm sec ) were not accompanied by EUV flashes large enough to produce SFD's ° -2 -2 -1 (A$(1-10 30A) £ 10 ergs cm sec , with a rise time <_ 1 min) . However, no SFD events have been found that definitely were not accompanied by a soft X-ray enhancement. In other words, often component A and B both occur, B may occur without A, but I have found no evidence that A occurs without B. This may be the result of high sensitivity in the satellite measurements of soft X-rays (B) compared to the measurements made of the A component (satellite hard X-ray and EUV measurements, SFDs and ground based microwave measurements) . The distribution of the time and wavelength integrated flux of component A relative to that of component B should be studied. 5.3 Ultraviolet Flashes (10 30-3000 A) o Few satellite observations of ultraviolet enhancements at wavelengths > 1030A o have been made except at wavelengths near HLymana 1215. 7A. Observations of Lyman a usu- ally show only small percentage enhancements (< 20%) even for flares of Ha importance 3 (Hallam, 1964, Hall and Hinteregger, 1969). Such enhancements are still relatively -2-1 ° large in ergs cm sec compared to the energy flux enhancements at £10 30 A because of the large non-flare energy flux in Lyman a; consequently, the value R$ in table 2 is quite large for Lyman a. The R$ results for Lyman a vary from one flare to another from 25% to 250%, perhaps because of experimental difficulties in observing the Lyman a flare. However, SOLRAD-8 satellite measurements of the proton flare of 1525 UT August o o 28, 1966, in the wavelength intervals 1050-1350A and 1225-1350A have been interpreted by Grebenkemper (1969) and McClinton (1968) (Friedman, 1969) , as being caused mainly by a continuum enhancement and not by a Lyman a enhancement , because of the large ratio of o A$max(1225-1350A) ^ n ,_ ._ , . ■,n*n\ ^ *.-> ±.u -, 4-u -, ^ b o- > 1/3 (Grebenkemper, 1969). Consequently, the wavelength depend- A$max(1050-1350A) ence of 1030-1350A impulsive flare radiation should be measured and studied further in order to resolve the true nature and intensity of this radiation. The time depend- ence of both Hall and Hinteregger ' s (19 69) observations of flares in this range and the NRL results for the August 28, 1966 proton flare agrees well with the corres- ponding EUV flash deduced from SFD data. No satellite measurements of flare radia- o tion in the 1350-3000 A range have yet been reported to my knowledge; I understand o however, that Nimbus 3 made solar flux measurements in the 1100-3000 A range after April 14, 1969 (Heath, 1969). 5.4 White Light Flares Mcintosh and Donnelly (1970) studied white light flare patrol films of Sacramento o _2 -1 Peak Observatory at times of SFD's when A max (10-1030 A) >_ 2.0 ergs cm sec at 1 AU and found the following: 29 (1) For the five such cases when white light patrol films were available, three were definitely white light flares and the other two were suggestive of faint white light emission but the seeing was too poor to be certain. o o (2) The white light flash (3500-6500A) and EUV flash (10-10 30A) were roughly comparable in timing and flux enhancement . (3) The white light emission areas and, hence, probably also the EUV emission cores, varied from about 2 to 15 arc seconds in diameter. (4) The white light emission cores and, hence, probably the EUV emission cores, lay adjacent to the penumbra of strong sunspots, sometimes covering small umbrae, but never over the larger and very strong spots. These cores occurred in two or more places located near and on either side of a longitudinal neutral line in the magnetic field. This line lay between the leader sunspot of a relatively young sunspot group and the follower of an older group to the west of the line, with the separation between groups being less than one heliographic degree. Considering the success of the above study, it may be possible to find more ° -2 -1 small white light flares by checking SFD's when A$(10-1030 A) >_ 1/2 erg cm sec at 1 AU, especially since white light flares are more easily seen for flares at large central meridian distances (CMD) while the EUV emission is relatively weak for large CMD. Further quantitative study of white light flare emission should be made because such observations could provide excellent information on the size and location of the source region of the most energetic portion of flares. High time- resolution white light measurements should be made (~ 1 frame per sec); such a high film speed could be automatically triggered at times of large impulsive flares by either SFD or microwave radio measurements. 5.5 Low-Chromosphere Optical Line Emission Unfortunately, white light flare emission is observable only for very large flares; Mcintosh and Donnelly (1970) estimate about 5-6 events per year during the present solar maximum. Optical observations other than white light provide good observations of the impulsive portion of flares smaller than white light flares. Several flares observed by Lockheed observatory (Ramsey et al . , 1968) at 5324 + 0.15A in an Fel line, which is a low chromosphere line, were studied with respect to the associated SFD. At that wavelength, sunspots and penumbra are visible, which makes it easy to determine the location of the flare emission relative to the sunspots and magnetic fields of the active region. Unfortunately, the flare emission at this wavelength is usually quite faint and moderately energetic flares are still required for quantitative study. Furthermore, good seeing is essential. Unfortunately, the particular flares studied, which were selected because the SFD data were of sufficiently good quality, were cases of fair to poor seeing. Figure 15 shows the best example studied. Based on the size of the EUV flux enhancement deduced from SFD data, Mcintosh and Donnelly (19 70) suspect this event to be a white light flare. Although a white light flare patrol film exists for this event, it has not yet been available for study to determine whether a white light o flare occurred. The 5324 A flare observations studied may be partly continuum emission , rather than just Fe line emission. Figure 15 shows a sketch of the preflare 30 Sunspots Approximate Penumbra Boundary PREFLARE SPOTS A-2O08 : 45UT only ~ 1° Longitude Fe LINE FLARE JWOOfrOOUT xand thereafter (Ha) for the impulsive component. 5.6.3 Sunspot Dependence Dodson and Hedeman (1970) have studied Ha flares of importance >_ 2 from nearly spotless active regions. Of their 83 events, only 21 occurred when SFD observations were made at Boulder. In 7 cases, SFD's were observed, so that 60% of their impor- tance 3 flares were accompanied by SFD's and only 33% of their importance 2 flares. These percentages for flares from spotless regions are noticeably smaller than the results in table 6, but not greatly smaller. The most remarkable result of this study was that none of the seven flares from completely spotless regions (their type A) were accompanied by SFD's or impulsive EUV flashes, while 50% of the events from regions with very small spots (type B) were accompanied by SFD's. This latter per- centage is consistent with the results in table 6. These results suggest that SFD's or EUV flashes do not occur with flares in completely spotless regions , at least small spots are required. More events should be studied for further verification. Another parameter, besides brightness, which probably influences the relationship between Ha flares and SFD's (or EUV flashes) and which has not been adequately studied is the impulsiveness of the Ha rise time. However, the study discussed in the next section is probably a more fundamental study of the dependence on Ha impulsiveness than just examining the start- to-maximum time of Ha flares. 5.6.4 Association of EUV Bursts with the Ha Explosive Phase Moreton (1964) has defined the explosive phase of the Ha flare as "the short period, commonly less than 30 seconds, during which time part or all of the flare borders undergo accelerated expansion" (~ 100 km/sec) . During some explosive Ha flares, traveling disturbances propagate away from the flare across the solar disk causing disturbances in quiescent solar prominences; sometimes matter appears to be blown off the sun; but in many cases, such associated effects are not observed. The explosive phase sometimes occurs when the Ha flare is well developed, when a portion of the Ha flare boundary suddenly expands; but for most of the events studied in the present paper, the explosive phase occurred during the flash phase or the rapid portion of the rise in Ha intensity. The explosive-phase classifications used in the present paper were made by the Lockheed Observatory staff * using single-frame projection of 35 mm Ha flare patrol films. Table 7 shows a strong association of SFD's with explosive phase flares. In Explosive-phase data were provided through the courtesy of Mr. Harry Ramsey of the Lockheed Observatory. 33 if) CD v_ _g L±_ CD if) O _£Z Q_ I CD > if) O QJ X LU "O cd 'c o CL E o o o < if) Q Ll_ CO CD -Q E 3 CO Q Li_ CO CD E 13 O .o 1 i i i i i i i 1 — 1 — CM CD CO CO J_ J 1 1 1 1 1 1 O o> T3 c rti • CD W CO - id Q ^ En Pn W 1 CD ^ > -H m 10 CD rH H Oj X! X rd w Eh c >i m o C X! CD -p CJVH C £ ■H +J CO a CD H u rd rH En r^ 3 X 0) H X! fd H CO OO CO LO ro OO |U80J8d ^ LU O +i s cd cd O CO CO 4^ a e « a> v-A ^ C^l >- 03 «+-, Ss -W Al V r£ CD O +i 3 rQ CO 2 O 0) £ O LO cd LU co £ CD Al V Z> o enee -pha The r de LU ■a cd a £ » • rQ OO CD LO CD U- 0) "^ Cl &, co r^ S CD O Cr; « • Al V "XJ ^ SL, ^ PX, cd C3i Q) *: CD B <-£ O ^ ^ (» -^ +iM OO < Eh co O ^ c,-V^ ^ CD LU V CL «0 3 S5^ r-H 4^ O CO CO '^ « 4^ Q CD - CO CD s £h d £ CD U_ 3 Pt, cd «k ^ CjjCo +i O CD (/) CO D 245 SFD's 00 00 2123 Ha Flare CD •— s ^-v > CD if) if) o o o OO 0O CM CT> ^ o O £ ro £, <\J Z Q.-C 00 O x 0- C\J UJ CD ««-* o i xplosiv Phase i CM ! r-l LO ON oo v 102 *CO CL X UJ >^ _Q TD CD "c o £ o o o < _co Q U_ CO CD E _co b CO o CD _Q E a .0 a _CO Q Li_ CO ID CD 'cz o Q. E o CJ O < CO CD CD CO O XI CL I CD > *co _o CL X CD E 13 CO CD ^_ _o Ll CD CO a CD > 'co _o CL X LU CD E Z3 O .O rO OJ PO CVJ >- \- 3 u. c -) ai o a en u. n o X QJ - — O *> X e i.o 0.8 — O O lost O rO O o E o < 0.6 0.4 0.2 — % S cr > a> a; 1500 3.0 1500 1505 UT -1 ' i IN V ' ' ' -!• October 24, 1966 / - -^^_ — 2.0 — / — i n ~l* ~~*~\ 1 "i | 1 1 I ft" : j \ Computed from SFD Dato — i i ^^y i 1 1 1 1 _L 1 1 1 1 1 * * j I i T | SFD j X — | 12 1 MHz WWI Havana, III to Boulder Colo. 1290 km 1 • 1 - J- i i ■ 1 , 1 I t 1 1505 UT Time Ftgure 18. Association of an EUV burst with a small bright Ha kernel 37 Events at 0° were alternately added to the 0-5° West and the 0-5° East range. The range marked 85-90° really means CMD > 85°. The number of Ha flares peaks sharply at CMD > 85°, probably because we're seeing many flares where the center of the emitting area is located beyond the limb so that the effective CMD range is greater then 5° . The number of Ha flares with SFD events decreases toward the limb with a weak peak at the limb, much weaker than the large peak for all Ha flares. Consequently, the percentage of Ha flares with SFD's or EUV flashes peaks at CMD £ 5° and decreases appreciably for large CMD for both the east and west limbs . This result is evident for each year studied as well as for all five years and also for Ha flares excluding subf lares. Data for 1964 and 1965 are not shown separately, because the number of events was too small to show significant results. Several of the secondary peaks at intermediate CMD values, e.g., 60-65° East, which is particularly evident in the percentage figure for 1967, are largely due to several flare-prolific active regions and are probably not indicative of a preference of EUV emission from those locations. The observed CMD dependence could be caused by a bias in Ha flare observations for near- limb flares, a decrease in EUV emission with increasing CMD, or both. The bias in seeing Ha flares centered beyond the limb, discussed above, probably prefer- entially selects large flares; therefore, since large Ha flares are statistically more frequently accompanied by SFD's than small Ha flares, this type of bias should result in the percentage of Ha flares with SFD's increasing at the limb. This may explain why the percentage of Ha flares with SFD's for CMD > 85° is comparable to that for 80° < CMD <_ 85° rather than being less, but this effect is opposite that required to explain the near-limb results compared to the center-of-disk results. There are undoubtedly other biases in Ha. flare observations that are not accounted for by present correction factors, e.g., the background Ha emission decreases with increasing CMD thereby, perhaps, making small faint flares easier to detect. The Ha flare area observed near the center of the disk is the area projected on the solar surface, but flares observed at the limb must depend partly upon the radial extent of the Ha flare. For further discussion of problems in the CMD dependence, see Smith and Smith (1963). Studies of soft X-ray emission from flares (Thomas, 1970, Dodson and Hedeman, 19 64) suggest that the importance of Ha flares near the limb tends to be underestimated. Considering the statistically increasing association of SFD's and Ha flares with increasing Ha importance, the bias in the CMD dependence of Ha flares is probably not the main cause of the CMD dependence of the percentage of Ha flares with SFD's. Consequently, the relative strength of impulsive EUV emission from flares probably decreases with increasing CMD , particularly near the limb . 5.6. 5. b Peculiar Long-Duration Events During the past ten years of SFD observations, a number of events stand out as being rather unusual compared to the average characteristics of SFD's, e.g., events with very extensive fine structure (see sec. 4.5) . Another type of peculiar event that has an unusually long duration of the frequency deviation exceeding the preflare 38 440 400 360 320 B 280 ? 240 -O E 3 120 40 32 CO Q Ll 28 CO .c 24 3 a> 20 D U_ 16 a T 1/ O a> 8 o F Z3 4 1? to Q Ll CO 10 „ -C a> — 8 0>"5 o 5 r- 10 6 a a T' 50 20 10 in 16 n Ll CO 14 T 3 12 ir> (I) o 10 Ll_ a 8 j. ■ i o 6 (i) -O F 4 3 Z 2 in 16 O ll CO 14 r 3 12 (/> 0) o 10 u. a 8 X "o 6 o O) 4 C7> O (I) 2 Ci 0) LL t — r— r 1968 i — i — ; — i — \ — i — i i i r I i I i — i — r During Boulder SFD observations when the solar zenith angle was S 80° n l i i — i — i — i — r 2 ,682 Events 1 — r 199 Events Average for all CMD^ I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 90° 80° 70° 60° 50° 40° 30° 20° 10° 0° 10° 20° 30° 40° 50° 60° 70° 80° 90° East West Central Meridian Distance of the Ha Solar Flare Figure 22. Dependence of EUV flashes on the central meridian distance of the assoczated Ha flare for 1968 for all Ha flare classifications 42 ° g. fe E 1° Ll a x a. P £ E _o — E^ o a. a> p w g»— Q 2 *- Ll c: O (f> if aj Q_ Ll o-| a x < o M 4 a x ~r~i — i — i — i — i — i — i — i — i — i — r~ i — r 1964-1968 During Boulder SFD observations when the solar zenith angle was S.80° "i — i — i — i — r 801 Events 217 Events .Average for oil CMD " I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 90° 80° 70° 60° 50° 40° 30° 20° 10° 0° 10° 20° 30° 40° 50° 60° 70° 80° 90° East West Central Meridian Distance of the Ha Solar Flare Figure 23. Dependence of EUV flashes on the central meridian distance of the associated Ha flare for 19 64-19 68 for Ha flares of importance >1. 43 en c J •p a) ►J a. e H T3 E-i c S W 1) e ■H Eh X ^^ rd S tu H a rH ■P h H m J? p 32 oi X H Ph u 03 X D m m m in ro ro in in co > H w w H s 3 s w w s 3 s s « * s H H H M 3 3 3 3 S 3 S 3 in u) io o o o cn rji "* rH rsl CM IN r^ iH cri in in in ID 10 ro m oj in H Oi H IB (Ji (^ 01 tu ^0 ID KD ID 1.0 ^0 X X \ V \ V, H T CO m H r~ o o H cn O CM \ S. •s, \ \ \ 10 CO CO o ro ro o o o ^H O o 3 s 3 3 n £ OJ fO 4J l-l ^H ■H Cn C C 03 CD < < o c 01 -H < -- 4J P •H ■H <1) 6 +1 J Ul ra C ■P H rd 03 H s o ~ rd ~ £ N MH X < -- 44 values is illustrated in figure 24. The value of X is unusually large relative to Y. Furthermore, the fine structure occurs early during the event and the latter portion of the positive frequency deviation is unusually free from fine structure. (Another example of a long-duration event is the event of March 1, 1969, discussed in sec. 5.1 and shown in fig. 13) . A special study was made of these long-duration events. Boulder SFD data from March 1966, through May, 1970, and published figures of SFD's from October 1960 through March, 1966, (Agy et al., 1965, Baker, 1965) were searched for such events using the following criteria: (1) X/Y > 5 min Hz 1 (2) X > 5 min (3) Y > 0.3 Hz (4) Fine structure does not dominate during the latter portion of the period of positive frequency deviation. (5) Negligible frequency deviation from nonf lare-induced ionospheric disturbances. (6) SFD observed on a propagation path reflected in the F-region and not at the bottom of the E layer or from Sporadic E. (7) The Ha flare associated with the SFD could be unambiguously identified. Criterion 3, which insures that the SFD was large enough to be accurately scaled, disqualified about 40% of all SFD's. Criterion 2, which insures the event is of long duration, disqualified another 40% of the SFD's. Criteria 1, 4, 5, 6 and 7 disqualified nearly all the rest of Boulder SFD's except those listed in table 8. Criterion 1 insures that the event has long duration relative to the size of the event, criterion 4 insures that it is not simply a long-duration impulsive event (e.g., the July 8, 1968 event, see sec. 6), and criterion 5 disqualifies noisy SFD's. Criterion 6 is necessary because SFD's observed on propagation paths reflected from the bottom of the E layer are produced mainly by soft X-rays, which have relatively slow enhancements during flares, such that these SFD's often satisfy criteria 1-5, even when the SFD's for the same event for F-region propagation paths are very impul- sive. One might suspect that the above criteria may be simply selecting weak events, i.e., selecting Af small enough that the first criterion is satisfied. However, peak frequency deviation and duration are roughly correlated so that if one decreases the EUV flare emission he would usually have to also decrease the duration of that emission. Among the seventeen events in table 8, there are some that look distinctly different from the event in figure 24, yet they still satisfy the above seven criteria. These events have essentially no impulsive structure, even at their beginning, and their Af(t) trace becomes very faint because of high ionospheric absorption of our probing radio waves induced by the associated soft X-ray flare. Such behavior is typical (and much more frequent) of SFD's observed on radio paths reflected from the bottom of E-layer, so a careful study of the preflare propagation paths and fuzziness and wiggliness of the preflare frequency deviations (F-region Af (t) traces are dist- inctly more fuzzy and wiggly than bottom of the E-layer or Es traces) had to be made to properly identify these events. These events suggest that more such long duration events exist than are indicated by table 8 because, if the absorption was too great and the SFD trace was lost, the X value could not be determined and the event was 45 LO CD E ro ro ro 0) » M ■fi •v^ 03 • O Ai Cu « 0) ^Ss o rQ s K o O •^ •^ +i -^ a a fc s •^ *X3 3 S Q 0) fei *X3 CQ RQ s CQ a H Q> 0) ■XJ s Si ^ O s 05 • 3 ^ Cr eg CD is 0) =K fc 3 535 •^ fe. 46 disqualified. Furthermore, because of their slowness and lack of sizeable frequency deviations, such events could easily have been ignored and not listed when the origi- nal data were processed to detect SFD's. One of the best examples of this type of event is the SFD of September 10, 1961, shown in figures 33 and 34, pp 52-5, of Baker (1965) . Nine of the seventeen events (53%) were at central meridian distances greater than 70°, while for all SFD's observed in Boulder only about 15% are at CMD > 70°. Considering the CMD dependence discussed in the previous section, a possible qualita- tive explanation of these peculiar long-duration events is as follows: The impulsive EUV radiation from source regions at relatively low heights suffer considerable absorption by the surrounding cool atmosphere in the active region and surrounding solar atmosphere, especially for the line of sight for flares near the solar limb, slow radiations including soft X-rays and certain EUV lines are emitted from relatively high source regions and suffer negligible absorption compared to the impulsive EUV emissions. Because the impulsive emissions are relatively weak, the peak frequency deviation (Y) is fairly small and the frequency deviations are impulsive mainly early in the event. Because the slow emissions are large and unattenuated, the duration of Af > (X) is anomalously large. If the same flare occurred near the center of the disk so that the EUV source were observed looking through the overlying ionized flare region, negligible absorption of the EUV flash would occur and a very large impulsive SFD would be observed. The recovery from the EUV flash would quickly drive Af below zero even through the slow emissions were the same as for the near-limb case. Hence, the same flare at the center of the disk would have a much larger Y and smaller X. An alternative explanation is that the emission of the impulsive component of these flares was relatively weak. However, events like that of March 1, 19 69, (discussed in sec. 5.1) exhibit large impulsive microwave and hard X-ray bursts which show that the impulsive component of these flares was quite large. The flares in table 8 at CMD < 70° have a slight tendency toward large solar zenith angles (\) which has a similar effect of attenuating the impulsive EUV emission relative to the slow emission in the ionosphere. This is because the latter emission o o is large in the 1-100 A range which is more penetrating than the 100-800 A range. The validity of this effect can be checked by studying SFD's observed at two locations where large solar zenith angles are involved at one location and fairly small angles at the other. Some of the events neither fit the large CMD or x explanations, e.g., those of June 1, 1967 and February 11, 19 70. They can perhaps still be explained by the same basic mechanism, i.e. by relatively high absorption of the impulsive EUV emission by the solar atmosphere along the line of sight through the active region. For example, filaments or prominences may act as an absorbing screen. However, Boulder Ha obser- vations for the active regions involved show no large filaments near the flare and no unusual limb prominence at either limb passage. Perhaps these two cases are better ex- plained by the alternative explanation, i.e. the impulsive component of these flares was relatively weak. Note that the flare of 0030 UT March 22, 1967, is one of the events listed in table 8. This event is the one for which the largest EUV enhancements observed by 47 satellite experiments has been reported in the literature (Hall and Hinteregger, 1969) . It is also the one on which many of the entries in table 2 are based (the ones with 1's in the column for number of events observed) . The present study suggests that the observed impulsive EUV flux may have been smaller relative to the slow emissions than for most flares. The main influence this may have had on the results in table 2 would be to reduce the Rf, $f , and R$ values for certain wavelengths if these wavelengths encounter greater absorption in the solar atmosphere compared to other wavelengths. For example, the radiation at wavelengths within the hydrogen continuum probably encounters more absorption than the flare radiation at wavelengths just above this continuum. 5.6.5.C Limb Flares Besides the above studies of the CMD dependence of EUV flashes, a special study of large limb flares was made. Solar Geophysical Data was searched for importance 2 Ha flares located at CMD >_ 80° from October 19 60 through June 19 70, during Boulder SFD observations when the solar zenith angle was <_ 80°. Only sixteen such flares were found and only four (25%) were accompanied by SFD's; whereas, from table 6, we would expect eight to have SFD's. Furthermore, the four SFD's were quite small. The SFD data were checked at the time of the other twelve events. Large absorption (SWF) was evident, which implies a large soft X-ray enhancement; but there was no evidence of a frequency deviation or EUV flash. These results also support the idea that the impulsive EUV emission of flares is relatively weak for limb flares; unfortunately the number of events involved is undesirably small, which may be partially a consequence of a tendency to underrate limb flares (Sawyer, 1967) . 5.6.6 Bright Impulsive Ha Kernels Several high quality photographs of impulsive bright kernels observed at the center of Ha have appeared in the literature. Some particularly good ones are shown by Tallant (1970, fig. 1) with a video scan across the main kernel for the flares of 1859 and 2003 UT , March 20, 1966 and 1510 UT, March 21, 1966. The SFD's for the March 20th events are shown in figure 39 in appendix A. Films of the Ha flare patrol show that the kernel measured by the video scan at 1859 UT flashed starting with the SFD spike at 1858. Similarly the kernel selected for the video scan at 2003 UT flashed at the same time as the SFD. The size of these kernels de- pends upon the intensity level used to define its boundary; the brightest part appears 3 to be less than 10 arc-sec in diameter (< 7x10 km) . Vorpahl and Zirin (1970) have reported that a hard X-ray pulse at about 2358 UT , September 11, 1968, was associated with the impulsive formation of a brilliant Ha kernel in the Ha flare. Their hard X-ray spike and brilliant Ha kernel were concurrent with a small spike SFD. The kernel for the flare of October 24, 1966, shown in figure 18 was only several arc sec. in diameter. The cross section (parallel to the photosphere) of a EUV source 17 2 region is therefore estimated to be about 3x10 cm . Large flares appear to consist of a number of such kernels . These kernels do not always appear to be round, some- times they look like very small arches low down near the edge of spots. 48 Broadband Ha observations or observations in lines weaker than Ha (see sec. 5.5) appear to me to be more suitable for studying the optical kernels of flares than the center of Ha, because only the brightest portion of flares appear and because the sunspots are more clearly visible so that one can better locate the position of the flare kernels with respect to the sunspots. The kernels are usually located at the edge of spots (see also sec. 5.4, 5.5, and 5.6.4) , but no quantitative study of their location has been made. Their relation to Rust's (1969) magnetic anomalies has not been determined. Such anamolies would mean the existence of magnetic flux tubes having anomalously low rates of convergence toward the photosphere compared to surrounding tubes that converge into the main spot; hence, these tubes may permit dumping of energetic electrons into low dense regions, namely the hot kernels, more so than most magnetic tubes through the flare region. In section 4.5, the hypothesis was suggested that flares with quasi-periodicity in the time structure of hard X-ray, EUV and centimeter-wavelength radio emissions were cases where a series of flare kernels at different locations occurred in a nearly periodic time sequence. Janssens and White (1970) have published filtergrams o , o spanning Ha ± 4 . 1A with a scanning rate of 0.29 5 A and about 2 sec in time between frames for the quasi-periodic flare of August 8, 1968. Their selected published frames are not at precisely the right time to determine whether new spatially-separate emission regions are appearing in time with each major spike in the hard X-ray, EUV, or centimeter-wavelength radio emission with the possible exception of the one at o 1816.6 UT , which is accompanied by several new small emission regions at Ha + 3A shown in their figure 1, frame 18. Observations of Ha for the flare of 1620 UT June 11, 1969, which were made at 5 sec intervals at Sacramento Peak Observatory and provided for study through the courtesy of Mr. Howard Demastus, were studied to check our hypothesis of Ha kernels separated in time of occurrence and spatial location being associated with the EUV fine structure. This event was the highly structured one discussed in section 4.5 and shown in figure 10. The pulses in this event are quite distinct and separated enough in time compared to Ha photographic observations that if the above hypothesis were correct, we would expect to see small bright Ha kernels popping off one at a time at different locations at the time of each EUV pulse. This was not the case; hence, this hypothesis is inadequate for explaining some quasi-periodic EUV bursts . The Ha observations included an off -band sweep so that the bright Ha cores were easily seen. Several small cores were present but were not very impulsive. The impulsive portion of the Ha flare spread along a tear-drop shaped arch where the pointed tail of the tear-drop pointed into the major sunspot of the flare region. The impulsive flare emission started at the blunt end of the "tear-drop" arch, then shot along one side of it toward the spot, then along the other side, etc. Hence, there were impulsive Ha emissions; but separate bright cores could not be identified with each EUV pulse. 49 In summary, impulsive Ha emissions are observed in association with EUV flashes. Sometimes these Ha emissions appear as explosive expansions of the Ha flare area (the explosive phase, see sec. 5.6.4) sometimes as impulsive stationary bright kernels, and sometimes as bright rapidly-spreading arch structures. Apparently no one simple spatial structure will suffice to explain the impulsive Ha emissions associated with EUV flashes. Perhaps a reasonably small set of basic types of structures will suffice. Further study of the impulsive optical-emission regions associated with EUV flashes or hard X-ray bursts should be made, including classifying the types of source regions and quantitatively studying the relationship between each type and the sunspots and magnetic field of the flaring active region. 5.7 Relation to Surges, Sprays and Eruptive Prominences A surge is a stream of chromospheric gas that shoots upward from an active region along the magnetic field lines, slows to a peak extension, dims, and may fall back along the field lines. A spray shoots out more rapidly, seemingly breaking the bonds of the active region magnetic fields and escaping off into space. Eruptive prominences involve pre-existing prominences that start ascending slowly, usually in arch form, and then accelerate to high velocities at a later phase some ten to twenty minutes later. Sprays reach velocities of 1000 km sec in about 3 minutes while eruptive prominences do so in about 30 minutes (Valnicek, 19 64) . These three phenomena, surges, sprays and eruptive prominences, appear as bright emissions when seen on the limb and sometimes as dark material when observed in Ha on the solar disk. The sprays and surges studied were those tabulated by Smith (1968) , Zirin and Werner (1967) , and Zirin (1969) . Table 9 shows the association of SFD's with surges and sprays when the latter events were observed during daylight at Boulder and SFD observations were being made. An SFD was assumed to be associated if its start time or end time was within 10 minutes of the spray or surge start-of-ejection time. Table 9. Association of SFD's with Surges and Sprays Number of Surges Number of Sprays 3 13 Events Accompanied by SFD's 37 17 Events Without SFD's 40 30 Total Number of Events There appears to be little relationship between surges and SFD's. When the events are associated, the SFD's were very small; and the timing of the events are in poor agreement with a tendency for the surge to follow the SFD. These results imply there o is little or no 1-1030A radiation enhancement associated with the surges studied -2 -2 -1 compared to 10 ergs cm sec above the earth's atmosphere. If one looks at the Ha sun, very small surges are rather common. The above comparison does not take into account how large the surges are ; unfortunately, surges are not routinely classified by size. The above study is believed to be dominated by small surges. 50 For all the SFD's for which the author has studied the Ha flare patrol films, few were accompanied by surges, and in those cases the surge occurred after the SFD. The association between sprays and SFD's is fair according to table 9, i.e., 43% of sprays are accompanied by SFD's. Furthermore, the spray start-of-e jection time agrees with the start time of the SFD to within less than one minute on the average. Smith (19 68) points out that some sprays are related to the explosive-phase of flares. Smith further argues that sprays and the explosive-phase are not always part of the same phenomena, i.e., that some explosive flares are not accompanied by sprays and vice versa. In section 5.6.4, we have seen that SFD's, or flashes of EUV emission, are closely connected with the explosive phase of Ha flares. In sections 5.1 and 5.6.5, we reported that EUV bursts from limb flares are anomalously low in intensity or that SFD's are relatively insensitive to limb flares. Conversely, sprays are most easily observed at the limb. Therefore, the association of sprays and EUV bursts is probably higher than indicated above, i.e., the opposing "seeing" bias of sprays and SFD's probably cause the apparent association to be weaker than their real association. Eruptive prominences are very weakly associated with SFD's. Again, eruptive prominences are mainly limb events and SFD's are relatively insensitive to limb flares. Furthermore, prominence eruptions are relatively slow events compared to the time scales of SFD's, and SFD's are quite insensitive to slow events. For disk events, e.g., the filament ejection of 1830 UT , September 3, 1962 (Nolan et al. p 20, 1970) , the corresponding SFD is concurrent with the associated impulsive Ha flare and appears less directly related to the filament ejection. The slowly rising loop structures observed after the flash phase of some large flares are also weakly associated with SFD's. Reeves et al. (1970) have reported on 0S0-6 EUV observations of a limb surge on September 15, 1969, which shows that with the spatial resolution and sensitivity to small EUV fluxes presently achievable in satellite experiments, EUV surges can be o observed. However, considering the slowness of surges and the minimum 10-10 30A flux -2 -2-1 ° enhancement detectable by SFD observations (-10 ergs cm sec at 1 AU) , the 10-10 30A emission from surges is not observed in SFD data. 5.8 Microwave Bursts Strauss et al . (1969), Basu and Chowdhurry (1968), and Chan and Villard (1963), denoted respectively by S, BC and CV below, have studied the relationship between tabulated SFD and microwave radio bursts; their results are as follows: (1) Radio bursts at frequencies below 500 MHz and gradual rise and fall events at higher frequencies are poorly correlated with SFD's (CV) . (2) Impulsive radio bursts of frequencies above 500 MHz are fairly well correlated in occurrence with SFD's. More than 30% of SFD's are accompanied by radio bursts (S, BC, and CV) and about 46% of impulsive 2800 MHz bursts are accompanied by SFD's (CV) . (3) The correlation of occurrence of SFD's and microwave bursts increases with the peak amplitude of the radio burst (BC, CV, S). (4) The correlation of occurrence increases with increasing frequency of the radio burst observation for radio frequencies above 600 MHz to a fairly flat peak at about 5 GHz (S) . (5) The start times of SFD's and impulsive radio bursts at frequencies greater than 600 MHz are correlated; most events having a difference in start times less than 2 min (BC, S) . 51 (6) Solar radio bursts at 4995 MHz (in radio flux units) and SFD's (in Hz) follow practically the same inverse square intensity distribution law (S) . (7) The correlation of occurrence of SFD's and microwave bursts depends on the spectrum of the radio burst. The correlation is small for radio bursts with spectra that decrease with frequency above 600 MHz; fairly large for radio bursts with spectra that increase with frequency in the 600-8800 MHz range, and largest for radio bursts with spectra that peak in the 600-8800 MHz range (S) . o Studies of the detailed time dependence of the 10-1030A flux enhancement deduced from SFD's in comparison to that of microwave bursts (e.g. Donnelly, 1968c) often (but not always) exhibit similar (but not identical) fine time structures during the SFD or during the early portion of the radio burst. The agreement in fine structure generally improves with increasing frequency of the radio burst observation and the transition from dissimilar to similar time structure usually occurs in the 2-5 GHz range, depending on the particular event involved. Richards (1970) studied the time dependence of EUV bursts observed by Hall and Hinteregger (1969) in comparison with microwave bursts. He found good agreement in start times and times for the first peak of EUV flares and centimeter wavelength radio bursts, but later peaks did not match up well. He also found that after the initial peak, the EUV bursts generally exhibited smoother structure and longer decay times than the radio bursts. More EUV measurements of flares with high time resolution should be made and studied quantitatively in comparison to radio bursts. o The peak intensity of the 10-1030A flux enhancements deduced from SFD data for the same events studied in comparison to hard X-ray bursts by Kane and Donnelly (19 70) were also studied in comparison to the associated microwave bursts. The average -2 -1 ratio of microwave flux (ergs cm sec. ) in the 3-10 cm wavelength range to the ° —8 10-1030A flux enhancement deduced from SFD's is about 2x10 at the time of the impul- sive peak of the 10-1030A enhancement (see table 2) . The regression diagrams in figures 25-30 exhibit an amount of scatter similar to that of the corresponding diagrams of hard X-ray peak flux versus peak radio flux and, for frequencies >_ 2695 MHz, similar o to that of the diagram of hard X-ray flux versus 10-1030A flux enhancement (Kane and Donnelly, 19 70) . The 606 MHz and 1415 MHz peak fluxes are poorly correlated with o the 10-1030A flux enhancements; the two types of data are poorly fit by a linear relation. The correlation coefficient jumps to about . 8 as the radio burst frequency increases to 269 5 MHz and then levels out and remains high as the frequency increases to 10 GHz. Similarly, a linear relation fits the peak radio flux and o 10-1030A flux enhancement fairly well at frequencies of 269 5 MHz and higher. There is slight evidence of a dependence of the ratio of radio flux to 10-10 30A flux enhancement on the central meridean distance (CMD) of the associated Ha flare, i.e. a relatively large ratio for limb flares compared to disk flares. However, any such CMD dependence is far less evident than the CMD dependence of the ratio of the hard o X-ray flux to 10-1030A flux found by Kane and Donnelly (1970) . There are only 46 events involved in figure 27. Radio flux measurements are available for far more events than hard X-ray measurements are available. Further study should be made of o the relation between microwave radio bursts and 10-10 30A flux enhancements as a function of radio frequency, radio spectra, and CMD of the associated Ha flare. 52 4-1 1 1 1 1 i Iffil 1 I I 1 INI 1 1 1 1 1 III 1 II 1 1 1 III I l l i 1 - z © - - © ®© - 3_ ® <§> — = - _i — UJ © _ • - © • ®* © © • - - © © - © © © • © NI = X — ^ • © ~ If) I 1 MM 1 1 1 1 Inn ii I i lllll 1 1 1 1 o -^ C> fo o <-1 to rV VI « co b tx O V) to X o> to • \_ ^ -w CD CO S a co E c. S CO a> « E to a CD o c o to a; E co CO c H N 3 jj ^ T->i 5: «k 0< o ro O ,_ZH 2_"JM 32.01 xn IJ OJPDy ^D8d ±1 1 1 1 1 1 hum i i inn II 1 III 1 1 1 1 1 hum i i = — 1 © i — - © - ®- © © © © — = © _i UJ - • - © • ® © © e © • - - © - © # © © © © - M — _ X - ^ © ' - CO — o — - CO - 1 1 II 1 lllll 1 1 1 Li i i o i CD o to r-H CVJ rii E a o CO &4 co CT CO ^~ 3 • O) to "M Ss S ^_ CO CO c » E o c: a c to -s; to K •ti CO E CO H UJ 3 t\J t-l 1 •* * V, — D U_ CO c> o< Co o r* rO CO O ft. ,.ZH g.LUM 22 .0I Xn|j OipDy >|D8d 53 o «q; +IH L E ® e 1 III 1 1 1 Mill 1 1 1 III 1 1 1 1 1 II 1 1 i i i i = - ® © - — ® © © — - LU • • (t) ® © • © - - © • ® © © © © - - X - ^ _ LO cn en llll II ! 1 Mill © 1 1 © 1 • 1 1 1 © ©1 1 Inn i i i i In M i i i i N5 Q -* r-1 o 1 a> Q w t\j a; 'E a o a. (/) 0> 03 s • -«— S E c cfl a> K O E O £ o> •t» a o tO »C c to K o •ti Cfl .c s c o< "W O a; ro a O ,_Z|H 2 .^M Z2 .0l Xn|j OipDy >|D9d s ■U 1 1 1 1 1 1 UN 1 1 1 1 Mill 1 INI 1 1 1 1 1 Mil 1 1 1 1 1 - - © 1 ' 1 ' — - © - © ■" © © © — © % ®® — - o X 13 ll N X o o _l LJ • • © © ® ffi®© ® © 8 '•• — *fi O _ — a> O © ® z - o C\J lO © • - > < ,cn CO Ml 1 1 1 CM 1 1 III 1 1 1 1 II 1 ® hn i i i i i IIIILI 1 1 1 T JH 2 _w/V\ 22 _OI xn|j OjpDy ^D8d E o a> E a> o c o .c c UJ x o< O ro O O +> CO « +i K "W Cfl co e ^ « S cfl s CO a rQ rC O K Cfl CO C H Q 3 ■ri T-*i to S-, to •<■»■=« F Q CO Ni C3> Ci T-H ■<■» 1 '« Q w *-l fc rV rV « « Cfl Cfl n, CK CO cni V t*i 54 #1111 ®' 1 III 1 1 1 I 1 Mill I I l MINI 1 i nil i i i i - - ® - ® ® ~ ® — — _i — u ® « ' _ - • ® ® ® ® - ® • ® ® — N — — X ® — ^ z o • ® — o N- O ® - III I 1 1 | I i i Inn i l I t 1 III 1 1 1 1 i mi ii i i 1 E o CD CD E CD O c o -£Z c LU X 3 o< o ro O i O o ^ a a, Cl> K » «a e K u O •ti K to CO ~ •ti K e i» ca C$ ft! fe ,_zh 2 .lum 22 .OI * n ld °!PDd ^9d CD CJi +1 1 1 1 1 1 1 III 1 1 I® 1 1 MINI 1 1 1 II MINI 1 III 1 1 1 1 = — 1 © = - © - ® © ~_ © ® ~ © — - © = © • *% ® ® - ®® ® ® «® - N @ % ~ Z X z ^ , — o © ® o 00 00 ® - II 1 1 1 1 1 1 III 1 1 1 1 in I I i i i Lin i i i III 1 1 1 1 i o CO CD C3 t/> ri CM C3 'e *< o tV to a ty cs Cl. CD CO s • CO +» CD CO CB E » E CD CD O K O c O R o ta « _c CO rC c eo K UJ •ri CD E X Cl> D t-i Ll_ £ % o< o O rO oo O a: i a o ,_zh 2 _w/\A 22 .0l xnt-j ojpDy ^Dad CD 55 6. SFD's FOR FAMOUS FLARES Over the past several years, a number of solar flares and their interplanetary and terrestrial effects have been studied by numerous scientists in many different disciplines. In this section, we present previously unpublished SFD observations of some of these famous events. Table 10 lists references to other studies of the same flares. The peak EUV flux enhancement for these events is given in table 4 in section 4.2. All the SFD's in table 10 are very large events. Figure 31 illustrates a type of distortion in the shape of Af(t) for certain transmission frequencies that occurs in such large events, namely, the ratio Af(t, 5.1 MHz)/Af(t, 9.9 MHz) decreases with time during the event. The 5.1 MHz observations involve a near-vertical path while the 8.9, 9.9 and 11.1 MHz observations involve a 1290 km oblique path. This distortion is caused by the height of reflection for 5.1 MHz lowering considerably (10's of km) during the flare because the flare-induced electron-density enhancement in the E and Fl regions is comparable to the preflare electron density. Lowering the height of reflection reduces the SFD's sensitivity to EUV radiation. The distortion is much greater for 4 MHz. The oblique paths are much less distorted, partly because their height of reflection lies above most of the flare-induced ionization enhancement at a height where the percentage increase in ionization is relatively small, and partly because changes in the equivalent vertical incidence frequency minimize the change in height of reflection. As the electron density increases, the height of reflection decreases but the ray-path take-off angle of elevation and equivalent vertical- incidence frequency (f ) increase for these oblique paths reflected in the F2 region of the ionosphere. Increasing f partially compensates for the increasing electron density resulting in a smaller decrease in the height of reflection than for a vertical path having the same frequency as the preflare value of f . The July 8, 19 68 SFD in figure 3 3 is a good example of an SFD with much fine structure. All the SFD's discussed in this section were part of the fine time-structure study of section 4.5. The three SFD records in figure -34b are almost identical even though their propagation paths are different, which shows how self consistent the SFD data are for small changes in transmission frequency or path length. The pre- flare ionosphere near Boulder was relatively quiet for this event, it is not unusual for small SFD's observed on near vertical paths to suffer distortions from ionospheric variations unrelated to flares. Oblique paths are less influenced by local ionospheric irregularities and are therefore less influenced by these distortions. However, the oblique records for the April 21, 1969 SFD in figure 35 suffer a related type of distortion in that the traces are fuzzy. The ionosphere at the midpoint of the Illinois to Boulder path was apparently disturbed while the ionosphere over Boulder was relatively calm, an unusual situation. The fuzziness of the oblique trace is believed to be a consequence of an averaging process of local irregularities near the midpoint of the propagation path for the small bundle of rays received in Boulder. 7. CONCLUSIONS, RECOMMENDATIONS, AND SPECULATIONS Observations of SFD's have provided considerable information about EUV flashes of solar flares. We have learned (sec. 5.1 and 5.8) that EUV flashes are closely related to hard X-ray bursts and 3 cm radio bursts; the quantitive relation between the EUV, radio, and hard X-ray energy fluxes has been determined (to within about 56 UH -P U 01 01 g UH Id w Ul 0) 01 rH Ul 0, a) •H a) Tj Oi a ■P 73 in V -p u id ai •H .c O ■J in in m p o a ■P Cm 111 Ul n a) 3 X 4-> O Q in 01 D •H 3 •O O . C r- T3 ~ id oi C r» f-H • id ko m OJ ■— ,-. ai 4-> rH a\ 0) rH — . •H rH >, ID rH — o J3 ■H Hu m r- 3 C TJ ai id rH 4J 3 ■H Cu > rH H OJ 01 C o ^-, rH 01 Ol rH * M n * r^ Ol rH rH OJ ^^ H m ~~. 01 kO * *-- *— ' UH 01 id m Ol rH ai r~ OJ \D O 01 VO ^^ rH kO c ■o rH 01 u Ol w ^-. Ol 0) c H ■a rH ■rH o rH Ul T3 «-* c -a ^ H c r* ^ Ul E C id c rH ai c J3 id C id c (U r-* e rH id O 01 *-* OJ rH o •rH w rH X •rH ^ 01 o ^-> Ul t-* 0] id I « Ol V) r~ Ol Ul Eh 01 ■a 01 kO c 01 kD c H T3 c p u "O 01 ra i-4 01 id •a -^ c 111 0) c rH X w r-^ X s id e \Q id ' i ^ 1 id T3 JZ ■p •a oi ■0 Ol C id o ■p c •» c M c c rH ■H M -H id M >-\ 111 rH OJ c g ■p « ■rH Eh o XI E ja -H £ Ul >H r* u TJ X o V HJ O OJ "0 M >0 Ol c C o c c M ■H > c id c rH •H n) -iH •H id « Ul id U id w J Eh « J Eh rH m 1*1 a OJ OJ M -H 3 Ol HH C >, r~- HH .C H H 01 H OH H •» S 0) — in c/i m c rn - c e * ■ 01 •P -H 0) -P ~ u -H Q OJ h ho 3 J3 3 r^ Oi Dlt) H 0> Qj Oi H OJ C ■H 0) rH Ik rH Id Ul t» k- 57 At 1Hz . ^' #*•» ■if rt «***r W** J * V*? i niM nrn i'i i- Mi t^ * w^aac a V v w A. w, -« 1920 UT 1922 f f 5 ^ 1924 III. I MHz 19.9 MHz 8.9 MHz 5.1 MHz 4.0 MHz 1926 Figure 31. Boulder SFD observations for the flare of 1924 UT May 21, 1967, 58 928 UT Figure 32. Normalized 10-1030A flux enhancement as a function of time for the flare of 1924 UT May 21, 1967. 59 o< O ro O x o £ < o< 4 — O ro O i O < c o > Q X o E cr \ 5 < 0.2 o E 1712 UT 1712 UT Figure S3. SFD and 10-1030A flux enhancement as a function of time for the flare of 1710 UT July 8 3 1968. 60 March 12, 1969 * 5 Hz i > a> Q o c a> cr 13.0MHz WWI Havana, III. to Boulder, Colo. 1290km Interference Trace Lost due to High Absorption 1 I.I MHz WWI Havana, III to Boulder, Colo. 1290 km Trace Lost due to High Absorption 1735 1 740 Time 1745 UT Figure 34. Boulder SFD observations for the flare of 174Q UT March 12* 1969 (a) Oblique-Path Observations WD ( 10 -10 20 A ^t) for this event is presented in Molntosh and Donnelly (1970), 61 March 12, 1969 5 Hz 1 5.1MHz KKE42 Sunset to Boulder, Colo. 23 km ' SFD j% l y_ Trace Lost due to High Absorption Station Identification Break Ground Wave I I I I > Q o c E Trace Lost due to High Absorption 8.9 MHz WWI Havana, III. to Boulder, Colo 1290 km *-*■ 5 I MHz KKE42 Sunset to Boulder, Colo. 23 km SFD /"V** 1 Trace Lost due to High Absorption Reflection from an Airplane \A+ \ Ground Wave 5.05 MHz KC2XIB Keenesburg to Boulder, Colo. 58 km s ■SFD 1 Trace Lost due to High Absorption 50 MHz WWV Ft Collins to Boulder, Colo 71 km SFD | Trace Lost due to High Absorption Ground Wove , ¥ , 2005 2007 2009 Time 2011 2013 U.T. Figure 35. Boulder SgD observations for the flare of 2009 LIT April 21, 1969 A$( 10-1030 A, t) for this event is discussed in section 5. 6.6 . 63 the absolute accuracy of the 10-10 30A flux enhancements deduced from SFD data, namely a factor of 4) . Large EUV flashes are closely related in time and occurrence with white-light flare emission and the relative energy fluxes of these emissions are roughly comparable (sec. 5.4) . Figure 14 summarizes our present picture of the time relation of EUV flashes with respect to other flare radiations, which is essentially that of DeJager (1964) . o Some information of the spectrum of the 10-10 30A flashes has been learned from comparisons of SFD data with satellite EUV measurements. These results (sec. 3) indicate that line emission from the more abundant solar constituents (H, He, C and 0) o contribute significantly to the 10-10 30A flashes; but we do not know yet whether continuum emission other than radiative recombination continua contributes signifi- es o cantly to 10-10 30A flashes. Further information on the spectrum of the 10-1030A flash will require more satellite observations with high wavelength resolution. Data on SFD have shown that the relative intensity of EUV flashes varies with the central meridian distance of the associated Ha flare location (sec. 5.1, and 5.6.5). Also the EUV flash appears to be associated with small impulsive portions of the Ha flare that are quite bright and usually located near the edge of sunspots (sec. 5.6.6, see also sec. 5.4 and 5.5). Recommendations for further study of SFD's have been made throughout the text; however, further experimental advances of our knowledge of EUV flashes from solar flares will probably depend upon satellite measurements, since any experimental advances will probably require high wavelength and/or spatial resolution as well as high time resolution. Measurements of SFD should, nevertheless, still be quite helpful o into the next solar cycle because their high time resolution of 10-1030A flashes should inexpensively provide useful information which is supplementary to satellite measurements with high wavelength or spatial resolution, since such measurements usually result in low time resolution compared to the fine structure of EUV flashes (see sec. 4.5) . Believing that the interaction of experiment and theory are vital for the vigorous growth of our knowledge, what is probably needed more than refinements of present EUV satellite experiments is more theoretical work to explain present observations and to determine which experiments will be decisive in evaluating alternative theore- tical explanations. Figure 14 is a first step toward a theoretical examination of the complex phenomena of solar flares, namely separating the phenomena into parts. It is quite an insufficient breakdown since it doesn't include the all important particle radiation, but it is probably sufficient for theoretical examination of EUV flashes. Our second step, and a logical extension of the study of the experimental results of SFD's, is to try to develop a theoretical model for the source region of the impulsive EUV emissions. A model for the EUV emissions must include impulsive line emission like those that have been observed (see table 2) . It must explain the following: (1) the observed detailed time agreement between the EUV flash and hard X-ray bursts and their energy flux ratios (sec. 5.1) , (2) the similar time dependence of white light emission with comparable energy flux (sec. 5.4), 64 (3) the similar time dependence of microwave flashes, their energy flux ratios, and dependence on microwave spectrum (sec. 5.8) , (4) the EUV flash dependence on the central meridian distance of the flare (sec. 5.1, 5.6.5) , and (5) the smallness and location of the associated Ha impulsive kernels, an extensive task: Fortunately, much work has been done in attempts to explain the relation between the hard X-ray burst and centimeter radio burst (e.g., Takakura, 1969). Hence our attempts at a model for the EUV source will build on these works. Four models for the impulsive EUV source region (component A) are proposed for consideration. They are illustrated in figure 36. Detailed arguments for or against these models will be the subject of a later report. Model A has been proposed by Kane and Donnelly (19 70) to explain the relationship between hard X-ray bursts and EUV flashes. In this model, the over-all time dependence of the impulsive burst, including its decay, reflects the time dependence of the unknown mechanism that energizes the particles. Each energetic electron quickly loses its energy in a time short compared to the duration of the impulsive burst. Model B is similar to A except that the mechanism that creates the energetic electrons operates during the rise of the hard X-ray burst and the decay is controlled by the electron loss from the trapping region. The average electron stays at a high energy until appropriately deflected so that it escapes the magnetic bottle and rapidly loses its energy in the dense chromosphere. Model C employs a continuum emission mechanism for the EUV emission. Bhatia and Tandon (1970) have suggested synchrotron emission as the cause of the EUV emission. This does not necessarily conflict with Hall and Hinteregger 's (1969) observations of line emission contributing to the EUV flash because this continuum emission would excite the underlying or surrounding chromosphere, which would subsequently emit EUV line emission. Model D is an attempt to connect with Nakagawa and Hyder's (1969) flare model wherein inf ailing material sets up a shock wave that hits the lower chromosphere. I favor model A but all four models should be considered further. An explanation of the EUV emission is not too important a goal in itself; but, if only one suitable explanation can be found, then in effect, we will have better defined the character- istics of the energetic particles involved in this portion of the flare. This in turn better defines what the particle acceleration mechanisms within the flare are required to do. The physical processes involved in accelerating particles and the mechanism that triggers the acceleration are major goals of solar flare research. The EUV flash should provide important restraints on the physical processes involved in the impulsive component of flares because the EUV energy flux is so large, at least with respect to hard X-ray or microwave radio emission (see table 2) . However, the total energy radiated in the EUV flash is not too large compared to 32 the 10 ergs estimated for the total energy of the largest flares. Assuming 10 -2-1 ° ergs cm sec at 1 AU for the 10-10 30A wavelength range for the largest EUV flashes (see table 4) , assuming a 10 sec duration (see fig. 9) , and assuming uniform EUV 65 66 emission over a hemisphere, the total energy radiated in the 10-1030A flash is only 31 32 £ 10 ergs. However, the estimates of 10 ergs for the total flare energy probably relate primarily to the slow component of flares; and the energy of the EUV emission may be a major portion of the impulsive component of flares. 8. ACKNOWLEDGEMENTS I am grateful to Dr. Kenneth Davies of NOAA , who is the main person responsible for ten years of good SFD measurements being made at Boulder. I would like to thank Mr. John E. Jones, who improved the instrumentation for Boulder SFD measurements, which resulted in particularly high quality SFD measurements from late 1966 through 1968. I am also thankful to Mr. Dennis Anderson, Mr. Henry Mai, and Mr. Dale Springer, who helped with the data analysis during the last three years of SFD observations. I am (wholeheartedly) thankful for the financial support for part of this work, par- ticularly that reported in sections 5.5, 5.6.5, and 5.6.6, from the NASA Marshall Space Flight Center under Government Order No. H-42710-A. Finally, I would again like to thank Dr. L. A. Hall of Air Force Cambridge Research Laboratories, who has provided unpublished information of his EUV measurement from OSO-3 and rocket flights, which have been very important in furthering our knowledge of SFD's. 67 9. REFERENCES A gy / V., D. M. Baker, and R. M. Jones (1965), Studies of solar flare effects and other ionospheric disturbances with a high frequency Doppler technique, NBS Tech. Note No. 306 (U.S. Government Printing Office, Washington, D. C). Angle, K. L. (1968) , Characteristics of the explosive phase of flares, Astron. J. 73 , S53. Angle, K. L. (1969), Ionospheric effects of solar flares, M. S. Thesis, Dept. of Meteorology, University of California, Los Angeles. Baker, D. M. (1965) , An atlas of solar flare effects in the ionosphere observed with a high-frequency Doppler technique September 19 60-December 19 62, NBS Tech. Note No. 326 (U.S. Government Printing Office, Washington, D. C). Baker, D. M. , N. Chang, K. Davies, R. F. Donnelly and J. E. Jones (1968) , A review of some ionospheric studies based on a high-frequency Doppler technique, ESSA Tech. Rept. ERL 78-SDL 1 (U.S. Government Printing Office, Washington, D. C). Basu, S., and S. R. Chowdhury (1969) , Correlation of sudden frequency deviations with solar microwave bursts, J. Geophys . Res. 74_, 4175-4177. Bennett, J. A. (1967) , The calculation of Doppler shifts due to a changing ionosphere, J. Atmosph. Terr. Phys . 29_, 887-891. Bhatia, V. B. and J. N. Tandon (1970), Extreme ultraviolet radiation from solar flares, Sci. Rept. No. PL-480-3, Dept. of Physics and Astrophys., University of Delhi, Delhi, India. Chan, K. L. and O. G. Villard Jr. (1963) , Sudden frequency deviations induced by solar flares, J. Geophys. Res. 6_8, 3197-3224. Davies, K. (1965), Ionospheric Radio Propagation, NBS Monograph 80 (U.S. Government Printing Office, Washington, D. C); (1966) (Dover Publications Inc., New York); (1966) (Corona Press, Tokyo, Japan, in Japanese). Davies, K., and R. F. Donnelly (1966), An ionospheric phenomenon associated with explosive solar flares, J. Geophys. Res. Tl' 2843-2845. DeJager, C. (1964), Solar ultraviolet and X-ray radiation, Research in Geophysics, 1, Sun, Upper, Atmosphere, and Space, ed. H. Odishaw, (MIT Press, Cambridge, Mass.) 1-42. Dodson, H. W., and E. R. Hedeman (1964), Problems o-f differentiation of flares with respect to geophysical effects, Planet. Space Sci. 1_2, 393-418. Dodson, H. W. , and E. R. Hedeman (1970), Major Ha flares in Centers of Activity with very small or no spots, Solar Phys. L3, 401-419. Donnelly, R. F. (1967), An investigation of sudden frequency deviations due to the immediate ionospheric effects of solar flares, ESSA Tech. Rept. IER 19-ITSA 19 (U.S. Government Printing Office, Washington, D. C). Donnelly, R. F. (1968a) , An analysis of sudden ionospheric disturbances associated with the proton flare of 1522 UT, August 28, 1966, ESSA Tech. Rept. ERL 92-SDL 6 (U.S. Government Printing Office, Washington, D. C.) . Donnelly, R. F. (1968b) , An analysis of sudden ionospheric disturbances associated with the proton flare of 00 26 UT, July 7, 19 66, ESSA Tech. Rept. ERL 86-SDL 3 (U.S. Government Printing Office, Washington, D. C). Donnelly, R. F. (1968c) , The X-ray and extreme ultraviolet radiation of the August 28, 1966 proton flare as deduced from sudden ionospheric disturbance data, Solar Phys. 5, 123-126. 68 Donnelly, R. F. (1968d), Early detection of a solar flare: a study of X-ray, extreme ultraviolet, H-alpha, and solar radio emission from solar flares, ESSA Tech. Rept. ERL 81-SDL 2 (U.S. Government Printing Office, Washington, D. C). Donnelly, R. F. (1969a) , Contribution of X-ray and extreme ultraviolet radiation of solar flares to sudden frequency deviations, ESSA Tech. Rept. ERL 95-SDL 7 (U.S. Government Printing Office, Washington, D. C.) . Donnelly, R. F. (1969b) , Contribution of X-ray and EUV bursts of solar flares to sudden frequency deviations, J. Geophys . Res. 1A_, 1873-1877. Donnelly, R. F. (1969c), Energetic X-ray and extreme-ultraviolet flashes of solar flares, Astrophys. J. 158 , L165-L16 7. o Donnelly, R. F. (1969d), The 10-1030A flux for the flares of May 23, 1967, as deduced from SFD data: preliminary results, Data on solar event of May 23, 1967 and its geophysical effects, Rept. UAG-5, World Data Center A, Boulder, Colo. (U.S. Government Printing Office, Washington, D. C.) . Donnelly, R. F. (1969e) , The X-ray and extreme ultraviolet radiation of the July 7, 1966, proton flare as deduced from SID data, Paper 32, Annals of the IQSY 3_, The Proton Flare Project (The July 1966 Event) , (MIT Press, Cambridge, Mass.) , 204-208. Drake, J. F. (1969), Characteristics of 2-12A solar X-ray flares, University of Iowa Tech. Rept. No. 69-41. Fortini, T., and M. Torelli (1970), The flares of July 6 and 8, 1968, Solar Phys. 11 , 425-433. Friedman, H. (1969), X-ray observations of solar flares, Solar Flares and Space Research ed. C. DeJager and Z. Svestka (North-Holland Publ. Co., Amsterdam) 87-94. Frost, K. J. (1969), Rapid fine structure in a burst of hard solar X-rays observed by OSO-5, Astrophys. J. 158 , L159-L163. Garriott, 0. K., A. V. DaRosa, M. J. Davis, and 0. G. Villard, Jr. (1967), Solar flare effects in the ionosphere, J. Geophys Res. 72_, 6099-6103. Garriott, 0. K., A. V. DaRosa, M. J. Davis, L. S. Wagner, and G. D. Thome (1969) , Enhancement of ionizing radiation during a solar flare, Solar Physics 8, 226-239. Grebenkemper, C. J. (1969), U.S. Naval Res. Lab., Washington, D.C., private communication . Hall, L. A. (1969), Air Force Cambridge Res. Lab., Bedford, Mass., private communication of their EUV flux measurements presented by H. Hinteregger at the IAGA Meeting, Sept. 1969, Madrid," Spain. Hall, L. A., and H. E. Hinteregger (1969) . Solar EUV enhancements associated with flares, Solar Flares and Space Research, ed. C. DeJager and Z. Svestka (North- Holland Publ. Co., Amsterdam) 81-86. Hall, L. A., J. E. Higgins, C. W. Chagnon and H. E. Hinteregger (1969), Solar-cycle variation of extreme ultraviolet radiation, J. Geophys. Res. 7A_, 4181-4183. Hallam, K. L. (1964) , Solar flares in the light of hydrogen Lyman-Alpha, AAS-NASA Symposium on the Physics of Solar Flares, NASA SP-50 , ed. W. N. Hess (U.S. Government Printing Office, Washington, D. C.) 63-64. Heath, D. F. (1969), NASA Goddard Space Flight Center, Greenbelt, Maryland, private communication. Hinteregger, H. E., and L. A. Hall (1969), Solar extreme ultraviolet emissions in the range 260-1300A observed from OSO III, Solar Phys. 6, 175-182. 69 Janssens, T. J., and K. P. White, III (1970) , Description of mass motions and brightenings in a class 2b flare, August 8, 1968, Solar Phys. 11, 299-309. Jones, R. M. (1966), A three-dimensional ray tracing computer program, ESSA Tech. Rept. IER 17-ITSA 17 (U.S. Government Printing Office, Washington, D. C). Kane, S. R. , and K. A. Anderson (1970), Spectral characteristics of impulsive solar flare X-rays >_ 10 keV, submitted to Astrophys. J. Kane, S. R. , and R. F. Donnelly (1970) , Impulsive hard X-ray and ultraviolet emission during solar flares, submitted to Astrophys. J. Lincoln, J. V. (1969) , Compiler, Data on solar event of May 23, 1967 and its geophysical effects, Rept. UAG-5, World Data Center A, Boulder, Colo. (U.S. Government Printing Office, Washington, D. C.) . Malville, J. M. , and E. Tandberg-Hanssen (1969), Magnetic fields in flares and active prominences I: The flares in active region McMath no. 8818, May 21 and 23, 1967, Solar Phys. 6, 278-289. McClinton, A. T., Jr. (1968), White light flares, NRL Space Research Seminar (U. S. Naval Research Lab., Washington, D. C.) , 63. Mcintosh, P. S., and R. F. Donnelly (1970), Relationships among white light flares, magnetic fields, and EUV bursts, to be submitted to Solar Phys. Moreton, G. E. (1964), The association of bremsstrahlung X-rays with explosive flares, AAS-NASA Symposium on the Physics of Solar Flares, NASA SP-50, ed. W. N. Hess (U. S. Government Printing Office, Washington, D. C.) 209-212. Nakagawa, Y., and C. L. Hyder (1969), Response of the transition region to infalling material associated with solar flares, Chromosphere-Corona Transition Region, NCAR High Altitude Observatory, Boulder, Colo., 231-241. Nolan, B., S. Smith and H. Ramsey (1970) , Solar Filtergrams of the Lockheed Observatory, Lockheed Solar Observatory, Burbank, Calif. Ohshio, M. , R. Maeda, and H. Sakagami (1966), Height distribution of local photo- ionization efficiency, Japan Radio Res. Labs. J. 2^3_, 245-577. Parks, G. K. and J. R. Winckler (1969a), Sixteen-second periodic pulsations observed in the correlated microwave and energetic X-ray emission from a solar flare, Astrophys. J. 155 , L117-L120. Parks, G. K. and J. R. Winckler (1969b), 16-second periodic modulations observed in hard solar X-rays, (abstract) Trans. Am. Geophys . Union 50_, paper STS13, 299. Ramsey, H. E., S. F. Smith, and K. L. Angle (1968), High resolution solar photography, Lockheed Missiles and Space Co., Tech. Rept. LMSC-68 1495. Reeves, E. M. , A. K. Dupree, L. Goldberg, M. Huber, R. W. Noyes , W. H. Parkinson, and G. L. Withbroe (19 70) , Observations of active regions of solar flares in the extreme ultraviolet, paper S4-2, presented at The International Symposium on Solar-Terrestrial Physics, Leningrad, U.S.S.R. Richards, D. W. (1970), Sudden frequency deviations, solar extreme ultraviolet bursts and solar radio bursts, unpublished preliminary report Air Force Cambridge Research Laboratories, Bedford, Mass. Richmond, A. D. (1970), Geomagnetic crochets and ionospheric tidal winds, PhD thesis and Sci. Rept. 1 of NSF Grants GA 1453 and GA 18132, Dept. of Meteorology, University of California, Los Angeles. Rust, D. M. (1968) , Chromospheric explosions and satellite sunspots, Structure and Development of Solar Active Regions, ed. K. 0. Kippenheuer (D. Riedel Publ. Co., Dordrecht, Holland), 77-84. 70 Sawyer, C. (1967), Correcting solar-flare data, Astrophys. J. 147 , 1135-52. Smith, E.V.P. (1968), Flare sprays, Mass Motions in Solar Flares and Related Phenomena, Nobel Sympos . 9, ed. Y. Ohman (Wiley Interscience Div. , John Wiley and Sons, Inc., New York) 137-153. Smith, H. J., and E.V.P. Smith (1963), Solar Flares, (MacMillan Co., New York). Solheim, F. (1970) , Ph.D. Candidate, Dept. of Physics and Astrophysics, University of Colorado, Boulder, private communication. Stickland, A. C. (1969) , general editor, Annals of the 1QSY 3, The Proton Flare Project (The July 1966 Event) (MIT Press) . Strauss, F. M. , M. D. Papagiannis , and J. Aarons (1969), The relation of sudden frequency deviations to the spectrum and other characteristics of solar microwave bursts, J. Atmosph. Terr. Phys . 3^, 1241-1249. Svestka, Z., and P. Simon (1969), Proton flare project, 1966, Solar Phys. 10_, 3-59. Takakura, T. (1969), Interpretation of time characteristics of solar X-ray bursts referring to associated microwave bursts, Solar Phys. 6_, 133-150. Tallant, P. E. (1970), A solar flare videometer, Solar Phys. 11, 263-275. Thomas, R. J. (1970), Solar soft X-radiation, Ph.D. Thesis and final rept of ORA Project 05567, Dept. of Astronomy, University of Michigan, Ann Arbor. Valnicek , B. (1964), Bull. Astron. Inst. Czech. 15, 207. Vorpahl, J., and H. Zirin (1970), Identification of the hard X-ray pulse in the flare of September 11-12, 1968, Solar Phys. 11, 285-290. Wright, J. W. (1967), Ionospheric electron-density profiles with continuous gradients and underlying ionization corrections. III. Practical procedures and some instruc- tive examples, Radio Sci 2, (New Series), 1159-1168. Zirin, H. (1969), Some interesting events observed in detail with the Caltech Photoheliograph 24 August 1967-9 September 1969, unpublished. Zirin, H., and S. Werner (1967), Detailed analysis of flares, magnetic fields and activity in the sunspot group of Sept. 13-26, 1963, Solar Phys. 1, 66-100. 71 APPENDIX A o 1. Procedures for Analyzing SFD Data to Deduce the 10-1030A Flux Enhancement: Brief Descriptions Method 1 Assumption of EUV Flash Spectrum (a) Use preflare ionograms , taken near the midpoint of the propagation paths employed to detect the SFD, with Newbern Smith transmission-curves (see Davies, 1965, pp 165-73) to estimate the preflare propagation paths. (b) Compute the electron density as a function of true height from the ionogram data (see Wright, 1967) . (c) Compute the preflare propagation paths in detail using a ray tracing program (see Jones, 1966). o (d) If preflare satellite measurements of the 1-1030A flux as a function of wavelength are available, compute the preflare rate of production of ionization as a function of height using the most recent model atmosphere appropriate for the preflare level of solar activity, e.g. the CIRA 19 65 model atmospheres. Comparing the pre- flare rate of production of ionization with the measured electron density as a function of height, determine the electron loss rates as a function of height. If no such satellite measurements are available either use recent laboratory measurements of the reaction rates for the reactions involved or assume a model for the effective recombination coefficient as a function of height (see Donnelly 1968b, pp 36-8) . This step is a major source of error in estimating o the 10-1030A flux, particularly for the decay stage of any event or for slow events . (e) Assume a spectrum for the 1-10 30A flash based on past satellite measurements (see table 2) . o (f) Assume a total 1-10 30A flux based on simpler methods described below. (g) Assume a first estimate of the radiation time dependence based on the simpler method 3 described below. (h) Compute the rate of production of ionization as a function of height and time using the standard production equations, recent values for absorption and ionization cross sections, and a model atmosphere suitable for the preflare level of solar activity (see Ohshio et al., 1966) . (i) Compute the time-rate-of-change of electron density ^p(h) and the electron density enhancement AN(h) as a function of time and height using the results in (b) , (d) and (h) above. (j) Compute the frequency deviation as a function of time for each of the propagation paths used for SFD observations starting with the results in (c) . Use the ray tracing program by Jones (19 66) , which evalutes the most general formula for frequency deviations (Bennett, 19 6 7) . (k) Based on the intensity of the computed frequency deviations compared to the SFD observations, particularly during the early portion of the event to its peak, o iteratively adjust (repeating h-1) the total 1-1030A flux to obtain a suitable fit between the computed frequency deviations and the observations. 72 (1) Based on the relative time dependence of the computed frequency deviations compared to the observations, iteratively adjust the time dependence of the o 1-10 30A radiation until a suitable fit is achieved. For large flares, a two component spectrum, where one set of wavelengths has a slow time dependence and the other an impulsive time dependence, may be necessary to obtain a satisfactory fit with observations. This method has the advantage of improved accuracy but has the disadvantages of a large amount of analysis time and large expense, mainly from the numerous ray tracing calculations. Method 2 Assumption of Height Dependence of the Rate of Production of Electrons . This method is very similar to method 1 above. It has been described in detail elsewhere (Donnelly, 1968a) and will not be described here. Method 1 has the advantage over Method 2 that a variety of spectra can be assumed and processed simultaneously, o providing an evaluation of the sensitivity of the A$ (1-1030A) results to the spectrum assumption. The computer programs presently used for methods 1 and 2 consist of one or more programs for each step; they have not yet been reorganized for operational use. Methods 3-5 below are all intended for quick, inexpensive, rough estimates of o A (10-10 30A) . They make use of a frequent situation, illustrated in figure 37, where the electron-loss time constant (t = l/2a CC N ) is nearly constant over the 110-200 km eft o height range. The model for aeff (h) used is that used by Donnelly (1969a) . N (h) is the preflare electron density. They also assume that Aq, the flare-induced enhancement of the rate of production of electrons, is negligible below about 100 km, approximately constant from 110 to about 200 km altitude, and then decreases with height fairly rapidly so that 200km I Aq (h) d - I Aq (h) dh = A qxlO cm 100km 100km /•<» -1030A \ IAq(h)dh - cos(X) / n e U)A is the flare radiation enhancement -2 -1 °-l ° in ergs cm sec A at 1 AU, A$(10-1030A) is the net radiation enhancement in ° -2 -1 the 10-1030A range in ergs cm sec at 1 AU, where A* ( 10-1030A) o o /1030A f 1030A A(A) dX, and f\ = / n A((idX e / - e ,»i y io a /A* (10-1030A) . 10A Since Aq(h) and t (h) are assumed approximately constant in the 110-200 km altitude range, then as long as the nonlinear loss term is negligible in the electron continuity equation, namely dAN .AN ... ,. M ,2 _ = Aq - _ - a eff (h) (AN) 73 I'M I I | I I 1 1 1 1 1 1 1 1 i i i N N \ - — \ 1 o ~i \ o — j - E V \ m - n 2E o X: CO ) from Ionogrc UT N 13 One-H 3 MHz, = j 1 PRIL 21, o CO \ \ X— — 'M f — x A_ x ^x- x _ — < i i i ill, 1 1 ::n — CD LlJ CO o> CD > O C\J K O s o Sh +i CO CB Q CO EX E X 63 to 3 6^ (ujh) aprnmv 74 then AN and ||^ will also be nearly constant over that height range. In that case, the frequency deviation formula developed by Agy et al . (19 6 5) applies for a propa- gation path reflected near 200 km or below; «c k dN (h - h ) (4) Af o = FT dt (h r V ' v where k = 8. 06x10 7 Hz 2 cm 3 , f v is the equivalent vertical-incidence frequency, c is the speed of light, h r is the virtual height of reflection, and ho a 100 km in this case. From (1) , (2) , and (4) A* (10-1030A, t) a a p^ h =^ 10 7 /Af(t) + i J Af(u)du\ ( 5 ) o {Af (t) + j I Af (u)du J f v secx K - h — 6 —2 — "? — 1 where a - 2.44x10 Hz km ergs cm sec Note that in figure 37, the height of reflection for 9.9 MHz for the flare on August 8, 1968, is well above the region where x is approximately constant; so (5) would not apply for that event. All the cases shown in figure 37 correspond to large EUV bursts and the nonlinear loss term is not completely negligible. Equation (5) is then used by adding the nonlinear term weighted at the height of reflection of the probing radio wave, namely f secx A4>(10-1030A) a a Z . h - h r o /t t Af (u)du + a __(h ) ( I Af (u)dul o effrllo I (6) (The weighting at the height of reflection is used because much of the observed freq- uency deviation comes from near the height of reflection where the effect of dN/dt is greatly magnified by deviative effects, i.e. by a — multiplying term where u is the phase refractive index) . Method 3 Assumption of Aq and x constant with height, (a) and (b) same as for Method 1. Determine h and f . (c) Assume the effective recombination coefficient as a function of height (see Donnelly, 1968b, pp 36-8) . (d) Compute the electron- loss time constant as a function of height. If x is approxi- mately constant in the 100-200 km, proceed with (e) . If no paths are reflected on the x curve where x - constant, this method does not apply. (e) Determine the true heights of reflection by determining the height at which the electron density corresponds to a plasma frequency equal to f . For a path with h - 200 km, scale Af(t) in detail. ( f ) Compute the solar zenith angle. (g) Evaluate (6) . Method 4 Six Point Method This method is the same as method 3 except it is used to compute only the o maximum value of A4> (10-1030A) rather than its detailed time dependence. Approxi- mation of Af(t) by five straight lines defined by six points. The number of points is of course, arbitrary, but in practice six points have been found to be a good compromise for almost all SFD's. The first point is the start, one is the Af peak, 75 and the sixth is that where Af equals zero at the start of its negative decay stage. Method 5 o Rough Estimate of A* max (10-10 30A) (a) Same as Method 1, and use data for f - 5 MHz. (b) Estimate the solar zenith angle from nomographs, Rx = cosx- -7 3 -1 (c) Estimate x ff in the 110-200 km altitude range assuming a gff = 10 cm sec and computing N for a plasma frequency equal f oE , the E layer critical frequency. (d) Approximate Af(t) by a linear rise to the peak value, measure the corresponding start- to-maximum time (t ) and the peak frequency deviation. (e) Determine R from a nomograph, where R fc in this case is given by (Donnelly, 1969a, p. 41) "V T eff, (7) (1 - e F ) t P (f) Compute A§(10-1030A) as follows: *f(10-1030A) Af , .,,„,. A$ ( io-1030A) = max , where m is the number of hops in the (8) max m R fc R x propagation path and $f( 10-10 30A) = 0.08. Methods 3 and 4 give results for A$max for the same event that agree closely. These results also agree closely with results obtained by methods 1 or 2 for the same event for the few cases when these more precise but cumbersome methods were employed. Method 5 has been found to give only order of magnitude results and to be inapplicable to slow or highly structured events. 2. Sensitivity of Frequency Deviations to Solar Bursts The frequency deviation (Af) produced by bursts of ionizing radiations from solar flares may be expressed by a function F that operates on the radiation enhance- -2 -1 °-l ment A(A,t) (ergs cm sec A at 1 AU) Af = F(X,x,t,f ,Ad>) Hz O) In general, F is a nonlinear operator that is a function of the wavelength A of the ionizing radiation, the solar zenith angle x» time t during the event, the hour of day, season, etc., the equivalent vertical incidence frequency f of the SFD probing radio wave, and other ionospheric disturbances as well as on A(A ,t) . In effect, every time method 1 above is applied, F is numerically determined for the particular event studied. For the purposes of qualitatively understanding the characteristics of SFD's as a detector of solar flare radiations, some simplifying but reasonable assumptions will be made below in order to develop analytic expressions that permit insight into the physical processes that influence SFD's. Assuming that the nonlinear electron loss term is small compared to the linear term in (3), then (9) can be written as a linear integral operation as follows: *d Af = / s (A,x,t,f ) A(A,t)dA, (10) J. ' — 1 2 where S is a general sensitivity function (Hz erg cm sec) , and A. is the longest g 1 o wavelength of radiation capable of producing ionization, which is less than 1030A for all major cpnstituents of the upper atmosphere. For impulsive radiations and an electron loss time constant that is approximately constant over most heights of interest, then s (A, X ,t,f ) - sU.f )K R (11) g (see Donnelly, 1969a, pp 40-42), i.e., S is separable into functions of x» t. and 9 76 R = cosx, and (12) " t/T eff A*(t)« e Mil R = 1- - — 1 — —r-r .where * denotes convolution. lx -" t T .^A* t ef f These results are not exactly true in the real ionosphere, however equations (11-13), can be used keeping in mind that the proper S(X,f v ) function should still have some dependence on x and t. In other words, the R and R. terms account for most of the dependence on x and t respectively, but S(A,f v ) will still have some residual dependence on x and t. Equation (10) then becomes: 1030A X)dX (14) sU,f v >Ai|>< Af (f = 5MHz) = R R o v X f (1-1030A) where the subscripts "o" and "f" refer to observed and solar flare, respectively, then from (14) >f(l-1030A) = / 1030A S (A ,f / 1030A (15) (16) Since *f(l-1030A) = 0.08 (sec. 3), the average value of S(X, 5 MHz) over the 1-1030A range should be about 12.5 if A(f> _ were nearly uniform over this wavelength range. For radio waves reflected at heights of 200 km and higher, and for wavelengths X £ 100 A and for most wavelengths in the 796-1030A range, the production of electrons occurs almost entirely below the height of reflection in the nondeviative region; hence , /%h Af = -^— I — dh = ■—— R I An dh = - R In AAfXldA . d 7 ) f c I dt f i X J Therefore, for 10A £ X £ 100A and 796A < X < 1030A, k n (X) X n S(X,f ) = — -S- = b — -^ (18) where b = 1.36x10 Hz erg -1 cm 2 sec A, X is the wavelength in A, n is the ionization efficiency in electron-ion pairs per erg while n eff is the ionization efficiency in electron-ion pairs per photon, and f y is the equivalent vertical-incidence frequency in Hz . The n eff term in (18) deceptively appears simple, particularly for wavelengths in the 796-1030A range. The ionization efficiency for each of the individual consti- tuents of the upper atmosphere are approximately known, but the effective ionization efficiency for the mixture of these constituents in the upper atmosphere is not a simple function of the ionization efficiencies of the constituents, particularly at wavelengths above 796A where the major constituent N 2 is a strong absorber without producing ionization, i.e. n N 2 =0 for X > 796A. The n eff values used in evaluating (18) were obtained from detailed computer calculations like those of Ohshio et al . (1966). The results depend on the particular model atmosphere used and somewhat on the solar zenith angle. o For X < 10A, (18) is inadequate mainly because of the rapid loss of electrons by attachment to form negative ions and by the fact that the free electrons that remain encounter such a high collision frequency because of the dense ambient gas density that they do not interact effectively with the SFD probing radio waves. The value of S fv accounts for this latter factor (Donnelly, 1968b, p. 26). The negative ion problem can be approximately accounted for by multiplying g£ by the factor (1+N /U-r) , where N e is the total electron density and Nj the total negative ion density after equilibrium 77 for the attachment process has been attained (see Donnelly, 1968b, p 39). This means we are considering only radiation bursts that are slow relative to the attach- ment time constant, which is consistent with all observations of solar radiations having time constants > 1 sec. Then (18) becomes I 1 + N e /N-y ^- — 7—1 1„ d9) BU.fJ « f I 1 + N e /N l) Xn eff S fV ** 1 + N /NT e i (20) 1 + N /N. / AqC.X,h)dh For wavelengths A >_ 10A, (S f / ( 1 + N /N . ) ) = 1, and this term is not important. For wavelengths X << 10A, it becomes very small and provides a low wavelength cut off for S(X,f ). Also at A << 10A, the ionization efficiency is approximately 35eV per electron- ion-pair. Equation (20) has been evaluated numerically using the model for N /Nj as a function of height used by Donnelly (1968b, p. 40) . The results depend on the solar zenith angle, i.e. the term, R = cos x does not completely account for the solar zenith angle rdependence. The value of (20) is mainly dependent upon how much of the height curve of the rate-of-production of electrons for radiation at wavelength A lies above about 75 km. For wavelengths in the 100-796A range and for radio waves reflected near 200 km altitude, (18) again becomes inadequate because the height curve of the rate of production of electrons is no longer mainly below the height of reflection. This has two effects, first h_ ) dh, lAq(h)dh ? lAq(h and secondly, Aq(h ) ? 0. The first effecct tends to reduce S(A) while the second tends r i to increase it because of the — amplification effects in the deviative region near the height of reflection where the phase refractive index u ■* 0. To account for this, r the area A under the Aq(h) curve is divided into three parts, A. =J. Aq(h)dh, r Aq(h r ) (h r -h Q ) , where h Q is at the bottom of the Aq(h) curve where Aq(h ) = Aq (h ) , A - A 1 - A 2 . The frequency deviation produced by A- will have the dependence given by (4) , and that produced by A. is given by (17) . ,A - A - A + A 2 h g - h oX b A n - / : — \ HenCS ' S(A,£ ) /•I I £ / (21) -(h-h )/H Assuming Aq(h) = Aq(h ) e r r for h >_ h , where H is the atmospheric scale height at h , and assuming Aq ma = A/(e H^) , where H m is the atmospheric scale height at the height of the Aq peak for a particular A (see Donnelly, 1967, pp 14-18) , then S(A,f ) a b X n ... / h-h -(h-h) (22) Aq(h r ) where Y = tz and h is the virtual height of h (h > h ) . Again, the evaluation of Aq o o o o max , (22) was made with computer calculations to determine n ^ £ , x, and h„ ; the results eff o depend upon the solar zenith angle and the particular model atmosphere used. In practice , method 1 described above is as easy to apply as to evaluate S(X, v) from (18) , (19) and (21) for a particular event and involves far fewer assumptions , approximations, and restrictions . Equations (18) , (19) and (21) were evaluated for a particular case where f = 5 MHz, fl = 200 km, h = 350 km, f = 13 MHz, and x = 60°, with the CIRA 1965 Mean Model Atmosphere. The absorption cross sections and ionization efficiencies used agree well with those used by Ohshio et al., (1966). The results are shown in figure 1 and in more detail in figure 38 together with S(X) results for f = 2 MHz. The drop-off in S(X) for 78 d b * the ( + s \/«) * both curves in figure 38 at A < 3A is caused by thef — ; Iterm. In other words. V + e 7 J the electrons produced by radiation at these wavelengths is produced at low heights in the ionosphere where they are quickly lost to form negative ions and restrained by collisions from interacting with our probing radio waves. Only that portion of the Aq(h) curve for these wavelengths which lies above about 75 km contributes much to o S(A) , and as the wavelength decreases this portion decreases. Above 6A, S(A) again decreases for f = 2 MHz because the portion of Aq(h) that lies above the height of reflection and does not contribute to Af(f = 2 MHz) increases. The f =2 MHz path v v is reflected from the bottom of the E layer where the electron-density height gradient is large, the deviative region is small, and the net — magnification is small. The line of S(A) jumps back up at wavelengths above 30A because the absorption cross section of the major constituent N drops abruptly there (because of its K-shell discontinuity) , which drops the Aq(h) curves back down to a low enough altitude to be seen by Af(f = 2 MHz) Strictly speaking, some similar minor peaks for S(X, f =2 MHz) occur at certain o ^ wavelengths in the 910-10 30A range, but they are of little importance. The line of S(A) for f = 5 MHz is fairly flat from 4A to 100A. Above 100A, the - V p magnification near the height of reflection is the main cause of increasing S(A) . Above o about 700A there is considerable fine structure in S(A) (not shown in figs. 1 and 38 but indicated by the wiggly line) caused by the fine structure in the absorption cross sections of the upper atmosphere constituents, particularly in N 9 and 0_. The line of S(A) in figure 38 has several limitations. It is computed assuming the ionospheric response to flare radiations is linear, which limits its use in the 1/2 - 8A -3 -2 -1 range to flux enhancements < 10 ergs cm sec . Many soft X-ray flares exceed that, which is not a problem to SFD studies, even though it invalidates S(A) at o A < 8A, because very little of an SFD observed on a path reflected in the upper-E or o F regions are caused by radiations at A < 8A (see table 2) and because simultaneous observations on paths reflected off the bottom of the E layer (see fig. 38) provide o a measure of the small frequency deviations caused by the radiations at A < 8A. Also, o one function for R, for the whole 1-1030A range, strictly speaking, only applies for about the first 10 sec of an event when R - 1 at all heights; after that, separate R functions for different height ranges or radiation wavelength ranges would be t o o required. The S(A) results for 300A < A < 800A depend upon the particular preflare electron-density (N (h) ) used, which influences the 1/y magnification. The sharp drop in S(A) at 972. 5A, the wavelength of the H Ly y line, is caused by a large spike in the N_ absorption cross section. Since absorption by N.. does not produce ioniza- tion, radiation at this wavelength produces ionization very inefficiently. The main effect, however, is that the ionization it produces is mainly at heights well above 200 km (Ohshio et al . , 1966) , i.e. above the height of reflection for the case considered. Other peaks and valleys in the N and 0, cross sections cause smaller o Z Z fine structure at 1 > 700A, which is indicated by the wiggly lines in figure 38. rate variations in the preflare electron density, solar zenith angle, or model atmosphere, the peaks and valleys in S(A) caused by the absorption cross-section fine structure also vary as the — magnification regior rises above or extends below the height of reflection. fine structure also vary as the — magnification region varies and the Aq(h) curve 79 -n — i — r TT i i I I I — r 0} Q.o< .Q C ■o-S Q. =3 O D o -g -,1111 I y< > o i_(DVI P t _39s e _luo s6>ia) zh (\) S t*4 80 3. Comments on SFD Observations The above complications in S(X) somewhat exagerate the problems in estimating o the 10-1030A radiation flux enhancement from SFD's because the flare radiation is spread out over many wavelengths (see table 2) and complications at a few isolated wavelengths do not greatly influence SFD's. In this section, various complications in SFD observations will be discussed; but first, to put these problems in proper perspective, the usual detailed consistency in SFD data should be emphasized. Figure 39 shows several SFD's observed one right after the other at about an hour interval. The observations at 5.1 and 5.0 54 MHz were at near vertical incidence but separated by nearly 1500 km. Detailed consistency in these data are evident even for the small fine structure. Similarly the oblique 10 MHz path, whose mid- point lies nearly 2,000 km east of Boulder, shows precisely the same relative shape. In general, SFD data have good consistency. See the SFD data for the March 12, 19 69 event presented in section 6 . The most common complexities in SFD data, particularly for small events (Af < 1Hz) , are caused by small frequency deviations produced by acoustic-gravity waves in the ionosphere that are normally present. These cause fuzziness of the SFD data for long oblique paths and swells in the SFD data for near-vertical paths. To minimize this noise, to simplify the SFD analysis, and to minimize distortions due to flare- induced drops in the height of reflection of near vertical paths (see sec. 6) , one-hop propagation paths with f ^10 MHz with great-circle path lengths in the 500-1500 km range are best. Probably about 800 km is optimum, since the angle between the one-hop E-bottom and one-hop F paths and that between the one-hop F and two-hop F paths are large enough that practical HF antennas can partially suppress the unwanted two-hop F path and the bottom of the E layer path. It is also long enough that the horizontal extent of the path in the F-region is large enough compared to the local ionospheric irregularities that the swells should be smoothed out, yet short enough that the fuzziness shouldn't build up too much. 4. Miscellaneous Unpublished Studies of SFD's A special study of Af (f v > was made primarily by Mr. Dennis L. Anderson of the NOAA Space Disturbances Lab. during the summer of 1968. The goals of the study were to determine the minimum number of transmissions needed for SFD observations, to search for patterns in Af (f ) that might be helpful in simplifying the analysis of SFD data and to evaluate several analysis procedures. A total of 70 SFD's from September 18, 1966 through August, 1967, each observed at Boulder on nine different transmissions (see table 3) , were studied. No simple pattern for Af(f ) was found. In general Af was small and slowly increased with f, up to 3 . 3 MHz. Above fv = 5 MHz, Af decreased with increasing f , 1 V but it wasn't always of the form Af « -s— . The variation in Af(f ) for 3.3 < f < 5 MHz varied greatly from one event to the next. Also, Af (f ) did not closely fit the form of (4) for many events (see Agy et al . , 1965, pp 22-31). This indicates that the Aq(h) assumptions in analysis methods 3-5 above may be inadequate, particularly for f < 5 MHz; however, for f - 5 MHz, that Aq(h) assumption is believed to be sufficient V Q V for 10-1030A flux estimates accurate in absolute flux intensity to within a factor of four. 81 i — i i. 1* i ID <1> O 2 iri X H c , I- 1 t i j in a > < niver rban ear- < [J 3 3 Z ( NJ \ X \ i ^ 1 i ■r-4 i If) 1 L CO K O <*5 O CO j J f ° o -\ o ■» J \ O CD id i en V o «•• lT) CM J o ^ > IT) if C a I \ 1 \ 1 1 M — * 1 K*H CO 1 i t— < J o * 3 ' — o 1 * — *" : 3 — x> O S o ~5 X) c 1) X) — ■ 3 (5 c> > ■ 3 ^ X LO g t- £ 3 « s Si O K 05 +i CO CO S o o -• CD ID a, 1 o CM i X o o — 2 o , o < a> -o - l 3 O CD ' IM 1 X 1 s \ "3" lO O iri J _ . N h^H _! x> -4- 5 _J ^r l-l _I XI = o C h-4 i 3 - >- o O 1 X > 3 5 S o —• $ lO — M X J \ s o 1 1 i # I > J \ I CD CO i CO S « H Pci 0) 3 (35 uoijoiasq ^ouanbajj 82 Additional results showed that Newbern Smith transmission-curve solutions for the propagation paths gave, for all practical purpose, as good of solutions for the ray path take-off angle, the virtual height of reflection (h r ) and the equivalent vertical incidence frequency (f ) as did detailed ray-tracing calculations for about 90% of the ray-path calculations. Also, two-hop propagation that dominated the SFD observations occurred far more frequently for 8.9 MHz and 9.9 MHz WWI , Havana, Illinois, to Boulder, Colorado, than had previously been realized. The minimum number of transmissions needed for SFD observations for accurate o estimation of A$(10-1030A) was not determined. A minimum of three transmission frequencies spaced in frequency for the particular path length and latitude involved to provide one-hop heights of reflection near 150, 180, and 200 km at summer solstice is recommended in order to provide redundancy against equipment failure, to minimize the cut-off effects of Es , to provide some height information, and to minimize the effects on just one path of diurnal and seasonal variations in the preflare electron density . Another short and inconclusive study was made of the association of SFD's with short events of particle interference observed by NRL Explorer 30 soft X-ray measure- o o o ments in the 0.5-3A, 0.5-8A, and 8-20A wavelength bands. Because the Explorer 30 satellite spun, one could easily distinguish the detector response to solar X-rays from its response to X-rays and energetic particles arriving from other directions. While scaling the X-ray flux as a function of time, it was noticed that the non- solar response due to "particle interference" often had an impulsive rise at the time of the SFD, sometimes having a time dependence more similar to the SFD than the solar soft X-rays. Out of one set of 13 SFD's and X-ray flares, two events showed no particle interference, five showed particle interference with a time depen- dence unrelated to that of the SFD, and six exhibited an impulsive time dependence very similar to the SFD and peaking at times in the range from the time of SFD maximum to one minute later. Certainly the number of events studied is too small to support any definite conclusions. Since small patches of particle interference frequently occur, these latter six cases could be simply the coincidental time agreement of unrelated phenomena; but since the timing agreed so closely, these phenomena might be related. For example, the ionization produced by the EUV flash at high altitudes, where the collision frequency for upward traveling electrons is small, may result in some of the freed electrons of moderate energy traveling upward along the magnetic field lines to the altitude of the satellite. Since this curious coincidence was not pertinent to the main goals of our work, we have not pursued it further. Finally, an inconclusive study was made of the relation between large SFD's and major proton events. One main problem was the lack of a well defined list of "major" proton events. About 80% of the major proton flares that occurred during Boulder SFD observations were accompanied by SFD's, most with peak frequency deviations ^ 3 Hz, i.e. large SFD's. Since some so called proton flares were not accompanied by SFD's, this study was never published. Proton flares tend to be large Ha flares and large Ha flares tend to be accompanied by large fluxes of all types of radiation, e.g., radio 83 bursts, hard X-rays, etc., so 80% of the flares being accompanied by SFD's was not considered to be important. However, SFD's are relatively insensitive to limb flares (sec. 5.6.5) and the events that were poorly associated with SFD's included events like the one at 1953 UT September 10, 1961, which was one of the long duration events (studied in sec. 5.6.5.b). Consequently, a more rigorous study of the rela- tion between SFD's and proton emission should be made which considers the CMD depen- dence of SFD's. 84 APPENDIX B UOULOtR SFO OBSERVATIONS OA TE MONTH DAY Y£AR UNIVERSAL TIME TRANSMITTER CALL FREQUENCY LETTERS MH2 PEAK FRE3UENCY OcVIATION 92 1 k 1968 161.3 .5 lb44.0 1648.0 HHI 3 .900 1 4 1968 lbii .6 1657.2 1706.0 HMI 3 .900 1 4 1)68 1719 .3 1720. 3 1725.0 HHI It .100 1 4 1968 1760 .6 1/92.4 1803.0 HHI 11 .100 1 4 H(i» 2238 .0 2239.8 2245.0 HHI 3 .900 1 6 1)68 1959 .9 2001.0 2004.0 HHI 11 .100 1 7 196 8 2123 .2 2121.. l 2127.0 HHI 3 .900 1 7 1 )63 2154 .0 2154.4 2204.0 HHI a .900 1 7 HbH 2242 .4 221.3.2 2252.0 HHI 9 .900 1 e 196* 191b .4 1816.9 1818. HHI n .100 1 e 1968 1913 .8 19m. 2 1915.0 HHI n .100 1 e 1968 2010 .0 2011.7 2018.0 HHI 8 900 1 9 I960 1653 .3 1659.8 1702.0 HHI 11 100 1 9 1968 1838 .3 1839. 8 181.0 .1. 1847.0 HHI 11 .100 1 10 1 168 2110 .3 2111.4 2114.0 HHI 9 .900 1 10 1963 2121 .0 2122.7 2127.0 HHI 9 900 1 10 1963 214= .0 2 1 4 7 . r 2152.0 HHI 9 9U0 1 n 1968 1615 .3 1617. 3 1618.3 1623.0 HHI 8 900 1 Ll 1968 1659 .2 1700.2 1703.2 171 J. HHI 11 100 1 12 1968 1712 .0 1713.0 1717.0 HHI 3 900 1 12 1368 1721 .0 17 2 5.5 1731. J HHI 3 90U 1 12 1968 18 J9 3 13U9.5 1820.0 HHI 11 100 I 13 1968 1847 u 1847.9 1857. HHI 11 100 1 14 1968 1838 7 1839.4 1345.0 HHI 11 100 1 t 4 3 Z-> 25 3p 1 4 2 6 3 3 4 3 1 1 3 1 4 6 » 1 5 45 6 9 4 4 0. 2-. 3 0. 35 Q. 4 z. 3 1 3 a 4 VERY LA^GE, STRUCTUREO FINE STRUCTURE LARiibT ONl-hOP FRl- OlV FINE STRUCTURE fUCH FINt sTRUCluRt 87 DATE MONTH 9AY YEAR UNIVERSAL TIME START MAXIMUM END TRANSMITTER CALL FREQUENCY LETTERS MHZ PEAK FREOUENCY DEVIATION HZ 10 6 1968 10 12 1968 10 16 1968 10 17 1968 10 17 1968 10 17 1968 10 18 1968 10 18 1968 10 19 1968 10 19 1968 10 20 1968 10 20 1968 ID 20 1968 10 20 1968 10 ?0 1968 10 20 1968 10 21 1968 10 21 1968 10 21 1968 10 22 1968 10 10 10 10 10 10 10 11 11 11 11 11 12 12 12 12 22 23 23 23 26 25 29 10 30 11 1 11 11 11 12 11 17 11 23 11 2". 27 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 1968 12 17 1968 12 20 1968 1731. .6 1731. 1735 .8 .3 1736. 2000 .2 2004 2005 .1 3 2012. 2138 .2 2139 .3 211.2. 1526 .6 1527 2 1531.5 2008 .8 2009 .3 2012. 2127 .3 2128 2131 2 211.0. 2025 .H 2026 2 2029. 2132 .2 2131. 1 2H.0. 1813 .3 1811. 1815 2 5 1828. 1850 .t 3 11.26. 11.26. 11.32 2 9 1 11.37. 172b 1 1729. 1730 1731 6 7 9 171.1. 2125 7 2126. 2128 2129. 2 2 7 211.1. 1658 2 1658. 8 1702. 1708 3 1708. 9 1712. 1739 1 171.3 1 1751. 1935 6 191.0 8 191.9. 2000 3 2001 i. 2007. 161.6 i» 161.9 2 1702. 181.5 3 1850 5 1857. 181.7 1.2 l.i. 0.65 0.55 0.35 1.3 0.1.5 0.2 0.35 0.2 0.2 O.i. 0.8 O.i. 0.3 0.1. 0.55 3.0 7.0 0.8 0.1.5 0.8 0.7 0.2 0.3 0.75 0.25 0.25 0.3 0.2 0.3 0.7 0.85 0.35 0.25 0.3 0.25 0.1.5 0.2 0.3 0.3 TIME ACCURACY ABOUT 2MIM LON3 OURATION LONS DURATION NO H-ALPHA FLARE REPORTO NO H-ALPHA FLARE REPORTO 89 JATE MONTH OAY YEAR JNIVtriSAL TIME START MAXIMUM END KAI.SMITTt*. CALL FREQUENCY LETTERS MHZ PtAK FRE3ULNCY DEVIATION HZ 12 22 1968 2033 .1 2036.3 2037. HHI 11 .100 12 2HA-?F rfljt ♦ PtAK FINE STRUCTURE FRti L'tV POSITIVE 3.5MI>* 2 SHARP PEAKS 2 SHARP PEAKS (OMPY PEAK 2 PEAKS LARit, SEVERAL PEAKS FINE STRUCTURE FINE STRUCTURE FINE STRUCTURE 89 DATE MONTH DAY YEAR 3 21 196") 3 21 1969 3 21 1969 3 21 1969 3 21 1969 3 23 1969 3 25 1969 3 25 1969 3 27 1969 UNIVERSAL TIME START MAXIMUM ENO 3 29 1969 3 28 1969 3 29 1969 3 29 1969 3 29 1969 4 2 1969 4 2 1969 4 2 1969 4 6 1969 4 m 1969 4 n 1969 19b3 2049.6 2050.0 2100. MMI 13 000 9 a 1969 1550.0 1551.0 1554. MMI 13 000 9 U 1969 1746.7 1749.9 1757. MMI 11 100 9 12 1963 2050.4 205 3.3 2054.5 2056. MMI 13 000 9 11 1963 1530.5 1530.9 1533. MMI 11 100 9 17 1963 1800.4 1804.8 1810.0 1811. MMI 13 000 9 17 1969 1859.7 1904.4 1909. MMI 8. 900 9 17 1969 2248.4 2251.3 2255. MMI 11 100 9 11 1969 2029.6 2035.5 2037. MMI 11 100 9 21 1969 2058.5 2101.0 2104. MMI 13 000 9 22 1963 14C8.0 1412.0 1418. MMI 13 000 9 27 1969 2125.4 2125. 7 2128. MMI 13 000 10 1 1969 1950.0 1953. b 1956. MMI 13 000 10 3 1963 1713.8 1715.3 1721. MMI 13 000 10 7 196 1 17Z2.i 172 3.9 1726. MMI 13. 000 10 i 196 3 1630.2 lbJO.8 1633. MMI 13 000 10 ia 1963 2223.5 2224.4 2228. MMI 8 900 10 i i 196 3 1632.5 1633.7 1635.3 lb36.3 1641. MMI MMI MMI 13 13 13 000 000 000 0.35 0.3 5.0 2.0 0.8 0.35 U.55 0.4 0.5 1.0 0.2 0.25 0.3 0.65 3.2 2.1 3.1 0.4 0.4 1.1 0.35 0.25 0.4 0.6 0.3 0.35 2.4 0.45 0.5 C.25 0.25 0.25 0.3 0.45 0.25 0.2 0.9 0.25 0.25 C.45 0.35 1.0 0.35 0.25 O.b 0.35 0.45 0.55 0.8 0.45 0.25 0.45 0.35 0.2 C.32 0.30 0.35 0.25 0.23 0.25 0.30 0.45 0.5 0.30 0.40 0.30 5.70 0.40 0.30 0.30 1.30 MUCH FINE STRUCIUR£ MANY SMALL PEAKS ASSOCIATED HITn THE TMRtE LA,19 6 1620 162 4 9 1625 2126 3 213 2131 6 9 2146 2123 6 212". 5 2129 1522 2 1521. 1525 2 3 1532 2012 3 2013 8 2020 1914 1916 1923 3 9 1933 1303 t a 197) 191.8.1. 1950.2 2005. HHI a 900 it i 1970 1816. 1821.. 1 1825. HHI 13 000 <• 1) 197J 11.31.1 11.32.2 11.31.. HHI 13 000 it 10 1370 162/ .2 1633. lo3i«.9 1637. HHI 13 000 it 10 1970 1 757.0 1757.2 1758. HHI 13 000 it U 1973 2230.it 223it. 8 221.2. HHI 13 000 it 11 1970 2312.1 2311.. 9 2317. HHI 13 ooo it 12 1970 1703.6 170i«. 2 1706. HH I 13 000 it 12 1973 1712. 1721. 171.2. HHI 13 000 it 13 197) 151.8.3 1552.2 1551.. HMI 13 000 •t 1 i 1970 2221.. 3 2227.1. 2230. HHI 11 100 it 1) 197) 1812.0 1815.2 181.8. HHI 8 300 it 2J 197) 2121.. 1 2127.0 211.5. HHI 13 000 it ?> 1970 1715.5 1717.3 1720. HHI 13 000 it 23 1970 17 3 5.3 173 7.0 1739. HHI 13 000 it 25 197) 0036.it 0037.8 0051.. HHI 8 900 it 2/ 197) 1715.1 1719.8 1721. HHI 13 000 it 2d 197) 2211.2 2211.. 3 2217. HHI 13 000 it 29 1973 20it2.2 roi.it'. g 201.7. HHI 13 000 5 2 197) 2212.0 2212.5 2211.. HHI 9 930 5 It 1970 1815.7 1816.1. 1818. HHI 11 000 5 t 1973 1829.2 1931.. 1937. HHI 13 000 5 6 197) 0029. C030. 0031. HHI 1 1 103 5 6 1973 1230. .Hi 8 900 5 7 1973 1601.2 1603.it 1608. HHI 13. 000 5 7 197) 1609.6 1611.2 1625. HHI 13 000 9 7 1970 ZiZT.i 2227.9 2231.. HHI 1 1. 100 5 7 1970 2329.6 2330.1. 2331. HHl 11 . 100 '. 7 197) 2331.. 1 i33i..9 2338. HHI 1 I . ioo 5 3 197 1 17 10.8 1718.9 1723. HHI 13. 000 9 1970 1951.5 1953.9 1959. HHI 13. 000 5 1 197) 2133. 2 .131.. 2 2135. HHl 13. 000 5 ) 197) 1559.8 1600.6 1608. HH I 13. 000 0.30 C.35 U.50 0.30 u.i.9 0.30 0.1.0 0.55 0.1.0 0.50 0.20 0.20 0.25 C.35 0.25 0.20 0.30 0.20 0.20 0.15 0.20 0.15 0.20 0.30 1.35 3.1.0 0.26 0.80 0.25 0.20 1.60 0.30 0.30 .80 0.60 0.90 0.20 0.20 0.30 0.20 0.20 1.10 0.10 0.30 0.15 0.20 0.80 0.15 0.25 0.25 0.1.5 0.80 0.20 0.20 0.25 0.20 0.20 0.U5 0.20 0.30 1.70 0.35 C.25 i. .6 PEAKED 0URIN5 STATION BREAK SLOH, LOH OtVlATION SLOH, LOh OttUATlOt PlAKEO APKGX. AT 1230UT 93 DATE MONTH DAY YEA* UNIVERSAL TIME START MAXIMUM END TRANSMITTER CALL FREQUENCY LETTERS MHZ PEAK FREQUENCY OEVIATION HZ 5 9 1970 1700.5 1701.3 1705. HHI 13 .000 5 9 1970 20(15. 3 2005.9 2009. HHI 13 .000 5 1J 1970 1420.5 H.22.0 1426. HHI 13 000 5 10 1970 1849.3 185 0.5 1853. HHI 13 .000 5 12 1970 171.3.2 171.1.. o 1745. HHI a .900 5 13 1970 1652.3 1655.9 1708. HHI 13 .000 5 15 1970 1153. 2 11.57.0 1500. HHI 8 .900 5 13 1970 1911.1 1914. 6 1922. HHI 8 900 5 la 1970 2201.. 6 2207.0 2219. HHI 13 000 5 lo 1970 11.