Документ взят из кэша поисковой машины. Адрес оригинального документа : http://srd.sinp.msu.ru/nymmik/models/bazilevskaya_1.pdf Дата изменения: Fri Jan 26 21:10:41 2007 Дата индексирования: Mon Oct 1 19:12:38 2012 Кодировка:

Reply to the Dr. Bazilevskaya refrees report (comments on Memorandum
No 1 Comment Implicitly, the Model states that the proton event with the peak flux f30 = 0.12 proton/(cm2s sr) has, in average, the fluence F30 = 105 proton/cm2. It would be useful to indicate to what extent it is valid. The smoothed Wolf numbers are widely used in the Model. What values (daily, monthly?) were used and what was a period of smoothing? A period of smoothing is mentioned only on the p.29 as equal 12 months (rigorously, the period of smoothing should be odd the obtained value being placed at the middle of it), but what does it mean "smoothed sunspot number on the day of the event generation" (figure caption of Fig. 2.6, p. 14, also discussion of Fig. 4.7, p.31)? The concept of "engineering" event has not physical meaning. It does not work for fundamental problems. However, extraction of a "physical" event is often not a simple task, feasible for users. If using "physical events" gains significant advantage for the Model reliability, the criteria for revealing physical events should be described in the Memorandum. What "indicators" were actually used (p. 27)? In the case one SEP event trails the other one, how the fluence F30 for each event could be calculated? Since the output of the Standard Model is the total proton fluence and maximum flux for the mission duration, the "engineering concept" seems to be good enough. Logarithmic averaging is proper for the analysis of the spectrum shape but not for the analysis of proton flux values as presented in Fig. 2.2. (p. 7) because summation of lg(F) is equivalent to multiplication of F (lg(F1)+lg(F2)= lg(F1*F2)). However, the procedure of logarithmic averaging complies with comparison of METEOR, GOES, and IMP data. I would propose to give "arbitrary units" on the y-axis of Fig. 2.2.

)

2

Reply According to Memorandum p.31 in case of "physical" events, the number of events with F30 106 protons/cm2is equal to the number of events with f301.2 protons/cm2ssr, and the distribution functions of by the proton fluence and peak fluxes are the same. Total number of events used for analysis is 200, so the situation is valid with statistical indefinite about 10%. It need to mention, that in the last versus of the model were used the 13-month smoothed month sunspot numbers and predicted by NOAA month sunspot number: by INTERNET (Eq.1 of Standard) ttp://www.sec.noaa.gov/Data/ In the experimental data analysis (Memorandum Fig.s 2.4 and 2.6) we proposed, that the13-month smoothed month sunspot numbers treat to the 15 day of month and for rest days the sunspot data were interpolated. For all events, used in the model, we used the same physical events, what were selected in the catalogues Bazilevskaya and Sladkova. For events dated late than 1994 in cases of multiple events, we divide the fluxes between several ones, if after the start of the first event the high energy protons flux again abrupt increases and it was not done by the particles, accelerated by power shock of the first event. We stated, that if we brake the fluence of first event at the moment beginning of following event, we lost less than 5% from the true fluence and added not greater than 10% to following event. In case, when the exactness of peak fluxes and fluences measurement results are equal to 30-40%, 5 and 10 % mentioned above we can neglect. In that method we calculated the mean value lg(F) of the lognormal distribution of fluences or peak fluxes for every energy of the SEP events group (I ­ events number)

3

4.

[]

1 Fj = I


i

I

log Fi

,j

After that, as we calculated

F

j

= 10

[F j ]

,the units of these

5a

The statement concerning the energy spectrum form (p. 19) in my opinion is too strong ("any attempt to approximate these data by functions containing an exponent has no perspective"). In particular, Fig. 2.12 is not a convincing argument against approximation of Ellison and Ramaty (1985). The author of this Standard has found that energy spectra for majority of events don't need an exponent for approximation. It is right. However, other researchers may find it fruitful to use an exponent in their analysis of some events. Comment on p. 19 concerning the energy spectrum of the peak proton fluxes should be done while the major definitions are introduced.

mean values are the same, as for the single events. Besides, the result of the analysis Fg.2.2 don't depend on of the units name on the y-axis. Besides, we analyzed all the events separately, the result is the same. If the experimental data set is small and the energy range is small too, we can approximate the data by any function ­ for example - by part of ring. It is very useful argument for justification every theoretical speculation. We must know and accept, that all energy spectra at E>30 are power law by rigidity and in this energy range the energy spectra never are the exponents. If we can these spectra at E>30 approximate by exponent, it mean, that these spectra are badly measured.

5b

OK! This comment is moved to the p. 17


6

The reliability of experimental data is a complicated and controversial point. For the purpose of the Standard, the GOES 7 not corrected data were proven by the author to be most proper. However, it is not right in all cases, and should be considered taking into account spectrum of a given SEP event (Smart D.F., Shea M.A. Comment on the use of GOES solar proton data and spectra in solar proton dose calculations. Rad. Measurements, 30, 327-335, 1999; R. Vainio et al.,Proc. 24th ICRC, v.4,p.131-134).

7

Unfortunately, the present version of the Memorandum is not free from vagueness and rather many inaccuracies, which must not be available in the accepted Standard.

8

9

Fig. 2.7 does not mach expression (2.4, as in Fig. 2.7 is not 1.32. What probability is along the X-axis of Fig. 2.7? ???? It should be stated in the text and figure caption Fig. 2.6 corresponds n 0.1W. Elsewhere in the Model n 0.015 W, which agrees with my estimation (n 0.012 W). It is desirable to comment Fig. 2.6 Since 101. 32 =20 rather than 2 (p. 29) expression (4.3) does not agree with (4.2). It should be 100. 32 , probably. Why consider fluences beginning from 105 cm-2 if approximation is good beginning roughly from 106 cm-2?

In our report on the last COSPAR session C0.1 "Standards in Space Environments for ISO "(E.Daly and W.K.Tobiska) - Mottl and Nymmik "THE ISSUES of RELIABILITY of SOLAR ENERGETIC PROTON FLUX DATA BASES and MODELS" we demonstrated, that the Smart and Shea paper conclusion is erroneous. There are very large variations between results of measurements of different SEP events by different instruments. We have shown, that besides the one GLE event, analyzed by Smart and Shea, rest 12 GLE events in 1989-1991 demonstrated, that uncorrected GOES high-energy data are rather overestimated, than on the contrary. CONCLUSION. If there are large fluctuations, caused by bad exactness of the results of measurements by different instruments, all conclusions about the SEP characteristics must be based on the large set of events!!!!!! The Memorandum is not official document of ISO. It is a working document for help of understanding of the Standard draft, based on the poorly known methods. Most of Standard projects have not any Memorandums. When the discussion is ended, only the Standard is the document, what should be official published. I am agree, that the Memorandum contents many inaccuracies and hope to correct them. As the Eq. 2.4 expressed the probability function (integral) there must be written =0.32 instead of =1.32. (corrected) On the p.13 is written, that the Fig. 2.6 data demonstrate the proportionality (or very close to proportionality) of the SEP frequency (the SEP (F30106) events/year) to the smoothed sunspot number. It's all. The coefficient for the frequency of the Standard (1.35·10-2 ) (F30105 ) in case of events/month is determined on the page 29 of the memorandum. Of cause! I was mistaken. Corrected!

10

11

The basis for this approval (in principle) was done in "Kurt

Victoria, and Nymmik R.A, 1997, The >30 MeV proton fluence size distribution of SEP events, Space Research ( ), v.35(10), 598-609". Besides, if we prolong this function to very small event sizes (may be 101 or 102) we can explain the low-energy component of the cosmic rays (at low SA conditions).
You are quite right! The expressions (2.4) and (4.5) are corrected (I replace =1.32 by =0.32). About (4.4) in the text is written, that it is the differential function, (4.5) of Memorandum.

12

13 14 15

I have failed to reproduce distributions in Figs. 2.7, 2.8, 4.1 - 4.8 using expressions (2.4), (4.4), (4.5). It seems that all expressions concerning the fluence or peak flux size distributions refer to differential values, and all corresponding figures (2.7, 2.8, 4.3-4.8) refer to integral distributions. Figs 4.12 and 4.14. Definition of event size should be added in the figure captions as it is given in the caption of Fig. 4.11. Expression =10A-1 (4.10) does not agree with A=+1. What is correct? Why condition (4.11) was introduced?

In the Fig. 2.12 we used the different argument (x-axis) compared with Fig. 2.12 and 2.14. By my opinion, it is reflected in the Figure captions and at the x-axis. Expression is corrected : =10A-1 These errors were by the computer caused but not noticed.... Very small <0 in nature are caused by the very large lowenergy (E<30 MeV) particle fluxes, what are the particles, accelerated by power shock. In GOES instruments, partially, they are sometimes the radiation belt particles, which cut up during


16 17 18

Fig. 4.17 is wrong. It is the same as Fig. 4.16. Fig. 4.19 is wrong. It is the same as Fig. 4.18. Results presented in Fig. 4.18 do not agree with results obtained using Tables in the Technical Specification for the same and probability.. On the other hand, calculation with Tables gives the results matching those of the web-site program Part 5 is very important. The uncertainties, estimated by the author, seems to be reasonable. Unfortunately, this part is written very complicated, so I could not understand details. Figs 5.1 and 5.10. How curves 2 and 3 were found? Parameter is wrong (p.42). Why expression (5.1) is taken for the relative error determination? Fig. 5.3, 5.7 ­ 5.9. What are Model parameters? Probability?

the large geomagnetic storm. There are only small number of that events, and we are not convinced, that they belong to SEP. Moreover, if the distribution function is used and the value of A generated, there are lot of cases with extremely small negative , what sixes are never measured. Yes they are. All these Fig. are recalculated. Figure 4.18 is recalculated. Fortran program, what was sent to you was tuned on the test of the capacity for work.. The working parameters of the program must be (File 10): DATA MCARL, MVERSUS /10000,100000/. MVERSUS must be not less than 100000!!!!!!! Excuse! Yes, it is. Is it a first attempt to estimate the SEP model exactness. It is only an attempt to rough estimation, aim for discussion about this problem. This point is not included into the Standard Specification. Curves 2 and 3 are found as best approximations for upper and lower statistical limits of the experimental data. For the Fig. 5.1 =0.32, for density of probability functions (used in calculations) =1.32. About (5.1) - it is relative hemi-width. Why no? The model parameters are shown on the Figures. Probabilities are, 0.9, 0.5, 0.1, and 0.01. are 32, 4 and 32. Relative errors on the Fig. 5.3 are caused by statistical errors in distribution function (Fig.5.1). Errors on Fig.5.7-5.9 were taken into consideration the errors in spectral index determination. Fig. 5.4. Experimental data. Lognormal distribution. Results of determination the standard declination of logarithmic mean <> and the lognormal function of declination . (Function Eq. 2.2) ? Excuse me, I don't understand the problem. Why? Fig 6.3 is data for peak flux. The caption at Y-axes was wrong. Corrected. OK. The discussion is paraphrased. Xapsos M.A., Summers G.P., Barth J.L., Stassinopoulos E.G., Burke E.A., Probability model for cumulutive solar proton event fluences, Proceedings of the RADECS-99, session A-27, 1999b. Yes, it is. Corrected. On the Figs 6.16-6.30 the experimental dots are the integral peak fluxes or fluences, calculated from GOES instruments differential measurement channels into different energy thresholds of the integral spectra. . The estimation is paraphrased. Corrected As "" is denoted the event frequency, as "n" ­ event number, probability (p) was replaced to "". In the text are these mistakes corrected. As Memorandum is not official document, if these mistakes are on the Figures, they were not corrected (corrections on the Figures need many time). D in Memorandum is replaced to C. Corrected Corrected

19

20

21

22 23 24 25 26 27a 27b 27c

I failed to understand the last 2 paragraphs on p. 44 and Fig. 5.4. Fig. 5.6. Figure caption is too complicated! Here, the errors are given versus energy, but it is not indicated. Fig. 5.13 seems to be wrong According to the figure captions, data of MSU Model should be similar in Figs. 6.1 and 6.2. Fig. 6.3 seems to be wrong. I did not understand discussion about "rate of fluences" on p. 60. In the text, there is no reference to Fig. 6.23 Paragraph 6.2.2.2. (in the text, mistakenly, 6.2.2.1, p. 72). Where GOES data in more than 20 energy channels are available? Is this just calculation of integral spectra from differential ones and approximation? I did not understand estimation of N in paragraph 6.2.3. (p. 74). There is no caption to Fig. 6.31. Number of SEP events is sometimes denoted as "" and sometimes, as "n", probability is "p" or "". The dates of events should be in the unified form, e.g., day, month, year. It is not always kept (e.g., figure captions of Fig. 2.10 and 2.11). Spectral coefficient is "C" in Technical Specification, but it is "D" in the Memorandum Examples of violation of numeration in the text: p. 67. Part 6.2.2.1. p. 70. Part 6.2.2.1. There are some references in the text not included into the Reference list (e.g. Barth et

28a 28b 29

30

31 32


al., 199, p. 9; Feynman et al., 2002) and vice versa (e.g., Berezhko et al., 2001; Blandford and Ostriker, 1978; etc.). In the reference list, there are items for the year with a, b, c, whereas in the text they are referred simply as year. The References in the list are not unified and sometimes not in alphabetic order