Документ взят из кэша поисковой машины. Адрес оригинального документа : http://srd.sinp.msu.ru/nymmik/models/hilgers.pdf Дата изменения: Fri Jan 26 21:10:44 2007 Дата индексирования: Mon Oct 1 19:11:32 2012 Кодировка:

Reply to the Dr.A.Hilgers refrees report (comments on Memorandum
2.1 However, there is no proof given here why GOES-7 correct and IMP are less good than Meteor or Neutron monitor data.=

)

This problem was discussed in COSPAR 2002 paper "DATABASES FOR SOLAR ENERGETIC PARTICLE MODELS: METHODOLOGICAL UNCERTAINTIES AND TECHNICAL ERRORS" (see http"//srd.sinp.msu.ru/nymmik/) and in COSPAR 2004 paper " THE ISSUES of RELIABILITY of SOLAR ENERGETIC PROTON FLUX DATA BASES and MODELS, and Fig.2.2 of Memorandum. 3 measurements:

Neutron monitor, GOES uncorrected and Meteor data are in good accordance, and IMP-8 differ from them. GOES corrected results are wrong, because the publication Smart D.F. and M.A. Shea, Comment on the use of GOES solar proton data and spectra in solar dose calculations, Radiation Measurements 30 (1999) 327-335 was erroneous (see "The issues...." above. 2.2.1 It must be noted that the choice of grouping events which are very closed is to end-up with a list of uncorrelated events such that Poisson Statistics may apply. What is eventually the definition of an event chosen for the proposed model? The criticism to Feynman et al. work is excessive. It must be noted that the approach of feynman et al. was to neglect the so-called solar minimum period when some TBD years of solar maximum period were included in the mission plan. Their claim is not necessarily incorrect. It is clear that their model do not apply (and is not built for use) when all the mission is planned during solar min. On Figure 2.5 the fluence should be divided by the W-bin size to get = read of possible artefact due to data processing. Idem on Figure 2.6 for the number of events per year.= What is the probability of the vertical axis of figure 2.7 Idem on figure 2.8. The criticism of the use of the log-normal distribution for the data fit in Feynman et al. is excessive. Indeed, Feynman et al. use only the high intensity part for the fit making the results almost independent of the low energy part.

For the grouping events all the regularities, used as a base of MSU model are unfounded. Grouping events have not physical base, they are technical happenings. In the model were use single events, the definition of what were exploited by authors of SEP catalogues, edited by Logatshov (Bazilevskaya et al) In the Memorandum were quoted the Feynmans declared ideas about her model. You comment contradict to the allowances. Feynmans publications. Feynmans model is not correct for Solar active years too, there is not data for Quite Sun years. How can you planned the 3 year missions in interplanetary space, if you can't compare the radiation environment at active, quiet and mixed years for choice the best mission time by Feymans model??????_ Of cause. They are divided by the time interval (years), with solar activity size in the common intervals. Its elementary! The same comment, as above. Probability = normalized to 1 integral distribution
Feynman et al. use only the high intensity part for the fit making the results almost independent of the low energy part.!!!! Ten years ago I repeated all Feynmans calculations. I wonder, how anybody can calculate the parameters of lognormal distribution without data of low-intensity part of distribution. Yes, you can use the high intensity part of distribution, but there is no lognormal distribution parameters without its left side!

2.2.2

2.2.3

2.2.5

Wonderful approval!
2.2.6 The differential flux rather than the flux should be plotted in figure 2.10 to avoid artefacts. Since the model is strongly depending on the quality of input data which are subject to continuous improvement the data should be made available together

2.3

? Sorry. I cannot find the word "artefact" in largest dictionary of 100 000 entries..... I cannot use you recommendation. It's not seriously. The database is the part of the dissertation of doctor degree of D. Mottl. This database was repeatedly checked but not formed in needed mode. I think, i can do it in November.


with the model.

But! You have all data about experimental data about SEP, you have our model outputs. The reliability of the

model is determined by the accuracy with which they can describe experimental data.
The additional requirement #1 does not seem to be a requirement. The additional requirement #6 seems not to be justified yet by clear correlation with W since improvement of graph 2.5 and 2.6 are requested (cf comments above). One cannot write that the 4 year quiet period does not exist. One cannot write that the use of the log-normal distribution leads to an erroneous over-estimation. It is ok to say that it is not perfect but is not clear wether it over-estimates or not.However, one can write that it provides estimates higher than some other TBD models. 3.5 One cannot write that ESP model neglects the probability of SEP to occur during the quiet part of a solar cycle. However one can write that it does not apply for missions occuring mainly during the quiet part of a solar cycle. The last statement on dependency with new data has been demonstrated for the distribution of peak fluence (section 2.2.5) but not for the model output. A clear and quantitative definition of an event should be given together with a list of these events characteristics. The probability of figures 4.1 and 4.2 should be explicited. Although a discrepancy between data set is noted it is not clear why there is a discrepancy and which data are the best. Density of probability should be used for the correlation. It is not clear why the uncorrected GOES data where used.

Why?

The comments above were erroneous (see above)!

3.2

? Please read carefully the Feynmans et al. papers! See, please the Figures 2.7, 6.16-6.18 and 6.28 and comments to these Figs.

"Only solar active years are considered in this model" Xapsos, 1998b, p.2949. "It is reasonable procedure to consider solar active years" Xapsos et al. 1999a, p. 1482. Missions cannot be divided as Active or quiet Sun periods. Most of missions should be mixed. And models needed for distinguish what periods are better! The model outputs for small probabilities depend immediate on distribution function( see Fig. 6.16-6.18 and 6.28). Besides, what is the peak fluence?

4.

4.1

Once again. Probability = normalized integral distributions of events. There is no problem, what is best. There is problem about large systematical errors in data, measured by IMP-8 instruments. I have not words........to answer on this "recommendation". Why ­ see Fig. 2.2 and quoted above papers. Besides. Explain, please, why by Zwikl ­ GOES data corrections explanation in Internet ­ claim, that corrected data must be less, than measured, but the real corrected data are large (Internet) than uncorrected...... Thank you, for this recommendations! They are not yet funny, but sad..... No comment. Here You are right. This is the normalized density of probability. Corrected. n=1 is minimum number for our Fortran program . For model, it may be less.

4.2 4.3

4.4 4.5.2

Density of probabilities should be used for the graphs 4.4 to 4.8 What is the probability in Figure 4.10 How this minimum number is specified?


4.5.4 4.5.5 5.

What is the probability in Figure 4.16 and 4.17? Differential fluence should be plotted in Figure 4.18 and 4.19. This is a very interesting and usefull section with a very valuable effort to compute errors. It can be noted that a previous effort to compute some of the = uncertainties in JPL-85 and JPL-91 models has been published by Rosenqvist and Hilgers [2004] (Rosenqvist, L., and A.Hilgers, Sensitivity of a statistical solar proton fluence model to the size of the event data set, Geophys. Res. Letters, 30, No 16, 1865, 2003). The author of this memo could apply a similar technique for the parameters of its model as he has done actually for the ESP model in section 2.2.5 Density of probability should be used for Figure 5.1 and 5.4 Differential fluence should be used for Figure 5.2 nd 5.5 Density of probability should be used for Figure 5.10. Differential flux should be used for Figure 5.11 and 5.13 The vertical axis of figure 5.15 has no label. Differential flux should be used for Figure 5.20. Section 6.=20 Differential fluence or flux should be used for figures 6.1 to 6.31 (unless it is the fluence or flux above or below a given energy which is plotted).

The same, as everywhere, Fig.4.10 excepted. No comments.... Thank Your, for information.

5. 1

Thank You for Your recommendations, but I cannot use them.

5.1.2

5.3

The authors remarks. The only criterion of the fairness of model is the model ability to explain the experimental data. In you remarks there is no any comments about it. Your comments are minor technical remarks, what don't touch the principal problems. Why do you not comment the main results of model calculation ­ the section 6 of the memorandum?
Why do you not comment the points of the Technical Specification? ISO needs from author the document, what reflect the remarks about Technical Specification only. .