1: \documentstyle[fleqn,epsfig]{article}
2: \textwidth 15.0 cm
3: \textheight 24.0 cm
4: \headsep -1 in
5: \footheight 2 cm
6: \footnotesep .5 cm
7: %\linespread{1.3}
8: \linespread{1.6}
9: \begin{document}
10: \title{About EAS size spectra and primary energy spectra
11: in the knee region}
12: \author{S.V.~Ter-Antonyan\footnote{e-mail: samvel@jerewan1.yerphi.am} ,
13: L.S.~Haroyan \\[0.5cm]
14: \emph{Yerevan Physics Institute, Alikhanyan Brothers 2,
15: Yerevan 375036, Armenia}}
16: \date{}
17: \maketitle
18: \begin{abstract}
19: Based on the unified analyses of
20: KASCADE, AKENO, EAS-TOP and ANI EAS size spectra,
21: the approximations of
22: energy spectra of different primary nuclei have been found.
23: The calculations were carried out using the SIBYLL
24: and QGSJET interaction models in 0.1-100 PeV primary energy range.
25: The results point to existence of both rigidity-dependent steepening
26: energy
27: spectra at $R\simeq200-400$ TV and an additional proton (neutron)
28: component
29: with differential energy spectrum
30: $(6.1\pm0.7)\cdot10^{-11}(E/E_{k})^{-1.5}$
31: (m$^2\cdot$s$\cdot$sr$\cdot$TeV)$^{-1}$
32: before the knee $E_{k}=2030\pm130$ TeV
33: and with power index $\gamma_{2}=-3.1\pm0.05$ after the knee.
34: \end{abstract}
35: \vspace{0.5cm}
36: {\emph{PACS:}} 96.40.Pq, 96.40.De, 98.70.Sa\\
37: {\emph{Keywords:}} cosmic rays, high energy, extensive air shower,
38: interaction model
39:
40: \newpage
41: \section{Introduction}
42: High statistical accuracy in modern EAS experiments in the knee region
43: encouraged the investigation of the fine structure of EAS size spectra.
44: Although the origin of the knee is still a matter of debate, recently
45: the series of publications \cite{EW} appeared, where the sharpness and
46: the spectral structure in the knee region were interpreted
47: by the contribution of heavy nuclei from a single local supernova.
48: Along with this, the absolute differential EAS size spectra measured at
49: different atmosphere depths and different zenith angles are not
50: explained yet from the point of view of a single $A-A_{Air}$
51: interaction model and a single model of primary energy spectra
52: and elemental composition. Such an attempt has been made in
53: \cite{KHR} based on the QGS interaction model and rigidity-dependent
54: steepening primary energy spectra \cite{Peters} for a description of vertical
55: MSU, AKENO and Tien-Shan EAS size spectra. In the present work
56: we worked out a formalism of the inverse problem solution -
57: reconstruction of the primary energy spectrum and elemental composition
58: based on the known EAS size spectra of KASCADE \cite{KAS}, AKENO \cite{AKE},
59: EAS-TOP \cite{TOP} and ANI \cite{ANI} measured at different zenith angles.
60: The calculations were done in the frames of QGSJET \cite{QGS} and SIBYLL
61: \cite{SIB}
62: interaction models. As a primary spectrum we have tested the
63: modified rigidity-dependent steepening primary energy spectra and the
64: hypothesis of the additional component in the knee region
65: \cite{JK,JL,TG,PB2}.
66: In this case the type
67: of nucleus of the additional component was considering as
68: unknown and determining
69: by the best fit of the fine structure of EAS size spectra in the knee
70: region.
71: \section{EAS inverse problem}
72: In general, the energy spectra ($\partial
73: \Im_{A}/\partial E_{0}$) of primary nuclei ($A$)
74: and detectable EAS size spectra ($\partial I
75: / \partial N_{e}^{*}$) are related by the integral equation -
76: \begin{equation}
77: \frac{\partial I(E_{e},\overline{\theta},t)} {\partial N_{e}^{*}}=
78: \sum_{A}
79: \int_{E_{min}}^{\infty}
80: \frac{\partial \Im_{A}} {\partial E_{0}}
81: W_{\theta} (E_{0},A,N_{e}^{*},\overline\theta ,t)
82: dE_{0}
83: \end{equation}
84: where $E_{0},A,\overline{\theta}$ are energy, nucleon number (1-59) and
85: average zenith angle
86: of primary nuclei, $E_{e}$ is an energy threshold of detected EAS
87: electrons, $N_{e}^*(E>E_{e})$ is the estimation value of EAS size
88: obtained by the electron lateral distribution function.
89: Here, by the EAS size ($N_{e}(E>0)$) we mean the total number of EAS
90: electrons
91: at given observation level $(t)$. The kern ($W_{\theta}$) of integral
92: equation (1) is determined as
93: \begin{displaymath}
94: W_{\theta}\equiv
95: \frac{1} {\Delta_{\theta}}
96: \int_{\theta _1}^{\theta _2}
97: \int_{0}^{\infty}
98: \frac{\partial\Omega(E_{0},A,\theta ,t)}{\partial N_{e}}
99: P_{\theta}
100: \frac{\partial \Psi(N_{e})}{\partial N_{e}^{*}}
101: \sin\theta d\theta dN_{e}
102: \end{displaymath}
103: where $\partial\Omega / \partial N_{e}$ is an EAS size spectrum at
104: the observation level $(t)$ for given $E_{0},A,\theta$ parameters
105: of a primary nucleus and depends on $A-A_{Air}$ interaction model;
106: $\Delta_{\theta}=\cos\theta_{1}-\cos\theta_{2}$;
107: \begin{displaymath}
108: P_{\theta}\equiv P(Ne,E_{0},A,\theta)=
109: \frac{1}{X\cdot Y}
110: \int\!\!\int D(N_{e},E_{0},A,\theta,x,y)dxdy
111: \end{displaymath}
112: is a probability to detect an EAS by scintillation array
113: at EAS core coordinates $|x|<X/2$, $|y|<Y/2$ and to obtain the estimations
114: of
115: EAS parameters ($N_{e}^{*}$, $s$ - shower age, $x^{*},y^{*}$ - shower core
116: location)
117: with given accuracies; $\partial \Psi / \partial N_{e}^{*}$ is a distribution
118: of $N_{e}^{*}(N_{e},s,x,y)$ for given EAS size ($N_{e}$).\\
119: One may achieve significant simplification of equation (1)
120: providing the following conditions during experiments:\\
121: a) selection of EAS cores in a range where $P_{\theta}\equiv 1$,\\
122: b) the log-Gaussian form of the measuring error
123: ($\partial\Psi /\partial N_{e}^{*}$) with an average value
124: $\ln(N_{e}\cdot\delta)$ and a RMSD $\sigma_{N}$,
125: where $\delta$ involves all transfer factors (an energy threshold
126: of detected EAS electrons, $\gamma$ and $\mu$ contributions) and
127: slightly depends on $E_{0}$ and $A$,\\
128: c) transformation (standardization) of the measured EAS size spectra to
129: the EAS size spectra at observation level
130: \begin{displaymath}
131: \frac{\partial I(0,\overline{\theta},t)} {\partial N_{e}}\simeq
132: \eta\frac{\partial I(E_{e},\overline{\theta},t)} {\partial N_{e}^{*}},
133: \end{displaymath}
134: where
135: $\eta=\delta^{(\gamma_{e}-1)}\exp\{(\gamma_{e}-1)^{2}\sigma_{N}^{2}/2\}$
136: and $\gamma_{e}$ is the EAS size power index,\\
137: d) consideration of either all-particle primary energy spectrum
138: $\partial \Im_{\Sigma}/\partial E_{0}$ with
139: effective nucleus $A_{eff}(E_{0})$ or energy spectra of primary nuclei
140: ($\partial \Im _{A_{\xi}}/\partial E_{0}$)
141: gathered in a limited number of groups
142: ($\xi=1,\dots\xi_{\max}$) as unknown functions.\\
143: (a-d) conditions make EAS data of different experiments more
144: comparable and equation (1) converts to the form
145: \begin{equation}
146: \frac{\partial I(0,\overline{\theta},t)} {\partial N_{e}}=
147: \eta\int_{E_{min}}^{\infty}
148: \frac{\partial \Im_{\Sigma}} {\partial E_{0}}
149: \frac{\partial\Omega(E_{0},A_{eff}(E_{0}),\overline{\theta},t)}{\partial N_{e}}
150: dE_{0}
151: \end{equation}
152: or
153: \begin{equation}
154: \frac{\partial I(0,\overline{\theta},t)} {\partial N_{e}}=\eta
155: \sum_{\xi=1}^{\xi_{\max}}
156: \int_{E_{min}}^{\infty}
157: \frac{\partial \Im _{A_{\xi}}}{\partial E_{0}}
158: \frac{\partial\Omega(E_{0},\overline{A}_{\xi},\overline{\theta},t)}{\partial N_{e}}
159: dE_{0}
160: \end{equation}
161: However, even in this form the determination of primary energy spectra
162: by measured EAS size spectra and solution
163: of integral equations (2,3) in general is unsolvable problem.
164: At the same time, using the a priori information about energy spectra
165: of primary nuclei ($\partial\Im_{A_{\xi}}/\partial E_{0}$)
166: and the EAS size spectra $\partial I/ \partial N_{e}^{*}\equiv
167: f_{i,j}(N_{e,i}^{*},\overline{\theta}_{j},t)$ measured in $i=1,\dots m$
168: size
169: intervals and $j=1,\dots n$ zenith angular intervals, one may transform
170: the
171: inverse problem into $\chi^{2}$-minimization problem
172: \begin{equation}
173: \min\{\chi^{2}\}\equiv
174: \min\Big\{
175: \sum_{i}^{m}\sum_{j}^{n}
176: \frac{(f_{i,j}-F_{i,j})^{2}}{\sigma_{f}^{2}+\sigma_{F}^{2}}
177: \Big\}
178: \end{equation}
179: with unknown (free) spectral parameters.
180: Here $F_{i,j}\equiv F(N_{e,i}^{*},\overline{\theta}_{j},t)$
181: are the expected EAS size spectra determined at the right
182: hands of equations (1-3) and $\sigma_{f},\sigma_{F}$ are the
183: uncertainties (RMSD) of measured ($f_{i,j}$) and expected ($F_{i,j}$)
184: shower size spectra.\\
185: One may also unify the data of different experiments applying
186: minimization $\chi^2_{U}$ with re-normalized EAS size spectra
187: \begin{equation}
188: \min\{\chi^2_{U}\}\equiv
189: \min\Big\{\chi^{2}\Big(
190: \frac{f_{i,j,k}}{\sum_{i}\sum_{j} f_{i,j,k}},
191: \frac{F_{i,j,k}}{\sum_{i}\sum_{j} F_{i,j,k}}
192: \Big)\Big\}
193: \end{equation}
194: where index $k=1,\dots l$ determines the observation levels ($t$)
195: of experiments. Expression (5) offers an advantage for experiments
196: where the values of methodical shift ($\delta$) and measuring error
197: ($\sigma_{N}$) are unknown or known with insufficient accuracy.\\
198: The energy spectra of primary nuclei are preferable to determine (a
199: priori) in the following generalized form
200: \begin{equation}
201: \frac{\partial \Im_{A}} {\partial E_{0}}\simeq\beta\cdot\Phi_{A}\cdot
202: E_{0}^{-\gamma_{1}(A)}\cdot
203: \Big(1+ \Big(\frac{E_{0}}{E_{knee}(A)}\Big)^{\epsilon}\Big)^
204: %{\frac{\gamma_{1}(A)-\gamma_{2}}{\epsilon}}
205: {(\gamma_{1}(A)-\gamma_{2})/\epsilon}
206: \end{equation}
207: Unknown (free) spectral parameters in approximation (6)
208: are $\beta$,
209: $E_{knee}(A)$ (so called "knee" of energy spectrum of $A$ nucleus),
210: $\gamma_{1}$ and $\gamma_{2}$ (spectral asymptotic slopes before and after
211: knee),
212: $\epsilon$ (sharpness parameter of knee, $1\leq \epsilon \leq 10$).
213: The values of $\Phi_{A}$ and $\gamma_{1}(A)$ parameters
214: are known from approximations of balloon and satellite data \cite{PB1}
215: at $A\equiv 1,4,\dots 59$ and $E_{0}\simeq 1-10^{3}$ TeV. Parameter
216: $\beta\simeq 1$ determines the normalization of spectra (6) in
217: $10^{2}-10^{5}$ TeV energy range.\\
218: Thus, minimizing $\chi^2$- functions (4,5) on the basis of
219: measured values of $\partial I(\overline{\theta}_{i,k})/ \partial N_{ej,k}$
220: and corresponding expected EAS size spectra (2,3)
221: at given $m$ zenith angular intervals, $n$ EAS size
222: intervals and $l$ experiments one may evaluate the parameters of the
223: primary spectrum (6).
224: Evidently, the accuracies of solutions for spectral parameters strongly
225: depend on the number of measured intervals ($m\cdot n\cdot l$),
226: statistical errors and correctness of
227: $\partial\Omega(E_{0},A,\theta,t)/\partial
228: N_{e}$ determination in the framework of a given interaction model.
229:
230: \section{Results}
231: Here, the parametric solutions of the EAS inverse problem are obtained
232: on the basis of KASCADE \cite{KAS} ($t=$1020 g/cm$^{2}$), AKENO \cite{AKE}
233: (910 g/cm$^{2}$), EAS-TOP \cite{TOP} (810 g/cm$^{2}$) and ANI \cite{ANI}
234: (700 g/cm$^{2}$) published EAS size spectra.
235: These experiments were carried out at different observation levels and were
236: chosen for two reasons: satisfaction of (a-c) conditions from the section
237: (2)
238: and a high statistical accuracy of presented data (especially KASCADE
239: experiment).
240: Unfortunately, during the standardization of EAS size spectra (condition
241: (c))
242: the value of $\eta$ parameter is not always known with proper accuracy,
243: which is the main reason of discrepancy in the results of
244: different experiments. In our calculations we included $\eta$ in the list
245: of unknown spectral parameters and determined by the minimization of
246: functional (4). The problem does not exist if the minimization of
247: re-normalized EAS size spectra is unified, since the linear parameters
248: ($\eta_{k}\cdot\beta$) are canceled out from the functional (5).\\
249: The differential EAS size spectra $\partial\Omega (E_{0},A,\theta ,t)/
250: \partial N_{e}$ for given
251: $E_{0}\equiv 0.032,0.1,\dots,100$ PeV,
252: $A\equiv 1,4,12,16,28,56$,
253: $t\equiv 0.5,0.6,\dots,1$ Kg/cm$^{2}$,
254: $\cos\theta\equiv 0.8,0.9,1$
255: were calculated using CORSIKA562(NKG) EAS simulation code \cite{COR} at
256: QGSJET \cite{QGS} and SIBYLL \cite{SIB} interaction models.
257: Intermediate values are calculated using
258: 4-dimensional log-linear interpolations. The estimations of errors of
259: the expected EAS size spectra $\partial\Omega /\partial N_{e}$
260: at fixed $E_{0},A,\theta,t$ parameters did not exceed $3-5\%$.\\
261: The basic results of minimizations (4,5) at a given number ($\nu$) of
262: unknown spectral parameters and the values of $\chi^{2}/q$
263: (or $\chi_{U}^{2}/q_{u}$ for unified data),
264: are presented in Tables~1-4, where $q=mn-\nu-1$
265: and $q_{u}=\sum_{k}(mn)_{k}-\nu-1$ are corresponding degrees of freedom.
266: The upper (lower) rows of each experiment in Tables~1 and 4
267: correspond to parameters obtained by the QGSJET (SIBYLL) interaction
268: model.
269: \subsection{Test of rigidity-dependent energy spectra}
270: Table~1 contains the approximation values of spectral parameters at
271: rigidity-dependent approach \cite{Peters}
272: \begin{equation}
273: E_{knee}(A)=R\cdot Z
274: \end{equation}
275: where $Z$ is a charge of $A$ nucleus and R is a
276: parameter of magnetic rigidity (or a critical (cutoff) energy
277: in more modern model \cite{PB2,PB1}).
278: The results were obtained by minimizations (4,5) applying (6,7).
279: The magnitudes of $\gamma_1(A)$ for all nuclei were taken from \cite{PB1}.
280: The number of unknown (free) parameters is equal to $\nu=4$ and
281: corresponding
282: solutions at two interaction models are presented in Table~1.\\
283: \begin{table}
284: \begin{center}
285: \begin{tabular}{|r|c|c|c|c|l|}
286: \hline
287: Experiment& R [TV] &$\gamma_{2}$ &$\epsilon$&$\eta\cdot\beta$&$\chi^{2}/q$\\
288: \hline
289: KASCADE&2390$\pm$190&3.46$\pm$0.12&2.2$\pm$0.3 & 1.05$\pm$0.08&1.3 \\
290: \cline{2-6}
291: $m\cdot n=24\cdot5$&2310$\pm$220&3.45$\pm$0.12& 1.8$\pm$0.2 & 0.69$\pm$0.05& 3.0 \\
292: \hline
293: AKENO &3150$\pm$120&3.50$\pm$0.14& 10$\pm$7.0 & 1.98$\pm$0.06& 2.2 \\
294: \cline{2-6}
295: $m\cdot n=20\cdot3$ &2820$\pm$110 &3.50$\pm$0.31& 10$\pm$? & 1.48$\pm$0.04& 3.1 \\
296: \hline
297: EAS-TOP&1450$\pm$120 &3.35$\pm$0.11&2.3$\pm$0.5 & 1.43$\pm$0.03& 1.2 \\
298: \cline{2-6}
299: $m\cdot n=24\cdot5$&1540$\pm$205&3.35$\pm$0.20 & 1.4$\pm$0.3 & 1.15$\pm$0.04& 0.5 \\
300: \hline
301: ANI&2030$\pm$245 &3.47$\pm$0.18& 2.1$\pm$0.5 & 1.07$\pm$0.02& 0.8 \\
302: \cline{2-6}
303: $m\cdot n=23\cdot3$ &2230$\pm$320&3.49$\pm$0.23 & 1.9$\pm$0.4 & 0.87$\pm$0.02& 1.0 \\
304: \hline
305: Unified data &2610$\pm$710&3.47$\pm$0.23 & 1.3$\pm$0.2 & - & 1.7\\
306: \cline{2-6}
307: $\sum m\cdot n=369$&3000$\pm$650&3.49$\pm$0.30 & 1.2$\pm$0.1 & - & 2.3\\
308: \hline
309: \end{tabular}
310: \caption{Rigidity ($R$), slope ($\gamma_{2}$), "sharpness"
311: ($\epsilon$), shift ($\eta\cdot\beta$) and corresponding $\chi^2/q$ values
312: obtained by approximations of EAS size spectra at
313: QGSJET (upper rows) and SIBYLL (lower rows) interaction models
314: and rigidity-dependent assumption (7).}
315: \end{center}
316: \end{table}
317: The stability of solutions (or the steep of $\chi^2$ global minimum)
318: of minimizations (4,5) is seen from obtained errors. So, it is seen that
319: the $\chi^2$ minimum for AKENO data does not depend on the
320: $\varepsilon$ sharpness parameter at SIBYLL and partly at QGSJET
321: interaction models. \\
322: The obtained slopes of primary energy spectra
323: after knee agree with the same calculations \cite{KHR} performed
324: by QGS model and exceed well known expected values ($3-3.1$)
325: in the $\sim10^{17}$ eV energy range \cite{PB1}.
326: Such a steep of primary spectra after the knee is a result of
327: $(\gamma_2,R)$ correlations in (4-7).
328: In case of $R\simeq2000$ TeV the $E_k(Fe)\simeq 5.2\cdot 10^4$ TeV and
329: the primary energy spectra in the large interval ($E_k(H)-E_k(Fe)$) at
330: fixed $\gamma_1(A_\xi)$ can be conformed with corresponding EAS size spectra
331: provided abnormally steep slope ($\gamma_2\simeq3.4-3.5$) after the
332: knee.\\
333: The results of expected EAS size
334: spectra in comparison with corresponding experimental data
335: are shown in Fig.~1 (the thin solid lines by QGSJET model, the thin dashed
336: lines by SIBYLL model).
337: Despite the satisfactory agreement ($\chi^2\sim1$) of EAS size spectra
338: with predictions of rigidity-dependent steepening spectra (6,7)
339: and QGSJET interaction
340: model the form (fine structure) of the measured EAS size spectra in the knee
341: region differs from the form of corresponding expected spectra.
342: It is worth mentioning that the difference is formally small and does not exceed
343: several percents in precise KASCADE data.\\
344: %%%%%%%%%%%%%%%%%
345: As a next step, we attempted to test the
346: rigidity-dependent approximation (7) directly.
347: The knee-energies $E_k(A_{\xi})$ were chosen as unknown (free)
348: parameters in the approximations of EAS data (4-7),
349: where $\xi=1,\dots \xi_{\max}$.
350: However, the stable solutions for free parameters
351: were obtained only at unified minimization (5), fixed values
352: $\gamma_1(A_{\xi}),\gamma_2,\varepsilon$
353: and number of nuclear groups $\xi_{\max}\le 5$.
354: In Table~2 the values of spectral parameters $E_{knee}(A_{\xi})$ obtained
355: by minimization of $\chi_{U}^2$ (5) for 5 groups
356: of primary nuclei ($H$), ($He,Li$), ($Be-Na$), ($Mg-Cl$), ($Ar-Ni$)
357: at given values of $\gamma_{2}=3.1$ , $\epsilon=4.0$ and $\nu=5$ are
358: presented.
359: It is seen, that approach (7)
360: is performed only for nuclei with $A>1$ and $R\simeq400$ TV
361: or (7) is valid for all nuclei at $R\simeq400$ TV
362: but there is an additional proton flux with
363: $E_{knee}^{(p)}\simeq3000$ TeV which shifts the
364: knee value of the total proton energy spectrum. \\
365: \begin{table}
366: %[h]
367: \begin{center}
368: \begin{tabular}{|r|c|c|c|c|c|l|}
369: \hline
370: Model&$E_{k}({\emph{H}})$[TeV]&$E_{k}({\emph{He,Li}})$&$E_{k}({\emph{Be-Na}})$
371: &$E_{k}({\emph{Mg-Cl}})$&$E_{k}({\emph{Ar-Ni}})$&$
372: \chi_{U}^{2}/q_{u}$\\
373: \hline
374: QGSJET&3070$\pm$160&790$\pm$70& 5550$\pm$60&6310$\pm$50& 9020$\pm$170&1.7\\
375: \hline
376: SIBYLL&3150$\pm$90&610$\pm$30& 6060$\pm$65 &6970$\pm$40&9780$\pm$90&2.3\\
377: \hline
378: \end{tabular}
379: \caption{Spectral parameters $E_{knee}(A)$ $[TeV]$
380: for different groups of primary nuclei
381: and different interaction models. The results obtained by unified
382: analyses of EAS data at $\gamma_{2}=3.1$ and $\epsilon=4.$}
383: \end{center}
384: \end{table}
385: The following testing of the rigidity-dependent
386: approach (7) is based on the investigation of the all-particle energy
387: spectrum. Toward this end the
388: fits of primary energy spectra $d\Im_A/dE_0$
389: for different nuclei ($A=1,\dots 59$) known from
390: \cite{PB1}
391: were extrapolated up to $E_{A}=10^5$ TeV energies taking into account
392: (6,7) at $R\simeq 600$ TeV \cite{PB2}.
393: The obtained expected all-particle energy spectrum
394: $dI_{\Sigma}/dE_0=\sum_{A} d\Im_A/dE_0$ was approximated by expression
395: similar to (6) at five free parameters
396: ($\Phi_\Sigma,\gamma_1,\gamma_2,E_k,\varepsilon$).
397: The average values of primary nuclei ($\exp(\overline{\ln A})$)
398: obtained from extrapolations of spectra \cite{PB1} were approximated by
399: step function
400: \begin{equation}
401: A_{eff}(E_{0})=a+b\cdot\ln\big(\frac{E_{0}}{E_{k}}\big)
402: \end{equation}
403: where $b=b_{1}$ at $E_{0}<E_{k}$ and $b=b_{2}$ at $E_{0}>E_{k}$.
404: The results of these approximations
405: are presented in a last row of Table~3 .
406: The value of sharpness parameter was equal to $\varepsilon=1\pm0.1$.\\
407: The first and second rows of Table~3 content the
408: parameters of the all-particle energy spectra
409: ($\partial \Im_{\Sigma}/\partial E_{0}$),
410: which were obtained by minimization $\chi_{U}^2$ (expressions 2,5-7)
411: of unified EAS size data at $\nu=6$ and
412: $\varepsilon=1$. The approximation (8) has been used
413: for $A_{eff}(E_0)$ which is at the right hand of expression (2).
414: It is necessary to note that the solutions for
415: $a,b_1,b_2$ parameters
416: can be obtained only by re-normalized EAS size
417: spectra (5) because of the strong
418: correlation between a linear parameter $\eta$ and
419: an effective nucleus $A_{eff}$.\\
420: It is seen from Table~3 that the model of
421: rigidity-dependent energy spectra predicts the
422: increase ($b_1>0$, third row of Table~3) of the mean
423: nucleus with energy, whereas the presented analysis of
424: EAS data points out a decrease of $A_{eff}$ with energy
425: ($b_1<0$, first two rows of Table~3) in the energy range of $E_0<E_k$.
426: It is obvious, that despite $\overline{A}$ could not be exactly
427: equal to $A_{eff}$, at least their dependence on energy
428: must be the same. \\
429: The results of recent precise experiments DICE \cite{DC} and
430: CASA-BLANKA \cite{CB} also point out to the decrease of $\overline{\ln A}$
431: with energy at $E_0<E_k$.
432: This dependence of $A_{eff}$ and $\overline{\ln A}$ on energy might be
433: explained by the contribution of an additional light
434: component in a primary nuclei flux.\\
435: \begin{table}
436: %[h]
437: \begin{center}
438: \begin{tabular}{|r|c|c|c|c|c|c|l|}
439: \hline
440: Model&$\gamma_{1}$ &$\gamma_{2}$ &$E_{k}$ [TeV]&$a$&$b_{1}$&$b_{2}$&$\chi^{2}_{U}/q_{u}$\\
441: \hline
442: QGSJET&2.72$\pm$0.01&3.05$\pm$0.03&2180$\pm$110&1.71$\pm$0.14&-0.30$\pm$0.13&0.86$\pm$0.12&1.5\\
443: \hline
444: SIBYLL&2.79$\pm$0.02&3.09$\pm$0.02&3680$\pm$230&10.5$\pm$1.00&-0.28$\pm$0.20&4.24$\pm$0.61&1.4\\
445: \hline
446: R=600TV&2.66$\pm$0.01&3.09$\pm$0.01&3400$\pm$200&7.0$\pm$0.5
447: &0.25$\pm$0.15 &3.0$\pm$0.5 &1.2\\
448: \hline
449: \end{tabular}
450: \caption{Parameters of all-particle energy spectrum
451: obtained by unified EAS data (the first and second rows).
452: The last row corresponds to results of extrapolation of
453: fits \cite{PB1} taking into account the assumptions (6,7) at $R=600$ TV.}
454: \end{center}
455: \end{table}
456:
457: From the above analyses follows that rigidity-dependent steepening energy
458: spectra in combination with QGSJET or SIBYLL interaction model
459: can not
460: explain the obtained results of the fine structure of EAS size spectra
461: \cite{KAS,AKE,TOP,ANI} in
462: the knee region (Table~1 and Fig.~1), the large values of knee $E_{k}(H)$
463: for Hydrogen component (Table~2) and dependence of the effective nucleus
464: $A_{eff}(E_{0})$ on primary energy before the knee (Table~3).
465: In this connection we have carried on the search of more adequate model of
466: primary energy spectra and elemental composition.
467: \subsection{Test of additional component}
468: Based on predictions \cite{JK,JL,TG,PB2} the primary energy spectra
469: in approximation (6) have been added by a new (polar cap \cite{PB2}) component
470: $\partial \Im_{Add}/\partial E_{0}$ with power energy spectrum
471: \begin{equation}
472: \frac{\partial \Im_{Add}} {\partial E_{0}}=
473: \Phi^{(p)}\big(E_{k}^{(p)}\big)^{\gamma_{1}^{(p)}}
474: \Big(\frac{E_0}{E_{k}^{(p)}}\Big)^{-\gamma^{(p)}}
475: \end{equation}
476: where $\gamma^{(p)}=\gamma_{1}^{(p)}$ at $E_{0}<E_{k}^{(p)}$
477: and $\gamma^{(p)}=\gamma_{2}$ at $E_{0}>E_{k}^{(p)}$.\\
478: New spectral parameters $\Phi^{(p)}, \gamma^{(p)},E_{k}^{(p)}$
479: and nucleon number $A^{(p)}$ of the additional component are considered
480: as unknown and determined together with parameters of
481: rigidity-dependent energy spectra (6,7) by minimization of $\chi^2$ and
482: $\chi^2_{U}$ (4,5) at $\nu=7$. The results of expected EAS size
483: spectra for each experiment (KASCADE, AKENO, EAS-TOP and ANI)
484: taking into account contribution of additional component (9) are shown
485: in Fig.~1 (the thick solid lines by QGSJET model and the thick dashed
486: lines by SIBYLL model). It is seen that the additional component
487: with high accuracy (2-5\% for KASCADE and ANI data) describes the
488: fine structure of EAS size spectra in the knee region.
489: The values of slopes of additional component before the knee turned
490: out
491: to be $\gamma_{1}^{(p)}\simeq1.5\pm^{0.6}_{0.2}$.
492: The nucleon number ($A^{(p)}$) of this component with high reliability
493: did not exceed of $A^{(p)}=1$ for most of experiments
494: especially at QGSJET interaction model
495: (except from AKENO ($A^{(p)}\simeq56$)).
496: The unified analyses of all experiments at QGSJET and SIBYLL interaction
497: models also gave a proton or neutron
498: ($A^{(p)}=1$) composition of the additional component.
499: The values of other spectral parameters at
500: $\gamma_{1}^{(p)}=1.5$ and $A^{(p)}=1$ are included in Table~4.\\
501: The obtained result disagrees with \cite{EW} where the alike component
502: consists of several heavy nuclei. However, our result is based on the high
503: accuracy of the coincidence ($\sim2-3\%$ for KASCADE data)
504: of expected and measured EAS size spectra by both $\chi^2$ criterion and the
505: overlapping of fine structure of spectra.\\
506: The comparison of parameters $\gamma_2$ from Tables~1,4
507: shows that spectral slopes after the knee from Table~4
508: roughly overlap with the expected slope ($3-3.1$) well known
509: from $N_e\gg10^7$ EAS data. This can be explained the fact that
510: the spectral brake of summary proton component
511: ($E_k^{(p)}\simeq2\cdot10^3$ TeV) is closer to the
512: iron component ($E_k(Fe)\simeq10^4$ TeV) at $R\simeq400$ TV.\\
513: The coexistence of rigidity-dependent primary energy spectra and
514: additional proton flux with spectral parameters $E_k^{(p)}\simeq2000$
515: TeV, $\gamma_1^{(p)}\simeq1.5$, $\gamma_2^{(p)}=\gamma_2\simeq3.1$
516: explains also the shifted value of the knee $E_k(H)$ in Table~2
517: and the decrease of $A_{eff}(E_0)$ at $E_0<E_k$ (see section 3.1).\\
518: Although the additional component
519: does not contribute to shower sizes below or above
520: the knee significantly , it is essential for the reproducing of the
521: sharp knee of EAS size spectra. This idea is taken from \cite{PB2} and
522: is confirmed here.\\
523: \begin{table}
524: \begin{center}
525: \begin{tabular}{|r|c|c|c|c|c|l|}
526: \hline
527: Experiment&R [TV] &$E_{k}^{(p)}$[TeV]&$\Phi^{(p)}\cdot10^6$
528: &$\gamma_{2}$&$\eta\cdot\beta$ &$\chi^{2}/q$\\
529: \hline
530: KASCADE&195$\pm$30&1960$\pm$150&11.6$\pm$1.0&3.15$\pm$0.04&1.11$\pm$0.02&0.8\\
531: \cline{2-7}
532: &150$\pm$20&1820$\pm$85&12.8$\pm$0.6 &3.14$\pm$0.03&0.74$\pm$0.07&2.3\\
533: \hline
534: AKENO &290$\pm$180&4240$\pm$370&4.04$\pm$0.10&3.25$\pm$0.05&2.60$\pm$0.07&2.0\\
535: \cline{2-7}
536: &290$\pm$210&3860$\pm$310&4.59$\pm$0.16&3.24$\pm$0.03&1.86$\pm$0.02&3.0\\
537: \hline
538: EAS-TOP&390$\pm$150&1960$\pm$30&6.08$\pm$0.10&3.16$\pm$0.02 & 1.43$\pm$0.01&1.1\\
539: \cline{2-7}
540: &150$\pm$95&2110$\pm$130&4.37$\pm$0.50&3.11$\pm$0.04&1.41$\pm$0.04&0.3\\
541: \hline
542: ANI&240$\pm$130&2060$\pm$145 &6.80$\pm$0.80&3.09$\pm$0.03&1.14$\pm$0.07& 0.5 \\
543: \cline{2-7}
544: &160$\pm$85&2050$\pm$130 &6.38$\pm$0.70&3.05$\pm$0.02&1.02$\pm$0.02& 0.8 \\
545: \hline
546: Unified data &390$\pm$30&2030$\pm$130&5.55$\pm$0.60 &3.09$\pm$0.01 & - & 1.6\\
547: \cline{2-7}
548: $\sum m\cdot n=369$&322$\pm$15&2150$\pm$100&4.92$\pm$0.35 &3.08$\pm$0.01 & - & 2.2\\
549: \hline
550: \end{tabular}
551: \caption{Spectral parameters taking into account the
552: contribution of additional proton component.}
553: \end{center}
554: \end{table}
555:
556: The final all-particle energy spectrum ($\partial I/\partial E_{0}$)
557: obtained by unified EAS size data at QGSJET interaction model
558: \begin{displaymath}
559: \frac{\partial I} {\partial E_{0}}=\beta\Big( \sum_{A}
560: \frac{\partial \Im_{A}} {\partial E_{0}}+
561: \frac{\partial \Im_{Add}} {\partial E_{0}}\Big)
562: \end{displaymath}
563: and corresponding energy spectra ($\partial \Im_{A}/\partial E_{0}$)
564: of 6 nuclear groups with additional component
565: ($\partial\Im_{Add}/\partial E_{0}$)
566: at normalization $\beta=1$ \cite{PB1} are presented in Fig.~2.
567: The solid (dashed) line is the all-particle energy spectrum
568: obtained by unified (only KASCADE) EAS size spectra.
569: Dotted lines are the energy spectra of different primary components
570: obtained by unified EAS data. Symbols in Fig.~2 are the data from
571: DICE \cite{DC}, CASA-BLANKA \cite{CB} arrays and reviews \cite{PB1,REV}.\\
572: The numerical values of all-particle energy spectrum (solid line in
573: Fig.~2), corresponding error, spectrum of additional proton component
574: and average nucleus versus primary energy are presented in
575: Table~5.
576: \begin{table}
577: \begin{center}
578: \begin{tabular}{|r|c|c|c|l|}
579: \hline
580: & $\partial I/\partial E_0$ & $\Delta(\partial I/\partial E_0$)&
581: $\partial \Im_{Add}/\partial E_0$ & \\
582: $E_0$ (TeV)& $(m^2\cdot sr\cdot s\cdot $&$\pm(m^2\cdot sr\cdot $ & $(m^2\cdot sr\cdot
583: s\cdot $&$\overline{\ln A}$ \\
584: & $ \cdot TeV)^{-1}$&$\cdot s\cdot TeV)^{-1}$ &$ \cdot TeV)^{-1}$&\\
585: \hline
586: 0.200E+03 & 0.174E-06 & 0.15E-07 & 0.196E-08 & 1.71 \\
587: 0.250E+03 & 0.964E-07 & 0.87E-08 & 0.140E-08 & 1.72 \\
588: 0.312E+03 & 0.534E-07 & 0.50E-08 & 0.100E-08 & 1.73 \\
589: 0.391E+03 & 0.296E-07 & 0.31E-08 & 0.719E-09 & 1.73 \\
590: 0.488E+03 & 0.161E-07 & 0.18E-08 & 0.514E-09 & 1.76 \\
591: 0.610E+03 & 0.884E-08 & 0.10E-08 & 0.368E-09 & 1.78 \\
592: 0.763E+03 & 0.486E-08 & 0.56E-09 & 0.263E-09 & 1.79 \\
593: 0.954E+03 & 0.260E-08 & 0.33E-09 & 0.188E-09 & 1.81 \\
594: 0.119E+04 & 0.140E-08 & 0.18E-09 & 0.135E-09 & 1.82 \\
595: 0.149E+04 & 0.762E-09 & 0.10E-09 & 0.965E-10 & 1.81 \\
596: 0.186E+04 & 0.418E-09 & 0.57E-10 & 0.681E-10 & 1.79 \\
597: 0.233E+04 & 0.224E-09 & 0.32E-10 & 0.395E-10 & 1.81 \\
598: 0.291E+04 & 0.117E-09 & 0.18E-10 & 0.199E-10 & 1.88 \\
599: 0.364E+04 & 0.609E-10 & 0.10E-10 & 0.100E-10 & 1.94 \\
600: 0.455E+04 & 0.316E-10 & 0.58E-11 & 0.502E-11 & 2.00 \\
601: 0.568E+04 & 0.163E-10 & 0.33E-11 & 0.252E-11 & 2.05 \\
602: 0.711E+04 & 0.843E-11 & 0.18E-11 & 0.126E-11 & 2.11 \\
603: 0.888E+04 & 0.435E-11 & 0.10E-11 & 0.634E-12 & 2.16 \\
604: 0.111E+05 & 0.222E-11 & 0.57E-12 & 0.318E-12 & 2.19 \\
605: 0.217E+05 & 0.280E-12 & 0.77E-13 & 0.402E-13 & 2.19 \\
606: 0.827E+05 & 0.448E-14 & 0.14E-14 & 0.642E-15 & 2.19 \\
607: \hline
608: \end{tabular}
609: \caption{Expected
610: all-particle flux, error, flux of additional component
611: and average nucleus at different primary energies.}
612: \end{center}
613: \end{table}
614: \section*{Conclusion}
615: High statistical accuracy of experiments KASCADE, EAS-TOP, AKENO and
616: ANI allowed to obtain approximations
617: of primary energy spectra and elemental composition
618: with accuracy $\sim15\%$ in the knee region. Along with this,
619: KASCADE and ANI EAS size spectra at $\theta < 37^0$ are described
620: with the accuracy of $\sim2-5\%$ in a whole measurement interval.\\
621: Obtained results
622: show the evidence of QGSJET interaction model at least in $10^5-10^7$ TeV
623: energy range, rigidity-dependent steepening primary energy spectra at
624: $R\simeq200-400$ TV and existence of the additional proton (or neutron)
625: component
626: with spectral power index $\gamma^{(p)}_{1}\simeq 1.5\pm^{0.6}_{0.2}$
627: before the knee
628: $E_{k}^{(p)}\simeq 2030$ TeV. The contribution of the additional
629: proton (neutron)
630: component in all-particle energy spectrum turned out to be
631: $20\pm5\%$ at primary energy $E_{0}=E_{k}^{(p)}$.
632: \section*{Acknowledgements}
633: We thank Peter Biermann for extensive discussions and
634: Heinigerd Rebel, Johannes Knapp and Dieter Heck for providing the CORSIKA code.
635:
636: \newpage
637: \begin{thebibliography}{15}
638: \bibitem{EW} A.D.~Erlykin, A.W.~Wolfendale, Astropart. Phys. 7 (1997) 1
639: //Astropart. Phys. 7 (1997) 203 // Astropart. Phys. 8 (1998) 265.
640: \bibitem{KHR}G.B.~Khristiansen, Yu.A.~Fomin, N.N.~Kalmykov et al.,
641: Proc. 24th ICRC, {\bf{2}}, Rome (1995) 772.
642: \bibitem{Peters} B.~Peters, Nuovo Cimento (Suppl.) {\bf{14}}, (1959) 436
643: // Nuovo Cimento {\bf{22}}, (1961) 800.
644: \bibitem{KAS}R.~Classtetter et al. (KASCADE Collaboration), Nucl.Phys. B
645: (Proc.Suppl.) 75A (1999) 238.
646: \bibitem{AKE}M.~Nagano, T.~Hara et al., J.Phys.G:Nucl.Phys. {\bf{10}}
647: (1984) 1295 // ICR-Report 16-84-5, Tokyo (1984) 30 p.
648: \bibitem{TOP}M.~Aglietta et al. (EAS-TOP Collaboration), INFN/AE-98/21
649: (1998) 15 p.// M.~Aglietta et al., Astropart. Phys. (10) 1 (1999) 1.
650: \bibitem{ANI}A.~Chilingaryan et al., Proc. 26th ICRC, Salt Lake City,
651: {\bf{1}} (1999) 240.
652: \bibitem{QGS}N.N.~Kalmykov, S.S.~Ostapchenko, Yad. Fiz. {\bf{56}} (1993)
653: 105 // Phys.At.Nucl. {\bf{56}} (3) (1993) 346.
654: \bibitem{SIB}R.S.~Fletcher, T.K.~Gaisser, P.~Lipari, T.~Stanev, Phys.Rev. D
655: {\bf{50}} (1994) 5710 // J.~Engel, T.K.~Gaisser, P.~Lipari, T.~Stanev,
656: Phys.Rev.D. {\bf{46}} (1992) 5013.
657: \bibitem{JK} J.~Kempa et al., J.Phys.A: Math.,Gen. {\bf{7}} (1974) 1213.
658: \bibitem{JL} J.~Linsley, Proc. 18th ICRC, Bangalore (1983) {\bf{12}}, 135.
659: \bibitem{TG} T.K.~Gaisser and T.Stanev, Proc. 22nd ICRC, Dublin (1991).
660: \bibitem{PB2} T.~Stanev, P.L.~Biermann, T.K.~Gaisser, Astron. Astrophys.
661: 274 (1993) 902 // P.L.~Biermann, Preprint MPIfR, Bonn, No 700 (1996) 6.
662: \bibitem{PB1} B.~Wiebel-Sooth and P.~Biermann,
663: Preprint Max-Planck Inst. f$\ddot{u}$r Radioastr., Bonn, No.772,
664: (1998) 50 p.
665: \bibitem{COR}D.~Heck, J.~Knapp, J.N.~Capdevielle, G.~Schatz, T.~Thouw,
666: Forschungszentrum Karlsruhe Report, FZKA 6019 (1998) 90 p.
667: \bibitem{REV}S.~Petrera, Proc. 24th ICRC, Rapp. Papers, Rome (1995) 737.
668: {\bf{2}}, 101.
669: \bibitem{DC} S.P.~Swordy, D.B.~Kieda, Astropart. Phys. 13 (2000) 137.
670: \bibitem{CB} J.W.~Fowler, L.F.~Fortson et al., Preprint, arXiv:astro-ph/0003190(2000).
671: \end{thebibliography}
672:
673: \newpage
674: \begin{figure}[htb]
675: \begin{center}
676: \mbox{\epsfig{file=ne.eps,width=14cm,height=15cm}}
677: \end{center}
678: \vspace{-1cm}
679: \caption{KASCADE, AKENO, EAS-TOP and ANI EAS size spectra
680: (symbols). The thin (thick) lines correspond to predictions
681: via rigidity-dependent steepening primary spectra (with the additional
682: proton component). The solid and dashed lines correspond to
683: the QGSJET and SIBYLL interaction models respectively.}
684: \end{figure}
685:
686: \newpage
687: \begin{figure}[htb]
688: \begin{center}
689: \mbox{\epsfig{file=p2.eps,width=14cm,height=15cm}}
690: \end{center}
691: \vspace{-1cm}
692: \caption{Primary energy spectra and elemental composition.
693: The solid (dashed) line is the all-particle energy spectrum obtained by
694: unified (only KASCADE) EAS data. The dotted lines are the energy spectra
695: of different nuclear groups. The A.c.~dotted line is the energy spectrum
696: of additional proton component. The symbols are the data from \cite{DC,CB}
697: and reviews \cite{PB1,REV}.}
698: \end{figure}
699:
700: \end{document}
701:
702:
703: