1: %%%%%%%%%%%%%%%%%%%%%%%%%% author.tex %%%%%%%%%%%%%%%%%%%%%%%%%
2: %
3: % sample root file for your contribution to a "contributed book"
4: %
5: % "contributed book"
6: %
7: % Use this file as a template for your own input.
8: %
9: %%%%%%%%%%%%%%%%%%%%%%%% Springer-Verlag %%%%%%%%%%%%%%%%%%%%%%%%%%
10:
11:
12: % RECOMMENDED %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
13: \documentclass[multphys,vecphys]{svmult}
14:
15: % choose options for [] as required from the list
16: % in the Reference Guide, Sect. 2.2
17:
18: \usepackage{makeidx} % allows index generation
19: \usepackage{graphicx} % standard LaTeX graphics tool
20: % when including figure files
21: \usepackage{multicol} % used for the two-column index
22: \usepackage[bottom]{footmisc}% places footnotes at page bottom
23: % etc.
24: % see the list of further useful packages
25: % in the Reference Guide, Sects. 2.3, 3.1-3.3
26:
27: \makeindex % used for the subject index
28: % please use the style sprmidx.sty with
29: % your makeindex program
30:
31: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
32:
33: \begin{document}
34:
35: \title*{Altitude Effect in \v{C}erenkov Light Flashes of
36: Low Energy Gamma-ray-induced Atmospheric Showers}
37: % Use \titlerunning{Short Title} for an abbreviated version of
38: % your contribution title if the original one is too long
39:
40: \titlerunning{Altitude Effect in the \v{C}erenkov Light Flashes}
41:
42: \author{A. Konopelko\inst{1,2}}
43: % Use \authorrunning{Short Title} for an abbreviated version of
44: % your contribution title if the original one is too long
45: \institute{Max-Planck-Institut f\"{u}r Kernphysik, Heidelberg
46: \texttt{alexander.konopelko@mpi-hd.mpg.de} \and
47: Humboldt-Universit\"{a}t zu Berlin, Institut f\"{u}r Physik,
48: Adlershof \texttt{konopelk@physik.hu-berlin.de}}
49: %
50: % Use the package "url.sty" to avoid
51: % problems with special characters
52: % used in your e-mail or web address
53: %
54:
55: \maketitle
56:
57: \begin{abstract}
58: At present the ground-based Very High Energy (VHE) gamma-ray
59: astronomy is racing to complete construction of a number of modern
60: gamma-ray detectors, i.e. CANGAROO III, MAGIC, H.E.S.S., and
61: VERITAS. They should be fully operational in a year's time. After
62: much debate, the further development of this gamma-ray astronomy in
63: the foreseeable future must be widely anticipated to proceed with
64: the designing and building of a new instrumentation, which is
65: primarily intended for the further drastic reduction of the energy
66: threshold in gamma-ray observations down to about 10 GeV. On the
67: ground one can hardly reach such low energy thresholds without
68: considerably larger, up to 30 m diameter, optical telescopes, which
69: might be able to collect sufficient amount of \v{C}erenkov light
70: from the atmospheric gamma-ray showers of that low energy. If not
71: taken off the ground entirely (like GLAST), then it seems to be
72: profitable to mount future low energy \v{C}erenkov telescopes at
73: higher altitudes in the atmosphere in order that they will be able
74: to detect substantially more unabsorbed \v{C}erenkov light from a
75: shower. There are a few sites up on the high mountains of roughly 5
76: km height worldwide, which can be used for such a venture. However,
77: one has to remember that actual time profile, and in particular a
78: two-dimensional distribution (image) of \v{C}erenkov light flash
79: from atmospheric gamma-ray showers, undergo a rapid change after an
80: increase in the observational level. This paper briefly describes
81: the results of a topological analysis of \v{C}erenkov light images
82: calculated at both conventional and desirable altitudes of 2.2 and 5
83: km above the sea, respectively. A discussion on major upgrades of
84: image topology at high altitude is also given.
85: \end{abstract}
86:
87: \section{Introduction}
88: The usual way to proceed with design studies for a future project,
89: at least in a field of VHE gamma-ray astronomy, is to perform full
90: scale simulations of detector response for various anticipated event
91: types. Apparently, gamma-ray-induced atmospheric showers represent a
92: sample of signal events, whereas cosmic-ray showers form a
93: dominating background (for review of \v{C}erenkov imaging technique
94: see \cite{tw04}). For a future low-threshold instrument a correct
95: tuning of detector design to optimize its response to gamma-ray
96: events becomes a most important issue, due to the fact that the
97: sensitivity of such a detector will be given by {\it angular and
98: energy resolution for signal events}. Both of these will finally
99: determine an efficiency of background rejection. Here we have
100: studied how a single parameter of detector design, particularly the
101: height of the observational site, may affect a topology of signal
102: events. Therefore, we have calculated the response of a ground-based
103: \v{C}erenkov light telescope of 30~m diameter, assuming an ideal
104: focal-plane detector. Comparative analysis of simulated events
105: helped us to understand what are the major differences in
106: parametrization of the \v{C}erenkov light flashes from gamma-ray
107: induced air showers, registered at two observational levels, i.e. 2
108: and 5 km above sea level. Considering the analysis results, there is
109: a discussion at the end of this paper as to which observational
110: level might be considered as more favorable for effective shower
111: imaging.
112:
113: \section{Simulations}
114: Numerous comparisons of a few shower simulation codes
115: %available in
116: %the H.E.S.S. collaboration,
117: have been recently performed by different groups.
118: %the Monte Carlo group,
119: They have revealed a rather good level of agreement in basic
120: parameters of \v{C}erenkov light emission in gamma-ray-induced air
121: showers over broad energy range, starting from 100~GeV and expanding
122: up to 20~TeV. Calculations presented here have been carried out
123: using one of those simulation codes, namely ALTAI code \cite{altai}.
124: This numerical code was extensively used for production of the
125: simulated data for the HEGRA system of five imaging atmospheric
126: \v{C}erenkov telescopes at La Palma \cite{ko99}.
127:
128: Shower simulations have been done for the standard continental
129: atmosphere (US standard atmosphere) \cite{elterman} for the
130: wavelength range of \v{C}erenkov light photons from 300 to 600~nm.
131: Absorption of \v{C}erenkov light photon in the atmosphere due to
132: Rayleigh and Mie scattering was modelled according to the data given
133: in \cite{elterman,valley}. The detector simulation procedure used
134: here accounted for all efficiencies involved in the process of the
135: \v{C}erenkov light propagation and registration \cite{ko99}. It
136: includes (i) mirror reflectivity; (ii) the acceptance of the funnels
137: placed in front of the photomultipliers (PMTs) (iii) the
138: photon-to-photoelectron conversion inside the PMTs (bi-alkali
139: photocathode).
140:
141: Shower simulations were made here in the so-called "batch" mode. A
142: shower propagating time, corresponding atmospheric depth, and a
143: number of emitted \v{C}erenkov photons were saved for each
144: multiple-scattering segment of all electron trajectories in a
145: shower. The actual segment size was chosen as small as 0.1~[$\rm
146: gr/cm^2$]. CPU time needed for simulation of such low energy
147: gamma-ray shower ($E_o$=10~GeV), using customary computers, is
148: short, and it is not an issue for any scheme's optimization. One
149: record for a single multiple-scattering segment was treated as an
150: individual "batch" of emitted \v{C}erenkov photons. At the second
151: step of this simulation procedure all recorded batches were restored
152: and finally used in the calculation of the response of a number of
153: \v{C}erenkov telescopes, situated at different atmospheric
154: altitudes. Such an approach provides an opportunity to use exactly
155: the same simulated showers for various telescope arrangements at
156: different observational levels. It should be noted that the
157: estimated statistical error for the parameters of the \v{C}erenkov
158: light emission given below is $\leq 10$\%.
159:
160: For the next generation of ground based \v{C}erenkov telescopes
161: %H.E.S.S. Phase II telescopes,
162: a dish of roughly 30 m diameter is foreseen.
163: %\cite{wh03}.
164: Issues around the design of such a big reflector are addressed in
165: \cite{wh01}. The simulation setup here consisted of 12 such
166: telescopes, which were arranged along one line at distances from 0
167: to 300 m from the shower axis. Note that the same showers were used
168: in calculations for each of these telescopes. It allows for a direct
169: comparison of the telescope's responses at different distances from
170: the shower axis, and the accurate study of fluctuations in
171: \v{C}erenkov light flashes at different shower impacts.
172:
173: \section{Results}
174:
175: Distribution of \v{C}erenkov light emission from atmospheric
176: showers can be characterized using a smooth function
177: \begin{equation}
178: \eta=\eta(t,\vec{r},\vec{\theta}), \label{aaa}
179: \end{equation}
180: which gives a mean number of photons per unit area arriving at the
181: observational level at time {\it t}, with space, $\vec{r}=\{x,y\}$,
182: and angular, $\vec{\theta}=\{\theta_x,\theta_y\}$, coordinates,
183: calculated with respect to the shower axis. The presentation
184: (\ref{aaa}) presumes averaging over a number of photons at any local
185: spot, because Monte Carlo simulations have provided the list of
186: individual photons with their coordinates. In the ideal case a
187: space-angular distribution of \v{C}erenkov photons in the
188: observational plane is simply a sum of $\delta$-functions
189: constructed for each individual photon.
190:
191: The lateral distribution of \v{C}erenkov photons at the
192: observational level, which is supposed to be perpendicular to the
193: shower axis, is given by
194: \begin{equation}
195: \rho(r) = \rho(|\vec{r}|)= \int_0^\infty \int_{2 \pi} \eta(t,
196: \vec{r}, \vec{\theta} ) dt ~d\Omega.
197: \end{equation}
198: The function $\rho(\vec{r})$ is the density of \v{C}erenkov light
199: (the number photons hitting the unit square at $\vec{r}$).
200:
201: In a similar way one can derive a temporal distribution of a
202: \v{C}erenkov light pulse, and a two-dimensional angular
203: distribution (image) of a \v{C}erenkov light flash
204: \begin{equation}
205: p(t) = \int_{A(\vec{r}_o)} \int_{\Omega_o} \eta(t, \vec{r},
206: \vec{\theta} ) rdr ~d\Omega_o,
207: \end{equation}
208: \begin{equation}
209: q(\vec{\theta})=\int_{A(\vec{r}_o)} \int_o^\infty \eta(t,
210: \vec{r}, \vec{\theta} ) rdr~ dt,
211: \end{equation}
212: where $A(\vec{r}_o)$ and $\Omega_o$ are the area of the reflector
213: and the angular camera size, respectively, of the telescope placed
214: at $\vec{r}=\vec{r}_o$.
215:
216: The function $\eta(t,\vec{r},\vec{\theta})$ can be well described
217: by a set of functions $\rho(r)$, $p(t)$, and $q(\vec{\theta})$,
218: given for a number of telescope locations,
219: $\vec{r}^{(i)}_o,~i=1,n$.
220:
221: \subsection{\v{C}erenkov light density}
222: \label{ss:1}
223:
224: It was emphasized in \cite{ap01}, that a substantial increase in
225: \v{C}erenkov light density at high altitudes in the atmosphere might
226: be very promising for a further reduction of the effective energy
227: threshold of a telescope array, which can be erected at a height of
228: about 5~km above the sea. One can see in Figure~\ref{fig:1} that for
229: a 10~GeV gamma-ray-induced atmospheric shower the density of
230: \v{C}erenkov light at 5~km above sea level, in a range of distances
231: of the telescope to the shower core limited by $r \rm \leq~100~m$,
232: is about a factor of 2.5 higher than the corresponding density at
233: 2~km altitude. At the same time {\it in a range of relatively large
234: impact distances, $r \rm \ge 125$~m, the \v{C}erenkov light density
235: remains the same at both observational heights.} (see
236: Figure~\ref{fig:1}).
237:
238: The atmospheric depth of the shower maximum can be estimated as
239: $X_{max}=t_o ln(E_o/E_c)$, where $t_o$ is the radiation length in
240: air ($t_o \simeq \rm 37~g/cm^2$), $E_o$ is the primary energy of
241: air-shower, and $E_c$ is the so-called critical energy ($E_c \simeq
242: \rm 80$~MeV). Thus for a 10~GeV $\gamma$-ray-induced air-shower the
243: atmospheric depth of its shower maximum is about 180~$\rm g/cm^2$. A
244: substantial fraction of \v{C}erenkov light photons emitted from the
245: shower maximum will be absorbed while traveling down to the
246: observational level. Between 30\% and 16\% (in the wavelength range
247: of 300-600 nm) for the heights of 2.2 and 5~km above sea level,
248: respectively. At the same time the \v{C}erenkov light pool shrinks
249: significantly at higher altitudes. The approximate radial size of
250: the light pool at 2.2~km is about 130~m (see Figure~\ref{fig:1}),
251: whereas at 5~km it might be roughly limited by $\simeq 90$~m. It
252: results in a corresponding increase of \v{C}erenkov light density by
253: approximately a factor of 2. This geometrical effect has a major
254: contribution on the increase of \v{C}erenkov light density at high
255: altitude as shown in Figure~\ref{fig:1}.
256:
257: \begin{figure}[t]
258: \centering
259: \includegraphics[height=7cm]{ldf.eps}
260: \caption{Density of \v{C}erenkov light, $\rho(r)~[\rm ph.-e./m^2]$,
261: in a 10~GeV gamma-ray-induced atmospheric shower at two
262: observational levels of 2 and 5 km above sea level, respectively.
263: The density was measured in photoelectrons $\rm [ph.-e./m^2]$. A
264: photon-to-photoelectron conversion efficiency of $\chi= 0.1$ was
265: assumed here, $\rho~ \rm [ph.-e./m^2]=\chi\cdot \rho~ \rm
266: [photon/m^2]$. \label{fig:1}}
267: \end{figure}
268:
269: For a large telescope of a 30~m diameter, a requirement of a minimum
270: number of 15~ph.-.e.\footnote[1]{This requirement of a minimum
271: number of ph.-e. for the telescope trigger is in fact not a generic
272: value, and it might be slightly different in certain circumstances,
273: but it does not affect the general discussion given here.} in the
274: \v{C}erenkov light flash, which is sufficient to trigger the
275: telescope, will limit the allowed range of impact distances for
276: 10~GeV gamma-ray showers to roughly $\rm r \leq 300~m$ (see
277: Figure~\ref{fig:1}). It corresponds to the same effective detection
278: area of $\rm S=3 \times 10^5~m^2$ at both observational heights. On
279: the other side, assuming that the \v{C}erenkov light density scales
280: with primary shower energy as $\rho\propto E^{1.1}$, one can roughly
281: estimate the minimal primary energy of the gamma-ray shower, which
282: has still sufficient amount of light at the density profile plateau
283: (r$<$125~m) and can still trigger the telescope. Simple calculations
284: yield, accordingly, a factor of 4 and 10 for 2 and 5~km altitudes,
285: respectively. It means that one can catch gamma-ray events of energy
286: $\geq 2.5$~GeV and $\geq 1$~GeV at 2 and 5~km observational height,
287: respectively, using the same telescope. It is important to mention
288: that all these extremely low-energy events will be concentrated
289: within a radius of roughly 100~m, which is determined by the actual
290: shape profile of a lateral distribution function of \v{C}erenkov
291: light. The drastic drop in photon density beyond 120~m will prevent
292: the detection of such gamma-ray showers at larger distances to the
293: shower axis. As a result the detection area for these events will be
294: a weak function of primary shower energy. Furthermore, the detection
295: area of high energy gamma-ray showers will be of the same size at
296: both observational heights.
297:
298: \begin{table}[t]
299: \centering \caption{Parameters of the fit in Eq.~\ref{eqn:1}. D
300: [$\rm gr/cm^2$] denotes an atmospheric depth at a given
301: observational level. R is the impact distance of the telescope to
302: the shower axis. \label{tab:1}}
303: \begin{tabular}{lllllll}
304: \hline\noalign{\smallskip}
305: H [km] & D [$\rm gr/cm^2$] & R [m] & C & $\alpha$ & $t_o$ & $\beta$ \\
306: \noalign{\smallskip}\hline\noalign{\smallskip}
307: 5 &500 & 50 & 2.595$\times 10^{-5}$ & 8.533 & 3.463 & 13.778\\
308: & & 100 & 2.650$\times 10^{-10}$ & 14.149 & 4.670 & 19.076\\
309: & & 150 & 2.424$\times 10^{-11}$ & 12.495 & 6.500 & 15.827\\
310: & & 200 & 4.824$\times 10^{-13}$ & 12.154 & 9.085 & 15.081\\
311: & & 250 & 7.352$\times 10^{-14}$ & 11.277 & 12.240 & 13.770\\
312: \hline
313: 2.2 &843 & 50 & 1.360$\times 10^{-6}$ & 7.689 & 5.824 & 19.295\\
314: & & 100 & 2.871$\times 10^{-15}$ & 17.936 & 6.620 & 32.634\\
315: & & 150 & 4.031$\times 10^{-17}$ & 17.907 & 8.157 & 26.966\\
316: & & 200 & 1.225$\times 10^{-18}$ & 17.551 & 9.988 & 23.094\\
317: & & 250 & 3.512$\times 10^{-20}$ & 17.264 & 12.368 & 21.476\\
318: \noalign{\smallskip}\hline
319: \end{tabular}
320: \end{table}
321:
322: \subsection{Time profile of \v{C}erenkov light flash}
323: The longitudinal extension of an atmospheric shower, and its
324: location in space with respect to the telescope, finally determine a
325: time profile of the \v{C}erenkov light flash. For a gamma-ray shower
326: of a certain primary energy, e.g. of 10~GeV, recorded at fixed
327: observational level (e.g. 2 or 5~km above sea level), the shape of
328: the \v{C}erenkov light flash basically depends only on the actual
329: distance of the telescope to the shower axis. The arrival time of
330: \v{C}erenkov photons onto reflector is $t=t_e+t_{\check{C}}$, where
331: $t_e$ and $t_{\check{C}}$ is accordingly a propagation time of
332: emitting electrons and photons, respectively. Electrons of an
333: energy, which is sufficient for emission of \v{C}erenkov light in
334: the atmosphere, are apparently moving faster than the emitted
335: photons. Thus at relatively small distances of the telescope to the
336: shower axis ($r\leq$120~m) \v{C}erenkov photons, emitted at later
337: stages of the shower development in the atmosphere, are actually
338: arriving earlier than the photons emitted at the beginning of shower
339: development. It swaps over at a large distance from the shower axis,
340: because of the rather long travel path in dense atmosphere for the
341: photons emitted by electrons out of a dying particle
342: cascade\footnote[2]{This is a well-known effect, sometimes called
343: the "sea-gull" effect, in the shape of \v{C}erenkov light pulses
344: (e.g. see \cite{kruger})}. Therefore, in general, the larger the
345: distance of the telescope to the shower axis, the broader the
346: corresponding \v{C}erenkov light pulse.
347:
348: Time pulses of \v{C}erenkov light flashes always show a very steep
349: rising edge, and a slow fall-off. They can be well fitted by
350: \begin{equation}
351: p(t)=C\cdot t^{\alpha}(1+(t/t_o)^{\beta}). \label{eqn:1}
352: \end{equation}
353: Parameters of the fits of time pulses simulated for two
354: observational levels and for a number of impact distances are
355: given in Table~\ref{tab:1}. The normalization condition was
356: $\int_o^{50ns}p(t)dt=1$. The contribution of \v{C}erenkov photons
357: delayed by longer times than 50~ns is negligible.
358:
359: \begin{figure}[t]
360: \centering
361: \includegraphics[height=7cm]{pm.eps}
362: \caption{Parameters of \v{C}erenkov light time pulses, $t_1$ (1),
363: $t_2$ (2), $t_3$ (3), calculated at two observational levels of 2
364: (solid curves) and 5~km (dashed curves) above sea level, as a
365: function of impact distance of the telescope to the shower axis R.
366: \label{fig:2}}
367: \end{figure}
368:
369: \begin{figure}[t]
370: \centering
371: \includegraphics[height=7cm]{pulse.eps}
372: \caption{Time profile of \v{C}erenkov light flashes simulated for
373: impact distance of 150~m and two observational levels of 2 (1) and
374: 5 (2) km above the sea. Primary energy of simulated air-showers is
375: 10~GeV. Label a.u. along Y-axis stays for the arbitrary units.
376: \label{fig:3}}
377: \end{figure}
378:
379: Shape of time pulse can be characterized by a few parameters, e.g.
380: $t_1=t_{30\%}-t_{10\%}$, $t_2=t_{50\%}-t_{10\%}$,
381: $t_3=t_{90\%}-t_{10\%}$, where $t_{10\%}$, $t_{30\%}$, $t_{50\%}$,
382: and $t_{90\%}$ give the time tags, which are defined as, e.g.
383: $\int_o^{t_{10\%}}p(t)dt=0.1$. Results of calculations are shown in
384: Figure~\ref{fig:2}. One can see that for a 500~$\rm [gr/cm^2]$
385: observational level time pulses are substantially broader at impact
386: distances beyond 100~m. Hereafter we are dealing with \v{C}erenkov
387: light images averaged over a sample of simulated 10~GeV gamma-ray
388: showers. For an impact distance of ca. 250~m, the time pulse of
389: \v{C}erenkov light flash recorded at high altitude in the atmosphere
390: will be a factor of 2 broader than for the same impact distance at a
391: conventional altitude of 2~km above sea level (see
392: Figure~\ref{fig:3}). Integration over the time pulse yields a total
393: number of \v{C}erenkov photons in a flash. Therefore, for a given
394: flux of night sky background light, {\it a signal-to-background
395: ratio might be correspondingly lower by factor of 2 for a high
396: altitude site}. For high energy gamma-ray showers ($E_o\geq$100~GeV)
397: \v{C}erenkov pulses recorded at 5~km above sea level might be as
398: long as 50~ns. Registration of these pulses will occur in the regime
399: highly dominated by night sky background.
400:
401: It is worth noting that, at the time of writing, there is no well
402: established altitude dependence of a flux of night sky background
403: light available. In general this parameter is considered to be very
404: specific for each individual observational site. Apparently the high
405: altitude sites provide substantially reduced attenuation and
406: consequently more starlight from the individual stars, which is in
407: fact a background for the \v{C}erenkov telescopes. However, the
408: effect of bright stars might be diminished simply by switching off
409: the high voltage for those camera pixels (PMTs) collecting direct
410: star light. Such procedure usually runs in automatic mode while
411: taking the observational runs. At the same time one might expect a
412: significant increase in flux of background light photons within the
413: UV wavelength range (200-300~nm). However, conventional imaging
414: cameras are not very sensitive in a wavelength range well below
415: 300~nm.
416:
417: Even though there are good reasons to believe that this flux will,
418: in fact, be much lower (on average) for high altitude sites, it
419: still needs to be measured at any chosen observational site.
420:
421: For the sake of thoroughness one should mention that the high
422: altitude sites will noticeable increase the probability that the
423: ionizing particles, such as atmospheric electrons and muons of low
424: energy cosmic rays, can directly hit the camera PMTs. It will lead
425: to some random increase in background light over the camera pixels.
426: However, dedicated calculations are needed in order to quantify this
427: additional component of the background, which fall unfortunately out
428: of the area of this paper.
429:
430: \subsection{Images}
431: A two-dimensional distribution of the \v{C}erenkov light intensity
432: in the telescope's focal plane (image), $q(\vec{\theta})$, can be
433: effectively used to derive detailed information about shower
434: orientation and shape. Phenomenology of \v{C}erenkov light images
435: was discussed in \cite{hillas}. For an ellipsoid-like image, the
436: orientation of its major axis constrains the shower orientation in
437: space with respect to the telescope optical axis. Images recorded at
438: relatively small impact distances from the shower axis (R$\le$100~m)
439: have circular shape, and an accurate determination of the major axis
440: is quite difficult. For impact distances beyond 100~m the image
441: ellipsoid has a well defined shape, and the ratio of its angular
442: size measured along a major axis to the corresponding angular size
443: of the image measured along a minor axis is ca. 2 and above. At the
444: same time at large impact distances (R$\gg$200~m) the total number
445: of \v{C}erenkov photons in an image becomes rather low and high
446: fluctuations prevent accurate measurement of the image orientation.
447: Those two effects finally constrain the range of optimum impact
448: distances for effective shower reconstruction. As mentioned above,
449: the advantage of the high altitude site is mainly associated with an
450: enhancement in \v{C}erenkov light density at small distances to the
451: shower axis (R$\leq$100~m). However, in this range of impact
452: distances \v{C}erenkov light images tend to have poorly determined
453: orientation.
454:
455: Average \v{C}erenkov light images for two observational heights of
456: 2 and 5~km, respectively, are shown in Figure~\ref{fig:4}. Note
457: that scales used along X- and Y-axis are not identical. Detailed
458: comparison of those images revealed two major differences in their
459: shape.
460:
461: \begin{figure}[t]
462: \centering
463: \includegraphics[height=5cm]{img_r50m.eps}\hspace*{10mm}
464: \includegraphics[height=5cm]{img_r100m.eps}\\ \vspace*{10mm}
465: \includegraphics[height=5cm]{img_r150m.eps}\hspace*{10mm}
466: \includegraphics[height=5cm]{img_r200m.eps}
467: \caption{Images of \v{C}erenkov light calculated at different
468: altitudes above sea level for a number of impact distances of the
469: telescope to the shower axis as indicated in the pictures. Images
470: were averaged over a sample of 10~GeV gamma-ray showers. The
471: contour plots were drawn starting from the position of maximum
472: intensity with the isoline increment of $ln2$. \label{fig:4}}
473: \end{figure}
474:
475:
476: \begin{enumerate}
477: \item Images of 10~GeV gamma-ray showers recorded at high altitude
478: above sea level are substantially larger in size (see
479: Figure~\ref{fig:4} and Table~\ref{tab:2}). This is mainly because
480: showers are located at a relatively smaller distance to the
481: telescope, than for a conventional observational site of 2~km above
482: the sea. Calculations show that the angular size of an image
483: measured along the major axis, the image length, increases
484: considerably faster with altitude than the angular size of image
485: measured along a minor axis, the image width. This can be easily
486: understood by comparing the ratio of the actual scale of a shower
487: longitudinal development over the distance of shower maximum to the
488: telescope, which is located at two altitudes of 2 and 5~km above the
489: sea, respectively. In a simplified toy model the length of the image
490: will scale with height of observation level above sea level, $H_o$,
491: as
492: \begin{equation} L\propto (H_{max}-H_o),~
493: R<<(H_{max}-H_o).\end{equation} \item Images recorded at high
494: altitude must have considerably larger displacement from the
495: center of the focal plane. Coming closer to the shower maximum
496: ($H_{max}$ is the height of the shower maximum), the shower will
497: be apparently seen in \v{C}erenkov light at larger angle
498: \begin{equation}
499: \Theta \propto \tan^{-1}(\frac{R}{H_{max}-H_o}) \end{equation}
500: with respect to the optical axis.
501: % (see also \cite{memo}).
502: \end{enumerate}
503:
504: \begin{table}[t]
505: \centering \caption{Area $A,~[deg^2]$ and effective size,
506: $r_o=\sqrt{ab}~[deg]$, (in this Table both values are given in a
507: format of $A/r_o$) of \v{C}erenkov light images calculated for two
508: observational levels of 2 and 5~ km, and for a few impact
509: distances.\label{tab:2}}
510: \begin{tabular}{llcccc}
511: \hline\noalign{\smallskip}
512: H [km] & D [$\rm gr/cm^2$] & R [m] = 50 & 100 & 150 & 200 \\
513: \noalign{\smallskip}\hline\noalign{\smallskip}
514: 2.2 & 843 & 0.16/0.22 & 0.24/0.27 & 0.31/0.32 & 0.44/0.37 \\
515: 5 & 500 & 0.40/0.36 & 0.54/0.41 & 0.87/0.53 & 1.27/0.64 \\
516: \noalign{\smallskip}\hline
517: \end{tabular}
518: \end{table}
519:
520: One can see in Figure~\ref{fig:1} that the density of \v{C}erenkov
521: photons from a 10~GeV gamma-ray shower is approximately the same at
522: both observational heights for impact distances above 150~m from the
523: shower axis. Thus, at large impact distances, the difference in
524: image shape mentioned above is of purely geometrical origin, which
525: is independent of the image size (the total number of photons in the
526: image). At first glance, large images at high altitudes might offer
527: better resolution for a camera of crude pixellation. However, for a
528: 10~GeV gamma-ray shower images will always be significantly affected
529: by contamination of background light and reduction of background
530: light per camera pixel. Large images recorded at high altitudes
531: yield considerably lower \v{C}erenkov light density per 1~$str$ for
532: fixed image's size. {\it For the flux of night sky photons, as
533: measured at conventional observational level of 2~km above the sea,
534: the image of a 10~GeV gamma-ray shower recorded at high altitudes
535: will have higher contamination of background photons per camera
536: pixel of any size}.
537:
538: As mentioned above, the images recorded at high altitudes must have
539: large displacements from the camera center. This issue stringently
540: constrains the design of the camera for the telescope placed at high
541: altitudes, in particularly the field of view has to be larger. {\it
542: Cameras of a narrow field of view will be drastically limited in
543: their ability to detect gamma-rays at high energies}.
544:
545:
546: \begin{figure}[htbp]
547: \centering (a) \includegraphics[height=3cm]{b1.eps} \\
548: \vspace*{2mm} (b) \includegraphics[height=3cm]{b5.eps}\\
549: \vspace*{2mm} (c) \includegraphics[height=3cm]{b8.eps}\\
550: \vspace*{2mm} (d)
551: \includegraphics[height=3cm]{b10.eps}\\\vspace*{10mm}
552: \hspace*{4mm} \includegraphics[height=3cm]{b.eps} \caption{Average
553: images of \v{C}erenkov light from a 10~GeV gamma-ray shower,
554: calculated at an observation level of 2~km above sea level, and
555: for an impact distance of 150~m from the shower axis. An
556: additional selection on arrival time of \v{C}erenkov photons was
557: applied. Corresponding time windows were 7.25-7.5~ns (a);
558: 8.25-8.5~ns (b); 9-9.5~ns (c); 10-12~ns (d). Each of (a)-(d) plots
559: contains approximately the same number of photoelectrons. The
560: image shown in the lower panel was generated without time
561: selection. \label{fig:5}}
562: \end{figure}
563:
564:
565: \subsection{Time-dependent imaging}
566: One can try to suppress contamination of night sky background light
567: in an image by using a very narrow time gate. This may be tuned
568: exactly, for example, to a rising edge, maximum, or tail of a time
569: pulse. However, it leads to a trade off between losing a substantial
570: fraction of \v{C}erenkov photons and on the other hand a severe
571: reduction of night sky background. This approach is illustrated by
572: the images shown in Figure~\ref{fig:5}. Photons emitted at the very
573: beginning of the shower development in the atmosphere, which are
574: mainly arriving at the front of the time pulse of a flash, form the
575: so-called image "conk"\footnote[3]{Non-standard definition of a
576: strongly elongated part of the comet like image, which has a
577: relatively high photoelectron density.}. The image of shower
578: electrons, which propagate further into the atmosphere, shifts
579: further away from the center of telescope's focal plane. Multiple
580: scattering of low energy cascade electrons at later stages of the
581: shower development becomes very important and it results in a very
582: broad image as shown in Figure~\ref{fig:5}. One can see also in
583: Figure~\ref{fig:5} that \v{C}erenkov light {\it images for such
584: narrow time windows are in fact rather symmetric in shape, and they
585: are not so good for reconstruction of image orientation}. At the
586: same time, an image which contains all registered \v{C}erenkov
587: photons ($\rm 0~ns~<~t~<~14~ns$), has a regular ellipsoid-like
588: shape. Note that photons, which are significantly delayed with
589: respect to the front of time pulse, will be hitting the outer edge
590: of the image, whereas background light photons are apparently
591: dominant.
592:
593: Time-dependent imaging seems to be a rather promising approach in
594: improvement of gamma/hadron separation of extremely low energy
595: events. Despite that {\it present analysis does not reveal an
596: evident improvement using time-dependent imaging}, it could be
597: perhaps very effective in analysis based on centroid positions in a
598: few triggered telescopes in an array.
599:
600: \section{Conclusions}
601:
602: In this paper we have attempted to perform a comparative analysis of
603: \v{C}erenkov light images simulated for two observational levels of
604: 2 and 5~km above sea level, respectively. We used as the basis of
605: our calculations a 30~m diameter telescope, which is a conceivable
606: size for a future \v{C}erenkov telescope project. The system aspect
607: was not discussed here, but all conclusions are obviously relevant
608: to any telescope of a possible future array. One has to mention that
609: for a future detector, approaching a very low energy threshold of
610: about 10~GeV or even below, a contamination of night sky background
611: in the registered shower images is in fact a very important issue.
612: It might finally constrain the choice of the observational site for
613: such low threshold arrays of \v{C}erenkov telescopes.
614:
615: Results reported here generally confirm that the observational site
616: at higher altitude provides substantially higher \v{C}erenkov light
617: density. However this enhancement occurs only within the area
618: limited by roughly 100~m from shower axis. Therefore all recorded
619: gamma-ray showers well below 10~GeV (here we assume that the
620: telescope has in fact sufficient area of the reflector, see
621: discussion in Section~\ref{ss:1}) will concentrate in a region of
622: relatively small impact distances. Those images are not so clearly
623: elongated in shape, which makes reconstruction of the image's
624: orientation rather difficult.
625:
626: Flashes of \v{C}erenkov light from 10~GeV gamma-ray showers have
627: broader time pulses for the impacts beyond 100~m at higher altitude.
628: Corresponding images are considerably broader as well. Both of these
629: two effects might apparently substantially increase the background
630: light contamination in an image. Images recorded at high altitude
631: must be further displaced from the center of telescope's focal
632: plane, than for conventional altitude of ca. 2~km above sea level.
633: They also, most probably, require larger field of view.
634:
635: The time-depending imaging is a very promising approach in further
636: development of advanced analysis for observation of low energy
637: gamma-ray showers, but it might be not very effective in resolving
638: the problem of dominating night sky background light in the
639: recorded images of such low energy gamma-rays.
640:
641: One can briefly conclude that an observational site at high
642: altitude might provide {\it further modest reduction of the energy
643: threshold} of a future detector, even though the shape of time
644: pulses and in particular the topology of the two-dimensional
645: angular distribution of \v{C}erenkov light flashes recorded at
646: extremely high altitudes are palpably {\it less preferable} for
647: imaging of gamma-ray showers above 10~GeV.
648:
649: We hope that the results presented here may help to increase the
650: understanding of changes in topology of \v{C}erenkov light images
651: after an increase in the observational level from its conventional
652: height of ca. 2~km up to 5~km above sea level.
653:
654: Ultimately, the choice of a site for the next generation of
655: ground-based imaging atmospheric \v{C}erenkov detector, which is
656: widely believed to be a system of 30~m class telescopes, will depend
657: on both scientific and political issues relating to funding,
658: international collaboration etc. The move to lower energy threshold
659: is likely to remain a significant drive for the science. However,
660: one has to consider all trade-offs, i.e. reasonable altitude, low
661: level of night sky background, need for the robotic telescopes etc,
662: in selecting candidate sites for such a detector to optimize the
663: scientific goals.
664:
665: \section*{Acknowledgements} I would like to thank the referees, who
666: remain anonymous, for the invaluable comments and suggestions, which
667: have improved the quality of this paper.
668:
669: \vspace*{-2mm}
670: \begin{thebibliography}{99.}
671:
672: \bibitem{tw04} T. Weekes: "Very High Energy Gamma-Ray Astronomy",
673: IOP Publishing Ltd, Bristol and Philadelphia, Series in Astronomy
674: and Astrophysics (2003)
675:
676: \bibitem{altai} A. Konopelko, A. Plyahseshnikov: Nuclear Instruments
677: \& Methods in Physics Research A \textbf{450}, 419 (2000)
678:
679: \bibitem{ko99} A. Konopelko, et al. (HEGRA Collaboration):
680: Astroparticle Physics \textbf{10}, Issue 4, 275 (1999)
681:
682: \bibitem{elterman} L. Eltermann: Air Force Cambridge Res. Lab. Ref.
683: 40, AFC RL-68-153 (1968)
684:
685: \bibitem{valley} S.L. Valley: Handbook of Geophysics and Space
686: Environments, McGraw-Hill Book Company (1965)
687: %\bibitem{wh03} W. Hofmann, On the dish geometry of a large
688: %telescope, Internal Note 03/06 (2003)
689:
690: \bibitem{wh01} W. Hofmann: J. Phys. G: Nuclear and Particle Physics,
691: {\bf 27}, Issue 4, 933 (2001)
692:
693: \bibitem{ap01} F. Aharonian, A. Konopelko, H.J. V\"{o}lk, H.
694: Quintana: Astroparticle Physics \textbf{15}, Issue 4, 335 (2001)
695: %\bibitem{memo} A. Konopelko, H.E.S.S. Internal Note 03/03,
696: %November (2003)
697:
698: \bibitem{kruger} A. Konopelko: Proc. Kruger Park Workshop on TeV
699: $\gamma$-Ray Astrophysics "Towards a Major Atmospheric \v{C}erenkov
700: Detector -V", (ed. O.C. de Jager), Kruger Park, Source Africa,
701: August 8-11, 208 (1997)
702:
703: \bibitem{hillas} A.M. Hillas: Proc. "TeV Gamma-ray Astrophysics"
704: (Heidelberg) (Space Sci. Rev. 75) ed H.J. V\"{o}lk and F.A.
705: Aharonian (Dordrecht: Kluwer), 17 (1996)
706:
707: \end{thebibliography}
708: \end{document}
709: