EPJ Web of Conferences 89, 03009 (2015)
DOI: 10.1051/epjconf/20158903009
c Owned by the authors, published by EDP Sciences, 2015
The impact of clouds on image parameters in IACT at very high energies
Dorota Sobczyńskaa and Włodek Bednarek
University of Łódź, Department of Astrophysics, Pomorska 149/153, 90-236 Łódź, Poland
Abstract. The effective observation time with the Cherenkov telescopes arrays is limited to clear sky conditions due to
considerable absorption of Cherenkov light by the possible presence of clouds. However below the cloud altitude the primary
particles with high energies can still produce enough Cherenkov photons to allow the detection by the large telescopes. In this
paper, using the standard CORSIKA code, we investigate the changes of image parameters due to the absorption of Cherenkov
radiation by the cloud (for γ -ray and proton showers with various energies – from 2 TeV to 100 TeV and from 10 TeV to 200 TeV,
respectively). We consider the clouds with different transmissions located at various altitudes above the ground level (between
8 km and 3 km). We show that, for both simulated primary particles at fixed energy, the WIDTH and the DIST distributions are
shifted towards larger values in the presence of clouds in comparison to the clear sky simulations. This shift decreases with the
cloud altitude. The LENGTH distributions are shifted towards smaller values for images of primary γ -rays, while for primary
protons this shift is not observed. We conclude that the large Cherenkov telescopes with large camera FOV could be used for
observation of γ -ray showers with high energies in the presence of clouds.
1. Introduction
The imaging air Cherenkov technique (IACT) has been
successfully used to record γ -rays from cosmic sources.
The two dimensional angular distribution of the Cherenkov
light produced by the shower (so called shower image) is
measured using IACT. Parameters which describe shower
images were proposed by Hillas in 1985 [1]. He also
showed a method of γ /hadron separation based on those
parameters.
The observations using the Cherenkov telescopes are
carried out only during nights because the amount of
Cherenkov light created by the shower is relatively small.
The presence of clouds disturbs observations due to the
fact that Cherenkov photons are absorbed and scattered by
molecules of clouds. As a result, the amount of light can be
too small to trigger a telescope. Additionally, the recorded
images can be distorted. This may lead to deterioration
in the sensitivity of the telescope or telescope arrays.
Therefore the effective time of observation is limited to
clear sky conditions. Currently operating experiments like
HESS [2], MAGIC [3] and VERITAS [4] monitor the
atmosphere during data taking.
The influence of the atmospheric parameters on the
IAC technique has been already studied and presented in
[5]. The reconstructed energy of the shower and estimated
fluxes of primary particles (from the HESS data) can be
corrected for the presence of low-level aerosols [6]. The
data taken by MAGIC during nights with additional partial
absorption (caused by aerosols at a relatively low level) are
also corrected [7, 8]. However the applied method is based
on the correction of the image parameter called SIZE. It
has been shown in [9, 10], that also other Hillas parameters
can be affected by fully opaque clouds.
In this paper we investigate the influence of cloud
parameters (its altitude and transmission) on the Hillas
a
e-mail: [email protected]
parameters, which describe the shape and the position of
the image. We studied the problem for the γ -ray and proton
initiated showers with different fixed energies. Based on
Monte Carlo (MC) simulations, we show DIST, WIDTH
and LENGTH distributions, which are expected in the
presence of various clouds, for a detector area of 225 m2 .
We show that the changes of the image parameters are
similar for primary γ -rays and protons. Therefore the γ ray selection, based on the presented image parameters,
should be possible for the data collected under cloud
cover. We conclude that large mirror Cherenkov telescopes
with large field of view (FOV) camera should allow
for detection of very high energy γ -ray showers. Large
mirrors are needed because of significant Cherenkov light
absorption by molecules of clouds. Cameras with large
FOV are required because the images may be significantly
shifted outwards the camera center.
2. Monte Carlo simulations
We have used the CORSIKA code version 6.99 [12])
in order to simulate the shower development in the
atmosphere. Cherenkov photons (created by charged
particles from the shower) were absorbed in processes
of the Rayleigh and Mie scattering in the atmosphere
according to the Sokolsky formula [13]. Simulations have
been performed for vertical showers at the MAGIC site
[3] (the altitude of 2.2 km a.s.l.). To determine the two
dimensional angular distribution of the Cherenkov light,
we simulated a telescope with a mirror area of 225 m2
for eight different impact parameters (between 22.5 m
and 322.5 m). The histograms of angular distributions
were obtained by counting the Cherenkov photons in the
ranges −1◦ and 14◦ (in the x axis) and −7.5◦ to 7.5◦ (in
the y axis). The typical pixel size in operating IACT is
approximately 0.1◦ . This value was used as the width bin
of the histogram.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution,
and reproduction in any medium, provided the original work is properly cited.
Article available at http://www.epj-conferences.org or http://dx.doi.org/10.1051/epjconf/20158903009
EPJ Web of Conferences
all
7km
12000
45000
40000
35000
30000
25000
20000
15000
10000
5000
0
3
10000
8000
6000
4000
2000
0
2
1
y [d 0
eg.] -1
3
-2
-3 0
1
2
3
x [deg.]
4
5
6
6km
-2
-3 0
1
2
4
3
x [deg.]
-3 0
1
2
4
3
x [deg.]
5
6
5
6
5km
6000
2000
1800
1600
1400
1200
1000
800
600
400
200
0
5000
4000
3000
2000
1000
0
3
2
1
y [d 0
eg.] -1
2
1
y [d 0
eg.] -1
3
-2
-3 0
1
2
4
3
x [deg.]
5
6
2
1
y [d 0
eg.] -1
-2
Figure 1. The expected mean images of showers initiated by
20 TeV γ -rays with the impact parameter 112.5 m for different
altitudes of clouds. Note the different values on the z-axes.
The simulations have been performed without the
presence of any clouds and with clouds at the different
altitudes equal to 10, 7, 6 and 5 km a.s.l.. Those values
have been chosen because they correspond to the heights,
which are above and below the shower maximum in the
investigated energy range [9]. For each cloud altitude we
applied different transmission of the cloud (T), starting
from T=0 (fully opaque) up to T=1 (transparent) with
a step size of 0.2. Cherenkov photons, produced in the
shower above the cloud level, were partially absorbed
(with the probability equal to the cloud transmission).
We have simulated γ -ray showers with energies of 2, 5,
10, 20, 50 TeV and 100 TeV. The proton initiated showers
have been simulated with energies of 10, 20, 50, 100 and
200 TeV. The number of simulated events for each cloud
parameter was 1000 apart from γ -rays at 100 TeV and
protons at 200 TeV (500 events).
3. Results and discussion
The average images of the 20 TeV primary γ -rays at
impact parameter 112.5 m are presented in Fig. 1 for the
case of clear sky (marked by “all” in Fig. 1) and fully
opaque clouds at 7, 6 and 5 km a.s.l.. In the presence
of clouds the images contain less Cherenkov light and
they are less concentrated in comparison to clear sky
simulations. Images are also shifted away from the camera
center (located at the position of (0◦ , 0◦ )) due to the fact
that those Cherenkov photons which have been absorbed
by the clouds are produced higher in the atmosphere.
There are two important effects in the higher layers of the
atmosphere. First, the Cherenkov angle is smaller. Second,
the angular distribution of charged particles is narrower. As
a result, we expect that Cherenkov photons with very small
angles are not recorded on the camera for fully opaque
clouds. Therefore a part of the image close to the camera
center is missing (so called the geometrical effect). Finally,
showers can not be detected by cameras with small FOV.
The Hillas parameters should change in the presence of
clouds in comparison to clear sky conditions.
The impact of a cloud cover on the shower images
depends on the impact parameter of the shower because
the density of Cherenkov light is correlated with the height
of production of Cherenkov photons in the γ -ray initiated
shower [11]. The Hillas parameters called DIST, WIDTH
and LENGTH describe the position and the shape of the
image. They were calculated in few steps after the MC
simulations. First, we added the night sky background
(NSB) (with the level measured on La Palma [14]) to the
angular distributions of the Cherenkov light obtained from
MC simulations. Next, the images were cleaned. In the
calculations we neglected all pixels with a signal below
30 photons. Afterwards, image parameters were calculated
for all eight simulated impact parameters. Finally, the
distribution of the Hillas parameters was interpolated in the
range of impact parameters between 32.5 m and 322.5 m
for each simulation. All distributions presented below were
obtained for images with a total signal (SIZE) above
200 photons.
In Fig. 2 we show the dependencies of the Hillas
parameters on the transparency of the cloud located at
a height equal to 6 km a.s.l.. The DIST, WIDTH and
LENGTH distributions are presented as the top, middle
and bottom panel of figures, respectively. Four columns
of figures demonstrate different primary particle energy
from the left to right: 5 TeV γ -rays, 50 TeV γ -rays, 10 TeV
proton and 100 TeV proton. Protons and γ -rays with low
and high energies are shown in order to indicate possible
differences in the impact of cloudiness on the expected
image parameters.
The distributions of the DIST parameter of the γ -ray
and proton initiated showers become wider. They are more
shifted towards larger angles for lower cloud transmissions
(see Figs. 2a–d). At low energy (both primary particles),
the shift of the DIST distributions is significantly larger
for the fully opaque clouds due to the geometrical effect.
The distributions of the WIDTH parameter become
wider with decreasing transmission of the cloud for all
simulated primary particles and energies (see Figs. 2e–h).
The shift of the WIDTH distribution of the γ -ray and
proton showers towards larger values is expected for higher
energies. For lower energies a similar shift is obtained only
for the cloud transparency of 0 (which is due to the same
geometrical effect).
The distributions of the LENGTH parameter of the γ ray and proton images are shifted towards larger angles
for lower cloud transmission, except fully opaque clouds
(see Figs. 2i–l)). In this case, the LENGTH distributions
of primary γ -rays are significantly shifted towards lower
angles in comparison to the clear sky simulations due
to geometrical effects. This shift is not expected for
the primary protons (Figs. 2k,l). Similar changes of
the distributions of Hillas parameters with the cloud
transmission were obtained also for the cloud altitude of
7 km and 5 km. In the case of the cloud located at 10 km
a.s.l., the expected distributions of the image parameters
are very similar to the clear sky simulations for all
investigated transparencies.
We present dependencies of the Hillas parameters on
the cloud altitude for the fully opaque cloud (see Fig. 3).
The cloud transmission of 0 was chosen for demonstration
purposes because it shows an extreme case of the sky
conditions. Black and solid lines in all figures are the
results of the clear sky simulations. Panels and columns of
this Figure correspond to those in Fig. 2. The distributions
03009-p.2
AtmoHEAD 2014
T=0.6
5000
number of entries
4000
3000
2000
3000
2000
1000
0
0
1
2
3
4
5
o
DIST [
×10
6
7
2
3
4
5
o
DIST [
16
T=1.
e)
22
1
]
6
×10
18
number of entries
T=0.
14
12
10
8
6
2
3
4
5
o
×10
6
7
0.4
0.5
0.6
o
WIDTH [
×10
0.7
0.8
0.9
T=0.2
T=0.
4
0.2
0.3
0.4
0.5
0.6
×10
T=0.8
×10
g)
T=0.4
7000
T=0.2
T=0.4
T=0.4
0.9
number of entries
4
7
8
T=0.8
T=0.6
T=0.4
8000
6000
5000
4000
3000
T=0.2
T=0.
6000
4000
2000
0
1
0.1
0.2
0.3
0.4
]
0.5
0.6
0.8
0.9
0
1
×10
T=0.8
0.4
0.5
0.6
0.7
o
0.8
0.9
1
]
3
l)
T=0.8
8000
T=0.6
T=0.6
T=0.4
7000
T=0.2
T=0.2
6000
4000
0.3
T=1.
T=1.
k)
T=0.4
T=0.
0.2
WIDTH [
7000
6000
0.1
]
3
8000
j)
0.7
o
WIDTH [
T=0.2
8000
T=0.
6
]
h)
10000
T=0.
×10
T=0.6
6
5
o
T=1.
T=0.8
8000
3
T=0.8
10000
T=0.6
T=0.2
8
0.8
T=1.
T=1.
10
0.7
o
WIDTH [
i)
4
1000
0.1
]
6
12
3
3
T=0.6
6
0
2
DIST [
3
9000
8
1
1
]
T=0.6
number of entries
0.3
2000
T=1.
T=0.8
2
0.2
3000
0
0
8
2000
0.1
T=0.
1000
1
T=1.
f)
10
2
number of entries
2000
0
0
8
4
0
3000
DIST [
12
T=0.2
16
7
T=0.4
T=0.4
4000
6
14
T=0.6
6
T=0.2
4000
T=0.
]
T=0.8
20
T=0.4
T=0.2
1000
0
0
8
d)
T=0.6
T=0.4
T=0.
4000
T=0.8
5000
T=0.6
5000
3
T=1.
T=0.8
T=0.
number of entries
number of entries
5000
×10
c)
6000
T=0.2
T=0.
3
T=1.
b)
T=0.4
T=0.2
1000
number of entries
×10
T=0.8
T=0.6
T=0.4
6000
3
T=1.
6000
T=0.8
number of entries
×10
T=1.
a)
number of entries
3
number of entries
×10
7000
number of entries
8000
5000
4000
3000
2000
6000
T=0.
5000
4000
3000
2000
2000
2
1000
0
0.5
1
1.5
2
o
LENGTH [
0
2.5
0.5
1
]
1.5
2
o
LENGTH [
1000
2.5
0.5
1
]
1.5
2
o
LENGTH [
0
2.5
0.5
1
]
1.5
2
o
LENGTH [
2.5
]
Figure 2. Distributions of the Hillas parameters for the cloud at 6 km a.s.l.. Different colors of smooth histograms correspond to various
simulated transparencies (from 1 to 0). Primary γ -rays with energies of 5 TeV and 50 TeV are shown in the first and second column,
respectively. Primary proton with energies 10 TeV and 100 TeV are presented in the third and last column, respectively.
4000
3000
4000
3000
2000
1000
3
4
5
DIST [
×10
o
6
7
8
0
0
9
e)
20
16
h=10km
3
4
5
×10
o
6
7
16
number of entries
14
12
10
8
6
8
2
3
4
5
×10
o
6
7
8
9000
h=7km
8000
h=6km
0.3
0.4
0.5
0.6
WIDTH [
×10
o
0.7
0.8
0.9
h=5km
8
6
4
0
1
0.2
0.3
0.4
0.5
0.6
WIDTH [
×10
i)
T=1
h=10km
h=10km
8000
h=6km
h=5km
6000
4000
×10
g)
1
1.5
LENGTH [
2
o
]
2.5
o
0.7
0.8
0.9
5
o
6
7
8
9
]
T=1
h)
h=10km
h=7km
h=6km
h=6km
8000
h=5km
5000
4000
3000
0
1
h=5km
6000
4000
2000
0.1
0.2
0.3
0.4
]
0.5
0.6
WIDTH [
×10
h=7km
o
0.7
0.8
0.9
1
1.5
LENGTH [
2
o
×10
T=1
k)
8000
h=10km
h=7km
h=5km
5000
4000
3000
2000
2.5
0
]
0.2
0.3
0.4
0.5
0.6
WIDTH [
7000
h=6km
4000
0.1
]
6000
h=5km
0.5
0
1
3
7000
j)
6000
0
4
h=7km
o
0.7
0.8
0.9
1
]
3
T=1
l)
h=10km
h=7km
h=6km
6000
h=5km
5000
4000
3000
2000
1000
0.5
3
3
10000
h=10km
2000
2000
2
DIST [
h=6km
number of entries
8000
1
]
6000
3
T=1
h=7km
2000
1000
0.1
]
3
10000
3000
3
7000
number of entries
0.2
4000
T=1
h=10km
2
0.1
h=5km
0
0
9
2000
2
number of entries
1
T=1
10
h=6km
1000
0
0
9
4
0
2000
DIST [
12
h=5km
3000
]
f)
14
h=6km
4000
6
h=7km
18
number of entries
2
DIST [
T=1
22
1
]
6
h=5km
1000
number of entries
1000
h=7km
5000
h=6km
5000
d)
h=10km
h=7km
h=5km
3
T=1
6000
h=10km
6000
2000
2
c)
number of entries
number of entries
5000
1
×10
T=1
h=6km
5000
h=5km
0
0
3
h=7km
h=6km
6000
×10
7000
b)
h=10km
6000
h=7km
7000
3
T=1
h=10km
number of entries
×10
T=1
a)
number of entries
3
number of entries
×10
8000
number of entries
9000
1000
0.5
1
1.5
LENGTH [
2
o
]
2.5
0
0.5
1
1.5
LENGTH [
2
o
2.5
]
Figure 3. Distributions of the Hillas parameters for the clear sky and fully opaque clouds. Different colors of smooth histograms
correspond to various simulated cloud altitudes. Primary γ -rays with energies of 5 TeV and 50 TeV are shown in the first and second
column, respectively. Primary proton with energies 10 TeV and 100 TeV are presented in the third and last column, respectively.
of the DIST parameter of the γ -ray and proton induced
showers (Figs. 3a–d) become wider. They are shifted
towards larger angles for lower cloud altitude, except for
protons with low energy (Fig. 3c). In this case, shifts of the
DIST distribution are similar for cloud altitudes of 5 and
6 km a.s.l.. The WIDTH distributions also become wider
with decreasing altitude of the cloud (see Figs. 3e–h). The
expected shift (towards larges angles) is smaller at lower
energies than at higher energies for both simulated primary
particles. As expected, the changes of DIST and WIDTH
distributions are larger for lower altitudes of fully opaque
clouds. There are differences between the influences of
clouds (with T = 0) on the LENGTH distributions of the
γ -ray and proton initiated showers (see Figs. 3i–l). The
distributions of the LENGTH parameter of the γ -rays
are significantly shifted towards smaller angles for lower
altitudes of clouds (Figs. 3i,j). This shift was not obtained
for primary protons at low energy (Fig. 3k), while in the
03009-p.3
EPJ Web of Conferences
high energy range a small shift towards larger angles with
decreasing cloud altitude is observed (Fig. 3l).
We have checked the influence of the altitude of clouds
(with transmissions larger than 0) on the distributions
of the Hillas parameters. The changes of the Hillas
parameters due to cloudiness are similar for images of
the primary γ -rays and proton initiated showers. The
distributions of DIST and LENGTH are shifted towards
larger values for lower cloud altitude. The changes of the
WIDTH parameter depend on the simulated energy region.
The WIDTH distributions are shifted in the same direction
for high energies in contrast to low energies (shifts to the
smaller angles were obtained) for lower cloud altitude. We
have checked that, the differences between low and high
energies are correlated with the position of clouds with
respect to the shower maximum.
The γ /hadron separation is based on the differences
between the distributions of image parameters, but the
method is much more sensitive on the WIDTH than on the
LENGTH. The impact of the cloudiness on the WIDTH
distributions is similar for γ -rays and proton induced
showers in all simulated cases. Therefore, we expect, that
γ -ray events can be selected from the hadronic background
at very high energies even in the presence of clouds.
We have shown in [10], that for camera with FOV
limited to 8◦ the γ /hadron separation is still possible. The
Middle Size Telescope (MST) equipped with such cameras
will be built in the project CTA [15]. They might be used
for observations during cloud presence.
The influence of the cloud presence on an expected
trigger rate, a collection area, an angular resolution, an
energy resolution and a sensitivity curve will be considered
in the future studies for a particular telescope or IACT
system.
4. Conclusions
This work was supported by Polish Grant from Narodowe
Centrum Nauki 2011/01/M/ST9/01891.
The presence of clouds influences the expected image of
the shower because Cherenkov light is absorbed by the
particles forming the clouds. As the result, images contain
less light. They are shifted in the camera plane towards
larger angles. Therefore, IACT equipped with large mirror
(to keep reasonable high trigger rate) and with the cameras
with large FOV (to record shifted image) are needed to
observe γ -rays sources when the sky is covered by clouds.
As expected, the larger changes of the expected
DIST, WIDTH and LENGTH distributions were obtained
for the lower altitude and lower transmission of the
cloud. The presence of clouds (above 10 km a.s.l.) almost
do not influence the image parameters of the primary
γ -ray and proton initiated showers at high energies. This
altitude of clouds is above the shower maximum for
all simulated primary energies and particles. For clouds
at lower altitudes, the changes of the image parameters
can be explained by two effects: decreasing the image
concentrations (due to additional absorption) and the
geometrical effect. The second effect strongly depends on
the differences between altitudes of the shower maximum
and the cloud. The stronger impact of the cloudiness on
the image parameters is obtained for altitudes of the clouds
close and below the shower maximum.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
03009-p.4
Hillas A.M., Proc. 19th ICRC vol 3, 445 (1985)
Aharonian F. et al, A&A 457, 899 (2004)
Aleksić J. et al, Astropart. Phys. 35, 435 (2012)
Weekes T.C. et al, Astropart. Phys. 17, 221 (2002)
Bernlöhr K., Astropart. Phys. 12, 255 (2000)
Nolan S.J. et al, Astropart. Phys. 34, 304 (2010)
Dorner D. et al, A&A 493, 721 (2009)
Fruck C. et al, 33th ICRC, [arXiv:1403.3591]
(2013)
Sobczynska D. and Bednarek W., 33th ICRC ID0335 (2013)
Sobczynska D. and Bednarek W., accepted in J. Phys.
G (2014)
Sobczynska D., J. Phys. G 36, 045201 (2009)
Heck D. et al, Technical Report FZKA 6019
(Forschungszentrum Karlsruhe, 1998)
Sokolsky P., Introduction to Ultrahigh Energy
Cosmic Ray Physics (Addison-Wesley, 1989)
Mirzoyan R. and Lorenz E., MPI-PhE/94-35
(preprint, Munich, 1994)
Actis et al, Exper. Astron. 32, 193 (2011)