Measuring Light Pollution with Fish= eye Lens Imagery from A Moving Boat – A Proof of Concept
Andrea= s Jechow1,2,*, Zoltán Koll&aacut= e;th3, Amit Lerner4, Andreas Hänel5, Nadav Shashar6, Franz Hölker1, Christopher C.M. Kyba<= span lang=3DEN-US style=3D'mso-bidi-font-size:13.0pt'>1,2
[1]=
1Leibniz
2GFZ German R=
esearch
Centre for Geoscience, Potsdam, 14473, Germany
3University of
West Hungary, Savaria Campus, Szombathely, 9700, Hungary
4National Institute of Oceanography,
Israel Oceanographic and Limnological Research, Haifa, 31080, Israel
5=
Museum am
Schölerberg, Klaus-Strick-Weg 10, Osnabrück, 49082, Germany
6Ben-Gurion University of the Nege=
v,
Eilat Campus, Beer-Sheva, 84105, Israel
Received
05 December 2016, Accepted 08 February 2017
Abstract
Near
all-sky imaging photometry was performed from a boat on the Gulf of Aqaba to
measure the night sky brightness in a coastal environment. The boat was not
anchored, and therefore drifted and rocked. The camera was mounted on a tri=
pod
without any inertia/motion stabilization. A commercial digital single lens
reflex (DSLR) camera and fisheye lens were used with ISO setting of 6400, w=
ith
the exposure time varied between 0.5 s and 5 s. We find that desp=
ite
movement of the vessel the measurements produce quantitatively comparable
results apart from saturation effects. We discuss the potential and limitat=
ions
of this method for mapping light pollution in marine and freshwater systems.
This work represents the proof of concept that all-sky photometry with a
commercial DSLR camera is a viable tool to determine light pollution in an
ecological context from a moving boat.
Keywords: photometry, ecological light pollution,
marine light pollution, night sky brightness, all-sky imaging
1.
Introduction
Artificial light at night (ALAN) allows hum=
ans
to extend activities up to a 24 hour-a-day period. While indoor use of ALAN=
can
affect human health as it disrupts the natural circadian rhythm and suppres=
ses
melatonin production [1], the direct photo-related environmental consequenc=
es
are usually small as the light is mainly confined in buildings. Outdoor
lighting on the other hand may have less dramatic consequences on human hea=
lth
[1], but ALAN can spill into the naturally nocturnal landscapes and cause l=
ight
pollution (LP). The increase of outdoor ALAN is a global phenomenon with gr=
owth
rates of 3-6 % per year in industrialized countries and higher in developing
regions [2-3]. Sustainable lighting technology may help to reduce the overa=
ll
LP [4]. Several studies have investigated the impact of ALAN on the environ=
ment
[5] and there is a growing concern that LP affects biodiversity [6,7]. Until
recently, the main focus of studies about ecological LP was on terrestrial
animals [5, 8-10] and plants [10, 11]. However, because human settlements
concentrate along freshwater reservoirs and coastlines, research on ecologi=
cal
LP has shifted towards aquatic systems [12-16].
Historically, the first evaluations of the
night sky brightness (NSB) in the context of LP were performed by astronome=
rs
[17, 18]. They concentrated their investigations on clear sky conditions and
the ability to observe celestial objects. However, illumination conditions =
due
to LP and artificial skyglow (the backscattered portion of the upwelling AL=
AN)
vary dramatically with meteorological conditions, especially clouds [19-22].
Recent work on clouds was linking NSB data with laser ceilometer measuremen=
ts
[22]. The combination of satellite data and radiative transfer modelling can
provide accurate estimates of the NSB for clear skies [23], but is not (yet)
applicable to overcast situations. Ground based NSB measuring methods can
provide information in overcast conditions. These include single sensors su=
ch
as inexpensive handheld sky brightness meters [24] with the advantage of
providing data from citizen scientists, hobby astronomers and researchers o=
n a
local to global scale [16, 25, 26]. The drawback of this method is that no
spatial and no spectral information is available, and most devices only mea=
sure
the NSB at zenith, potentially missing out higher fractions of LP near the
horizon [16, 21, 27]. Photometry with DSLR cameras and fisheye-lenses [18, =
28]
or mosaics [29] can provide this spatial information about the NSB, and have
become available for the public at reasonable prices since consumer electro=
nics
has become a mass phenomenon.
Usually all-sky photometry as part of
astro-photometry is done in a terrestrial context using special filters, lo=
ng
exposure times and fixed tripods or mounts, potentially involving even
compensation for the Earth¡¯s rotation [29]. In astro-photometry, well known
objects are used as references for extinction measurements and the studies =
very
often concern single stars and very dark sites [30]. A very interesting and
powerful device in this context is the ASTMON (All-Sky Transmission MONitor)
making use of all-sky imaging in several spectral bands with a filter wheel
[31], which was also used in areas with LP [32]. For the measurement of the=
NSB
in the context of ecological LP and to compare different sites this precisi=
on
is not necessary. The effects of ALAN can vary by orders of magnitudes with=
in a
radius of several tens of kilometers [21, 26] or by weather conditions [16,=
21,
22]. For most species, the amount of light is the important parameter rather
than single celestial objects. However, visual animals might react on the
directionality that cannot be observed by single sensor devices like lux
meters.
NSB measurements with either single sensors=
or
all-sky photometry are only sparsely done in an aquatic or even in a marine
context [25]. We have recently quantified the NSB at a freshwater lake in
Germany from a floating platform using both methods [16]. However, the plat=
form
was anchored and rocking only slightly. Here we demonstrate that (near) all=
-sky
photometry can also be done from a moving boat in a marine context by using=
a
commercial DSLR camera with a fisheye lens. The camera was mounted on a tri=
pod,
without using motion-compensation. Despite this, we obtained NSB measuremen=
ts
with spatial information from almost the whole hemisphere. In this
proof-of-concept study, we discuss the limitations and the potential of this
method in the context of applying it to investigate the spread of LP, and in
particular skyglow, from coastal towns into unpolluted areas across open
waters.
2.
Methods
2.1. The st=
udy
site
The measurements were performed during
astronomical night on the 5th of March 2016 near the city of Eilat in Israe=
l on
the Gulf of Aqaba close to the Interuniversity Institute for Marine Sciences
(IUI) at 29¡Æ29'33.0"N, 34¡Æ55'54.8"E (see Fig. 1) from an 6.7&nb=
sp;m
long and 3.45 m wide single engine boat. Parts of the western shoreline
north of the IUI are a marine protected area: Eilat's Coral Beach Nature
Reserve and Conservation area, a nature reserve and national park in the Red
Sea. It covers 1.2 kilometers of shore, and is the northernmost coral reef =
in
the world.
Fig.
1. NSB measurement site. Left image: the Gulf of Aqaba with the Sinai Penin=
sula
on the left and Jordan and Saudi Arabia on the right. Red circle marks the
measurement region. Right image – the northern tip of the Gulf of Aqa=
ba where
the fisheye lens observations were made. The blue arrow represents the
measurement location near the border of Egypt, Jordan and Israel. The red a=
rrow
and circle indicate the city of Aqaba, the white arrow and circle indicate =
the
city of Eilat and the yellow arrow shows the position of the Interuniversity
Institute for Marine Science (IUI) .
¡©
2.2. Digital
single lens reflex camera with fisheye lens
NSB measurements can be performed with
calibrated DSLR cameras that allow saving images in an unaltered raw format
[28, 33, 34]. After calibration, it is possible to convert the camera¡¯s
radiance observations in digital numbers to photopic luminance, and to prod=
uce
false color images that show spatially resolved NSB in units of cd/m©÷,
mag/arcsec©÷ or natural sky units (NSU), respectively. If carefully calibra=
ted,
it is possible to reach 10 percent precision with a commercial DSLR camera
[28]. The NSU is a useful translation of NSB into a relative unit, comparing
how much a site differs from a relatively unpolluted site defined to have a
value of 1 NSU. It was introduced to allow non-astronomers to quickly =
rate
the amount of LP at different locations [see e.g. 23] using values from the
widely used sky quality meter (SQM, Unihedron, Canada). NSU =3D =
1 is
defined at 21.6 mag/arcsec©÷ and 254µcd/m©÷, respectively. It ca=
n be
transformed by using the equation:
We used a Canon EOS 6D camera with a Sigma =
EX
DG circular fish-eye lens. The lens has a focal length of 8 mm and an apert=
ure
of F3.5. The camera has a 20.2 Megapixel full frame (36 mm x 24 mm) CMOS se=
nsor
and an integrated GPS tracker. The camera was cross-calibrated with a
thoroughly calibrated camera during an earlier measurement campaign [33] and
several DSLR cameras in a recent intercomparison campaign [34]. Pictures we=
re
obtained in full format (5472 pixel x 3648 pixel), with ISO setting of 6400,
and a varied exposure time between 0.5 seconds and 5 seconds. All
images were saved in raw format. The images were processed using ¡°DiCaLum =
Ver.
0.99" [37] an open source code based on the free software GNU Octave [=
38].
3.
Results and discussion
3.1. Fishey=
e-lens
images
In Figure 2, (near all-sky) fisheye-lens
pictures taken from the slightly rocking and turning boat are shown for
different exposure times a) 5 seconds (image #1517), b) 2 seconds (image
#1518), c) 1 second (image #1519) and d) 0.5 seconds (image #1520). All four
pictures were taken at 21:42 pm local time (19:42 pm GMT). The ca=
mera
was orientated approximately towards the northeast/southwest axis of the Gu=
lf,
with the lower part of the image pointing to the northeast. The camera was =
not
aimed directly towards zenith in order to illustrate the light reflected fr=
om
the water surface. Therefore, it is not exactly all-sky imaging. The two
distinct light domes that are visible at the bottom in Fig 2 a) are from the
center of Eilat (white arrow with black filling on the lower left) and Aqaba
(red arrow without filling on the lower right).
The rocking of the boat was relatively stro=
ng
and is apparent by a distinct ¡°smearing¡± of the stars in the pictures. To
illustrate the possible impact of this rocking on the NSB measurements, Fig=
. 3
a)-d) shows zoom in pictures from the same images as in Fig 2 a)-d). The up=
per
left area of the images was chosen, which is pointing to the southwest and
included Sirius (magnitude -1.46, marked with a red circle) and the constel=
lation
of Orion (indicated with a light blue circle).
The constellation Orion is best visible in =
Fig.
3 c) and d) with low smearing. Here, the belt of three stars (Alnitak, Alni=
lam
and Mintaka) aligned on an axis with Sirius can be seen. In all four images,
the smearing can be tracked best with Sirius. In Fig. 3 a) and b), the smea=
ring
is strong and spiral, while in Fig. 3 c) the smearing is less strong and
linear. In Fig. 3 d), the smearing is lowest but still perceptible by a lin=
ear
distortion of the stars that should be point sources.
As expected, the longer the exposure time, =
the
more susceptible to smearing by rocking and drifting. From the movement of =
the
constellation between the different images, it can also be seen that the bo=
at
was slowly rotating clockwise. The zoomed in sections also show that the si=
gnal
to noise ratio was slightly decreased when the exposure times were reduced.
However due to the smearing the signal to noise improvement by increased
exposure time is not as pronounced as with still images, as the signal now
spreads over several different pixels instead of summing up on one pixel.
Fig.
2. Near all-sky pictures of the night sky above the study site obtained fro=
m a
non-anchored, rocking boat using a commercial DSLR camera (Canon EOS 6D) wi=
th a
fisheye lens (Sigma EX DG 8 mm). An ISO setting of 6400 was used with
different exposure times of a) image #1517 with 5 seconds exposure b)
image #1518 with 2 seconds exposure time c) image #1519 with 1 se=
cond
exposure time and d) image #1520 with 0.5 seconds exposure time. Black=
and
white arrow points to Eilat and red arrow to Aqaba.
Fig.
3. Zoom in of the pictures of Figure 2 showing the detail of constellation =
of
Orion (blue circle) and Sirius (red circle).
3.2. Lumina=
nce
maps
The fisheye images have been converted to
luminance maps that are shown in Figure 4 a)-d) on a scale ranging from 0.5=
to
1000 NSU. The four luminance maps look relatively similar to each othe=
r.
The zenith brightness was on the order of about 5 NSU and increased to=
wards
the horizon with peak brightness values exceeding 1000 NSU at the hori=
zon,
which is attributed to direct city lights. The light domes above the cities
near the horizon reach NSB values of several hundred NSU. These large values
indicate the night sky in this region was affected by skyglow.
Despite the qualitative similarities, there= are some obvious differences in the luminance maps: First, the exposure time resulted in different distortion of the images due to the smearing as discu= ssed above, and only slightly better signal to noise ratio for longer exposure times. The smearing leads to an ¡°analog binning¡± effect that in conjuncti= on with the better signal to noise ratio makes Fig. 4 a) appear most homogeneo= usly or smoother than the other luminance maps. Fig. 4 d) is less smeared and ha= s a lower signal to noise ratio, and therefore appears grainier. Second, for lo= nger exposure times some pixels were overexposed, leading to saturation and loss= of information about certain particularly bright regions near the horizon. Thi= s is most notable when comparing the light emission from the two cities in Fig. = 4 a) and b). The high values in the range of 1000 NSU are resolved in b) but not in a). Third, the rotation of the boat resulted in a slightly different= pointing for the luminance maps. This is most notable when comparing the brightness distribution at the water surface in the lower part of the luminance maps.<= o:p>
a)
c)
Fig.
4. Luminance maps calculated from the fisheye images of Fig. 2. The color b=
ar
represents NSU values. Note that NSU =3D 1 represents approximately a natur=
al sky
without light pollution (0.25 mcd/m©÷).
3.3. Angular
luminance distribution
As can be seen from the luminance maps, the=
re
is a brightness gradient from the zenith towards the horizon. To understand=
the
limits of the method, it is useful to plot the angular brightness distribut=
ion.
Figure 5 shows the luminance as a function of the angle with respect to the
normal vector of the imaging plane (which is not exactly the zenith in our
case, but for typical all-sky images normal vector should point to the zeni=
th)
for a) the full angular range and b) values up to 70¡Æ. To obtain the angul=
ar
luminance distribution, the brightness values for a specific angle have been
averaged by integrating over the azimuth with DiCaLum [31] using a 2¡Æ
resolution in altitude (averaging over rings around the imaging normal vect=
or,
or zenith in true all-sky photometry).
a)
Fig.
5. Luminance as a function of the angle with respect to the imaging normal
vector for, a) the full angular and dynamic range b) zoom in to angles below
70¡Æ,where the values for different exposure times show very good agreement=
.
The average brightness near the zenith is in
the range of 6.0-6.4 NSU and increases to values of several tens of NSU
towards the horizon. For angles up to 74¡Æ, the luminance distribution is v=
ery
similar between the measurements with different exposure times. Differences
become apparent at higher angles, where saturation effects result in platea=
uing
of the luminance values for the three longer exposure times due to
overexposure. At overly long exposure times, the pixels saturate and theref=
ore
the average luminance is underestimated. With a static all-sky camera, this=
can
be avoided by using the high dynamic range method (combining images with
different exposure times). From a drifting and rocking boat, this is more
complicated, but could be achieved by using the brightest stars for image
co-registration. We judge that the short exposure time produces very good
results, at least for this LP situation in the range of 5 NSU.
3.4. Horizo=
ntal
and scalar illuminance
A common measure for the light incident at a
specific site is the horizontal illuminance
However, this is not valid for most scenari=
os,
as usually the NSB is not uniform [16, 20, 21, 39]. Therefore, the horizont=
al
illuminance can only be roughly approximated from narrow angle zenith
measurements alone [27, 39]. From all-sky brightness maps, the horizontal
illuminance (incident at the imaging plane of the camera) can be obtained w=
hen
incorporating a cosine correction [39]. The cosine corrected angular lumina=
nce
distribution is shown in Fig. 6 a) in NSU.
a)
Fig.
6. a) Cosine corrected angular luminance distribution. b) Mean luminance for
the four images in Fig. 2 for different cut-off angles and cosine corrected=
.
The contribution of the NSB at the horizon =
is
reduced and this results in a reduction of the impact of saturation effects=
for
this particular LP situation. The cosine corrected NSB yields similar resul=
ts
for 0.5 s, 1 s and 2 s exposure times. Now only the long
exposure time of 5 s differs significantly from the rest.
For comparison, the mean luminance values in
NSU obtained from the luminance maps (Fig. 3), are plotted in Fig. 6 b).
Without cosine correction and using all the data, the mean values of the
(scalar) illuminace range from 16.4 ¡¾ 1.6 NSU to
21.6 ¡¾ 2.2 NSU, while with cosine correction the values of =
the (horizontal)
illuminance vary only between 6.2 ¡¾ 0.6 NSU and
6.9 ¡¾ 0.7 NSU. Two cut-off zenith angles are also shown wit=
hout
cosine correction: only NSB values up to angles of 74¡Æ and 20¡Æ from the n=
ormal
vector have been used. The 20¡Æ angle is considered because it is similar t=
o the
opening angle of the SQM, a commonly used device in NSB measuring networks.=
The
74¡Æ was chosen to show up to what angle the NSB in this particular site is
relatively homogeneous. The values are summarized in Table 1. In addition to
all-sky images, handheld SQM measurements were taken at the same sites. The=
se
observations gave values of 6.3 ¡¾ 0.6 NSU
(19.6 ¡¾ 0.1 magSQM/arcsec©÷). This is in good agreement wit=
h the
values obtained for the 20¡Æ cut off using the DSLR camera.
Table 1. Mean luminance/NSB
values in units NSU for the four images from Fig. 2 a-d showing different
cut-off zenith angles and cosine corrected values.
|
Image<=
span
style=3D'mso-spacerun:yes'> # |
Exposure time |
Lumina=
nce
[NSU] |
|||
|
Total<= o:p> |
Cos. c=
orr. |
74¡Æ c=
ut
off |
20¡Æ c=
ut
off |
||
|
1517 (a) |
5 s |
16.4 ¡¾ 1.6 |
6.2 ¡¾ 0.6 |
10.4 ¡¾ 1.0 |
6.6 ¡¾ 0.7 |
|
1518 (b) |
2 s |
19.7 ¡¾ 2.0 |
6.7 ¡¾ 0.7 |
10.8 ¡¾ 1.1 |
6.7 ¡¾ 0.7 |
|
1519 (c) |
1 s |
21.6 ¡¾ 2.2 |
6.9 ¡¾ 0.7 |
10.6 ¡¾ 1.1 |
6.6 ¡¾ 0.7 |
|
1520 (d) |
0.5 s |
19.4 ¡¾ 2.0 |
6.4 ¡¾ 0.6 |
9.9 ¡¾ 1.0 |
6.2 ¡¾ 0.6 |
The error bars and margins in the plots and=
in
the table were chosen to be 10% due to the error stemming from the calibrat=
ion
procedure. This error and the systematic error from the pointing differences
between the images (shaking and rotating boat) is certainly higher than any
statistical error in the image analysis. An evaluation of the error between
exposure times can only be done with stable illumination conditions in the
laboratory. So far we have observed that inter camera values are relatively
consistent by comparing many cameras in the field [33, 34]. Apart from the
obvious saturation effects near the horizon especially for image 1517 (5s
exposure time), it can be stated that neither shaking of the camera nor cha=
nge
in exposure time does cause an error higher than from the calibration.
Considering this, we judge that the spread between the measurements in betw=
een
the individual images is not very large and that the noise level is accepta=
ble
for the short exposure time.
The differences mainly originate from the
turning of the boat and from saturation effects. Surprisingly, the relative=
ly
short exposure time of 0.5 s at an ISO setting of 6400 can produce
reasonable information of the NSB at a site with a luminance of about
6 NSU at zenith and non cosine corrected luminance mean values of
16.4 - 21.6 NSU, while capturing a dynamic range up to
1000 NSU.
Our data also shows that a zenith measureme=
nt
(here mimicked with a 20¡Æ cut off and confirmed with an SQM) has the main
drawback that it can underestimate the overall NSB. When inferring the scal=
ar
and horizontal illuminance from the zenith NSB, the deviation between the n=
ear
zenith values and the non cosine corrected integration over the whole
hemisphere is up to a factor of three, while the cosine corrected values ma=
tch
for this particular brightness distribution.
4.
Conclusion and outlook
We have conducted NSB measurements with a
commercial DSLR camera from a moving boat. Despite smearing effects, the
overall NSB distribution could be obtained from near all-sky brightness map=
s.
This proof of concept study shows that the method is well applicable for
investigation of skyglow in marine and also freshwater environments. By
undertaking quick measurements from moving vessels it is possible to acquire
NSB data that provides spatial information about the LP. By varying the
exposure time, it was shown that saturation effects hamper the measurements
more than noise. The data demonstrates, that exposure times as low as
0.5 s produce reasonable results, at least for this site and with this
camera lens combination, which is commonly used by amateur astronomers. We =
want
to point out, that averaging of many images acquired with short exposure ti=
mes
could improve signal to noise even further, like in the ¡°lucky imaging¡±
technique is used to correct for a turbid and rapidly changing atmosphere [=
40].
While the image quality does not sa=
tisfy
the standards of conventional astro-photography, this precision is not
necessary in the context of ecological LP. For this application, quick data
acquisition at many sites and comparability between them is more important =
than
absolute precision. The spatial information is in our opinion needed to inf=
er
the propagation of the light and the possible perception of different anima=
ls.
A simple measurement of the zenith NSB using e.g. an SQM might misinterpret=
the
portion of light at the horizon, and therefore the overall scalar illuminan=
ce.
Another potentially interesting application=
of
the method discussed here is the investigation of the propagation of skyglow
from single sources into a dark environment. This could be done from an
isolated coastal town out into open waters. In contrast to land-based studi=
es,
the measurement on the water has the advantage of the absence of obstacles,=
and
the ability to perform a straight transect. For this particular task a more
rigid construction, a larger vessel less prone to rocking and drifting, and
motion stabilization of the camera are recommended.
5.
Acknowledgements
This work was supported by ESSEM COST Action
ES1204 - Loss of the Night Network (LoNNe). The survey was designed and exe=
cuted
during a COST LoNNe meeting in Israel. AJ, FH and CCMK acknowledge funding =
by
the ILES of the Leibniz Association, Germany (SAW-2015-IGB-1), and the Verl=
ust
der Nacht project funded by the Federal Ministry of Education and Research,
Germany (BMBF-033L038A).
References
[1] Stevens, R. G., & Zhu, Y. (2015). Electr=
ic
light, particularly at night, disrupts human circadian rhythmicity: is that=
a
problem?. Phil. Trans. R. Soc. B, 370(1667), 20140120.
[2] Narisada, K., & Schreuder, D. (2004). Li=
ght
pollution and astronomy. In Light Pollution Handbook (pp. 115-138). Springer
Netherlands.
[3] Hölker, F., Moss, T., Griefahn, B., Klo=
as, W.,
Voigt, C. C., Henckel, D., ... & Franke, S. (2010). The dark side of li=
ght:
a transdisciplinary research agenda for light pollution policy.
[4] Lyytimäki, J. (2015). Towards eco-effic=
ient and
enjoyable lighting, UN-DESA.
[5] Rich, C., Longcore, T. (2006). Ecological
consequences of artificial night lighting, Island Press, Washington, D.C., =
USA.
[6] Hölker, F., Wolter, C., Perkin, E. K., =
&
Tockner, K. (2010). Light pollution as a biodiversity threat. Trends in eco=
logy
& evolution, 25(12), 681-682.
[7] Gaston, K. J., Visser, M. E., & Höl=
ker, F.
(2015). The biological impacts of artificial light at night: the research
challenge.
[8] Rowse, E. G., Lewanzik, D., Stone, E. L.,
Harris, S., & Jones, G. (2016). Dark matters: the effects of artificial
lighting on bats. In Bats in the Anthropocene: Conservation of Bats in a
Changing World (pp. 187-213). Springer International Publishing.
[9] Robert, K. A., Lesku, J. A., Partecke, J., &=
amp;
Chambers, B. (2015, October). Artificial light at night desynchronizes stri=
ctly
seasonal reproduction in a wild mammal. In Proc. R. Soc. B (Vol. 282, No. 1=
816,
p. 20151745). The Royal Society.
[10] Macgregor, C. J., Pocock, M. J., Fox, R., &=
amp;
Evans, D. M. (2015). Pollination by nocturnal Lepidoptera, and the effects =
of
light pollution: a review. Ecological entomology, 40(3), 187-198.
[11] Bennie, J., Davies, T. W., Cruse, D., &
Gaston, K. J. (2016). Ecological effects of artificial light at night on wi=
ld
plants. Journal of Ecology.
[12] Davies, T. W., Duffy, J. P., Bennie, J., &a= mp; Gaston, K. J. (2014). The nature, extent, and ecological implications of ma= rine light pollution. Frontiers in Ecology and the Environment, 12(6), 347-355.<= o:p>
[13] Perkin, E. K., Hölker, F., Richardson,=
J. S.,
Sadler, J. P., Wolter, C., & Tockner, K. (2011). The influence of
artificial light on stream and riparian ecosystems: questions, challenges, =
and
perspectives. Ecosphere, 2(11), 1-16.
[14] Moore, M. V., Pierce, S. M., Walsh, H. M.,
Kvalvik, S. K., & Lim, J. D. (2001). Urban light pollution alters the d=
iel
vertical migration of Daphnia. Internationale Vereinigung fur Theoretische =
und
Angewandte Limnologie Verhandlungen, 27(2), 779-782.
[15] Bruening, A., Hoelker, F., Franke, S., Preu=
er,
T., & Kloas, W. (2015). Spotlight on fish: light pollution affects
circadian rhythms of European perch but does not cause stress. Science of t=
he
Total Environment, 511, 516-522.
[16] Jechow, A., Hölker, F., Kolláth=
, Z., Gessner,
M. O., & Kyba, C. C. (2016). Evaluating the summer night sky brightness=
at
a research field site on Lake Stechlin in northeastern Germany. Journal of
Quantitative Spectroscopy and Radiative Transfer, 181, 24-32.
[17] Riegel, K. W. (1973). Light pollution.
University of California.
[18] Berry, R. L. (1976). Light pollution in
southern Ontario. Journal of the Royal Astronomical Society of Canada, 70, =
97.
[19] Kyba, C. C., Ruhtz, T., Fischer, J., &
Hölker, F. (2011). Cloud coverage acts as an amplifier for ecological =
light
pollution in urban ecosystems. PloS one, 6(3), e17307.
[20] Kocifaj, M., & Lamphar, H. A. S. (2014).
Quantitative analysis of night skyglow amplification under cloudy condition=
s.
Monthly Notices of the Royal Astronomical Society, 443(4), 3665-3674.
[21] Jechow, A., Kolláth, Z., Hölker=
, F. & Kyba,
C. C. M. (2016) Using fisheye-lens imagery to track the reach of skyglow un=
der
clear and partly cloudy conditions near Montsec Astronomical Park in Spain.
ALAN 2016
[22] Ribas, S.J. et al. (2017) How clouds are
amplifying (or not) the effects of ALAN. ALAN 2016, International Journal of
Sustainable Lighting special issue.
[23] Falchi, F., Cinzano, P., Duriscoe, D., Kyba=
, C.
C., Elvidge, C. D., Baugh, K., ... & Furgoni, R. (2016). The new world
atlas of artificial night sky brightness. Science advances, 2(6), e1600377.=
[24] Cinzano, P. (2005). Night sky photometry wi=
th
sky quality meter. ISTIL Int. Rep, 9.
[25] Kyba, C. C., Tong, K. P., Bennie, J., Birri=
el,
I., Birriel, J. J., Cool, A., ... & Ehlert, R. (2015). Worldwide variat=
ions
in artificial skyglow. Scientific reports, 5, 8409.
[26] Pun, C. S. J., & So, C. W. (2012).
Night-sky brightness monitoring in Hong Kong. Environmental monitoring and
assessment, 184(4), 2537-2557.
[27] Kocifaj, M., Posch, T., & Lamphar, H. S.
(2015). On the relation between zenith sky brightness and horizontal
illuminance. Monthly Notices of the Royal Astronomical Society, 446(3),
2895-2901.
[28] Kolláth, Z. (2010). Measuring and mo=
delling
light pollution at the Zselic Starry Sky Park. In Journal of Physics:
Conference Series (Vol. 218, No. 1, p. 012001). IOP Publishing.
[29] Duriscoe, D. M., Luginbuhl, C. B., & Mo=
ore,
C. A. (2007). Measuring night-sky brightness with a wide-field CCD camera.
Publications of the Astronomical Society of the Pacific, 119(852), 192.
[30] Romanishin, W. (2006). An Introduction to
Astronomical Photometry Using CCDs. University of Oklahoma, 17. from http://www.physics..csbsju.edu/370/photometry/manuals/=
/OU.edu-CCD-photometry-wrccd06.pdf
[31] Aceituno, J., Sánchez, S. F., Aceitu=
no, F. J.,
Galadí-Enríquez, D., Negro, J. J., Soriguer, R. C., & Gom=
ez, G. S. (2011).
An all-sky transmission monitor: ASTMON. Publications of the Astronomical
Society of the Pacific, 123(907), 1076.
[32] Aubé, M., Kocifaj, M., Zamorano, J.,=
Lamphar,
H. S., & de Miguel, A. S. (2016). The spectral amplification effect of
clouds to the night sky radiance in Madrid. Journal of Quantitative
Spectroscopy and Radiative Transfer, 181, 11-23.
[33] Kyba, C. C. M., Bouroussis, C., Canal-Domin=
go,
R., Falchi, F., Giacomelli, A., Hänel, A., ... & Tong, K. P. (2015=
). Report
of the 2015 LoNNe Intercomparison Campaign. Report of the ES, 1204.
[34] Ribas, S. J., Aubé, M., Bará =
S., Bouroussis C.,
Canal-Domingo R., Espey B., Hänel A., Jechow A, Kolláth Z., Mar=
ti G., Massana
P., Schmidt, W., Spoelstra, H., Wuchterl, G., Zamorano, J., & Kyba, C. =
C.
M. (2016(in press)). Report of the LoNNe Intercomparison Campaign.
[35] Pogson, N. (1856). Magnitudes of Thirty-six=
of
the Minor Planets for the first day of each month of the year 1857. Monthly
Notices of the Royal Astronomical Society, 17, 12-15.
[36] Hänel, A. et al. (2017 (in submission)=
).
Current techniques and problems involved in measuring light pollution, J.
Quant. Spectrosc. Radiat. Transfer.
[37] Kolláth, Z., et al. (in preparation)=
[38] Eaton, J.W. (1996). GNU Octave, from https://www.gnu.org/software/octave/=
[39] Duriscoe, D. M. (2016). Photometric indicat=
ors
of visual night sky quality derived from all-sky brightness maps. Journal of
Quantitative Spectroscopy and Radiative Transfer, 181, 33-45.
[40] Law, N. M., Mackay, C. D., & Baldwin, J=
. E.
(2006). Lucky imaging: high angular resolution imaging in the visible from =
the
ground. Astronomy & Astrophysics, 446(2), 739-745.
A. Jechow et al. /International Journal of Sustainable
Lighting 20 (2017) 15-25