SKY BRIGHTNESS LEVELS BEFORE AND AFTER THE CREATION OF THE FIRST
INTERNATIONAL DARK SKY RESERVE, MONT-MÉGANTIC OBSERVATORY, QUÉBEC,
Martin Aubé1,2 and Johanne Roby1
1 Cégep de Sherbrooke, 475, rue du Cégep, Sherbrooke, Qc, Canada, J1E 4K1
E-mail: email@example.com, firstname.lastname@example.org
2 Corresponding author. Tel.: +819-564-6350 ext: 4146
In 2007, the area around the Mont-Mégantic Observatory (MMO) was officially certified by the
International Dark-Sky Association and the Royal Astronomy Association of Canada as the first
International Dark Sky Reserve (IDSR). In order to be able to investigate the impact of Artificial Light at
Night on night sky brightness before and after the establishment of the IDSR, we used a heterogeneous
artificial sky brightness model including an implicit calculation of 2nd order scattering (ILLUMINA)
developed by Martin Aubé’s group. This model generates three kinds of outputs: the sky radiance at the
given site, observing angle and wavelength and the corresponding contribution and sensitivity maps. The
maps allow for the identification of the origin of the sky radiance according to each part of the
surrounding territory. For summer clear sky conditions, the results show that replacing light fixtures
within a 25 km radius around the MMO with cut-off High Pressure Sodium devices and reducing the total
installed radiant power to ~40% of its initial level are very efficient ways of reducing artificial sky
brightness. The artificial sky brightness reduction at zenith observed after the establishment of the IDSR
was ~50% in the 546 nm mercury spectral line, while the reduction obtained in the 569 nm sodium line
was ~30%. A large part of that reduction can be associated to the reduction in radiant power. The
contribution and sensitivity maps highlight critical zones where any changes in the lighting infrastructure
have the most important impact on sky brightness at the MMO. Contribution and sensitivity maps have
been used to analyze the detailed origin of sky brightness reduction. The results of this study are intended
to support authorities in the management of their lighting infrastructure with the goal of reducing sky
brightness. The results have been shared with MMO officials and are being used as a tool to improve sky
quality at the observatory.
Dark sky areas are now increasingly rare in the world and this is due to the constant growth of artificial
light at night (ALAN). This phenomenon is largely the result of human activities and its primary sources
are street lamps, advertising panels and lighted buildings. Astronomers were the first to point out that dark
skies are disappearing when they realized that sky observation was becoming more difficult because of
the bright halos caused by ALAN. It is only in the past decade that the multiple negative impacts of
ALAN on fauna, flora and human health have been documented more intensively in the literature [1, 2,
Artificial light produced by street lights has been identified as one of the major sources of night sky
brightness. As such, sky brightness can be reduced by improving their performance. For example, a
transition from the Cobrahead model street light, which emits ~6% of its flux upward, to the Helios
model, a cut-off street light which emits ~1% of its flux upward, clearly reduces the total amount of light.
In fact, the downward flux is largely absorbed by the ground, especially during the summer when there is
no snow cover.
In 1978, a professional astronomical observatory located about 1100 meters above sea level on top of
Mont-Mégantic in the Eastern Townships, Québec (Canada) was inaugurated. The Mont-Mégantic
Observatory (MMO) is equipped with a Ritchey-Chrétien telescope whose primary mirror is 1.6 meters in
diameter, making it the fourth largest in Canada and the largest in eastern North America. It is the best
equipped astronomical observatory in Canada [5, 6] and is located over 60 kilometers from the closest
urban center (Fig. 1). Its mission is to conduct astrophysical research and train young researchers for
work in other major observatories around the world, thus exporting expertise from the Observatory. Also,
the Observatory develops state-of-the-art instruments that are globally recognized for their high quality.
Fig. 1. Mont-Mégantic Observatory (MMO) with the Milky Way right above (a) and its location relative
to the cities in southern Québec (b). Credits for left pane picture: Guillaume Poulin. The map on right
pane has been created with OpenStreetMap, © OpenStreetMap contributors. MMO is identified by the
dark star on pane (b).
It is estimated that in 1979, the sky brightness at the summit of the MMO was around 25% higher than the
natural sky brightness value. This evaluation was, unfortunately, only made qualitatively by observatory
staff members. Despite being located far from major centers, sky brightness at the summit further
increased, almost doubling, between 1979 and 1998 (i.e. an increase of ~4% per year). This increase in
sky brightness became a real threat to scientific studies and to the basic objectives of the MMO.
Furthermore, the research, education and tourism activities in the region of Mont-Mégantic are based
primarily on the astronomical observatory; the protection of the night sky is thus crucial for the local
Faced with these challenges, it became imperative to stop the growth of artificial lighting in order to
reduce sky brightness. ASTROLab, a public outreach center, developed and implemented a light pollution
abatement project that included the establishment of an International Dark Sky Reserve (IDSR) covering
5500 square kilometers around the MMO. This regional-scale initiative provided us with the unique
opportunity to study the changes in artificial sky brightness parameters before and after the
implementation of these protective measures.
In this paper, we will describe how light fixture parameters, properties of urban and rural environments
(reflectance, obstacles and topography), wavelength, and atmospheric content may influence the level of
sky brightness as a function of the viewing angle. Comparison of artificial sky brightness levels before
and after the creation of the IDSR will also be shown. To achieve the level of sensitivity needed for the
present study, we used a heterogeneous numerical radiative transfer model [8, 9]. The results allow us to
represent as faithfully as possible the phenomenon of artificial sky brightness as can be seen from a
standing point at any horizontal and vertical viewing angle. In our study, the observer location was set to
The model also produces contribution and sensitivity maps, two powerful tools that allow the
identification of the origin of the artificial sky brightness and the most efficient ways to act in order to
reduce it. By comparing the results for 2005 (before the creation of the IDSR) with the results for 2009
(after the creation of the IDSR) we will be able to monitor the impact of creating the IDSR on artificial
2.1.1 The light pollution abatement project
In 2003, after considering several initiatives aiming to address the worsening sky brightness problem, a
local public outreach center, ASTROLab, developed an ambitious light pollution abatement project to
reduce artificial sky brightness at the MMO by 50% to revert back to the brightness levels of 1979 .
The project focused on two zones concentrated inside a 50 kilometer radius around the MMO (Zone 1: 0
to 25 km from the MMO, Zone 2: 25-50 km from the MMO) and a third zone corresponding to the city of
Sherbrooke, which is located approximately 60 kilometers from the MMO (Fig. 2). These intervention
areas were determined according to their respective estimated sky brightness contribution at the MMO.
The MMO is surrounded by several municipalities and, despite their small size, those located within a
radius of 25 kilometers contribute significantly to sky brightness. Fig. 3 shows a 360o panorama of the
light domes as seen from the MMO in 2007 during the lighting devices conversion.
Fig. 2. Map of the Mont-Mégantic area International Dark Sky Reserve (MMO IDSR) identifying its
three zones. Zone 1: 25km around MMO, Zone 2: 50 km around MMO. Zone 3: City of Sherbrooke,
Fig. 3. Light domes produced by municipalities surrounding the Mont-Mégantic Observatory.
Municipality characteristics (name: population, surface area, and distance from the Observatory): Notre-
Dame-des-bois: 985 inhabitants, 191km2, 10km; La Patrie: 810 inhabitants, 207km2, 10 km; Sherbrooke:
154 000 inhabitants, 366km2, 60 km; Scotstown: 836 inhabitants, 122km2, 12 km; Lac Mégantic: 6086
inhabitants, 26.4km2, 25 km.
As of 2005, following important public awareness efforts, the municipalities involved with the project
adopted regulations pertaining to the installation of new lighting devices. With the cooperation of
businesses, industries and residents, a lighting device conversion program was initiated in order to
convert existing light fixtures in the nearby municipalities to devices that were less powerful
(approximately 40% lower radiant power), but more efficient; these fixtures significantly reduced light
pollution without significant impact on the ambient light. In the conversion process, white lights like
metal halide (MH) or mercury vapor (MV) lamps were replaced by high pressure sodium (HPS) lights
(model: Helios). In 2007, over 3300 fixtures were replaced resulting in energy savings of nearly 2
GWh/year. The 40% radiant power average reduction has been obtained from statistical data provided by
the MMO's IDSR in their report to Hydro-Québec . The impact on the starry sky was immediate and
2.1.2 Establishment of the Mont-Mégantic area International Dark Sky Reserve
In order to continue to pursue the project's goals and help ensure their persistence in time, ASTROLab
proposed the creation of the first International Dark Sky Reserve (IDSR) covering an area of nearly 5500
square kilometers that includes two Regional County Municipalities, the city of Sherbrooke, 35
municipalities and over 225 000 citizens. Under the leadership of ASTROLab and with the help and
strong support of numerous regional partners, such as the MMO and the Mont-Mégantic National Park,
the Mont-Mégantic area IDSR was formally established in 2007 and is recognized by the International
Dark-Sky Association and the Royal Astronomy Association of Canada (RASC). This official recognition
favors the sustainability of the project by giving the local population a sense of pride about this
2.2 Modeling experiment
2.2.1 Light Pollution Numerical Model: ILLUMINA
The radiative transfer model used for this study is ILLUMINA [8,9] version 2. This model is distributed
under Gnu Public License and can be downloaded from Google Code . In version 2, a statistical
optimization procedure has been added for the selection of ground pixels and line of sight voxels to
reduce computing time.
Basically, ILLUMINA acts as a ray tracing software where a set of photons are thrown from light fixtures
above the ground cells and then reach the observer’s field of view after four different light paths: 1- single
scattering by molecules and aerosol inside voxels of the line of sight (I1); 2- single scattering after a
lambertian reflection on the ground (Ir1); 3- 2nd order of scattering in the line of sight after a single
scattering from an atmospheric voxel in a volume surrounding path between the source and the line of
sight voxel (I2); and 4- same as path 3, but after a reflection on the ground (Ir2). The geometry of these
light paths is illustrated in Fig. 4. Along with scattering processes toward the observer, extinction from
aerosols (scattering and absorption) and molecules (scattering only) is computed for all light paths
considered. In 2007, Aubé  showed that the 2nd order of scattering may have a significant impact on
artificial sky brightness, especially when the observer is far from cities. This phenomenon may be
explained by the fact that the single scattering dome of light acts as a large source for the 2nd order of
scattering process and thus its distance decreasing function is less steep when compared to point-like
Fig. 4. Light paths considered for the calculation of the artificial sky radiance in the model ILLUMINA.
“o” represents the observer, “n” is a given voxel falling into the line of sight, “s” is a light source cell, and
“m” are voxels intervening in the calculation of the 2nd order of scattering.
When 2nd order scattering is computed, the numerical approach is CPU time-consuming and requires
access to a high performance computing infrastructure. In the model, the atmosphere is subdivided into 50
prescribed vertical levels, from the lowest altitude of the modeling domain to 30 km above. Thereby, the
model accounts for about 99% of the atmosphere. Vertical level thickness is increasing with altitude in
order to be more accurate at low altitude voxels where atmospheric concentration and light flux are most
often higher. We assume exponential vertical profiles for aerosol and molecular concentrations. A scale
height of 2 km was used for aerosols while 8 km has been adopted for molecules.
ILLUMINA computes optical impact of size distribution and composition of aerosol content using Mie
theory for spherical particles; more complex particle shapes are not yet implemented, leading to a
possible overestimate of the backscattering efficiency . We are using the complex refractive index and
bi-modal lognormal size distributions suggested by Shettle and Fenn . The aerosol composition can
be adjusted in accordance to modeling experiment particularities in term of geography (rural, urban or
maritime) or to account for special atmospheric events like important biomass burning, volcanic eruption
or dust storms. ILLUMINA is a regional model, and thus the maximal domain size should not exceed a
few hundred kilometers. We assume a plane-parallel atmosphere. The horizontal resolution typically
ranges between 150 m and 1 km. Light Output Patterns (LOPs) for each horizontal cell are determined
from a linear combination of Illuminating Engineering Society of North America (IESNA) files
corresponding to the different lamps found in this cell. We assume that LOP is isotropic along azimuth
angles. This is realistic as long as many light fixtures with various azimuthal orientations fall in a grid
cell. Spectral lamp radiant power, LOP, ground reflectance, ground altitude & tilt, and lamp height
relative to the ground are defined independently for each grid cell. Reflection on the ground is assumed to
be lambertian. It is planned to implement bidirectional reflectance functions in the future. Detailed
computation of shadowing effects from masking by ground elevation is performed while a crude
calculation is done for subgrid masking by smaller obstacles like trees and buildings.
2.2.2 Modeling domain
The modeling domain includes the main light sources that may have a significant effect on the MMO. In
this study, we set the horizontal resolution to 1 km. This resolution has been chosen to match the worst
gridded dataset to be used, namely the Defense Meteorological Satellite Program Operational Linescan
System (DMSP-OLS) nighttime satellite radiances. The model boundaries have been chosen to ensure a
buffer region of 50 km between the observatory and its nearest domain limit. Giving that the model height
is 30 km, this means that we can accurately model zenith angles up to z≈60o. This limitation is only
encountered toward the south and in this specific orientation, no significant sources are found before a
distance of 100 km. Therefore, in our case, even toward the south the model calculations are accurate up
to z≈75o. The domain extent is 45o N to 47.5o N and 75o W to 70o W. This domain is covered by 395 x
290, 1 km by 1 km cells. The modeling domain can be seen in Fig. 1 (b).
2.2.3 Preparing input data
To perform a modeling experiment for a given spectral line, relevant gridded and non gridded datasets
have to be provided to the model. Gridded datasets include: 1- the light radiant power for the given
wavelength; 2- the LOP; 3- the lamp height relative to the ground; 4- a digital elevation model (DEM); 5-
the ground reflectance at the same wavelength; and 6- the land/water mask. The ground reflectance was
taken from NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) Surface-Reflectance
Product (MOD09A1, Vermote and Vermeulen 1999) . For the 546 nm and 569 nm lines, we retained
the MODIS Level 3 land band 4 (centered at 555 nm). The spatial resolution of the reflectance product is
500 m. Reflectance data is a combination of the 8 day L3 composite. The L3 composite contains the best
possible observation during an 8-day period as selected on the basis of high observation coverage, low
view angle, the absence of clouds or cloud shadow, and low aerosol loading. DEM is determined with the
Shuttle Radar Topography Mission (SRTM). We used the 3 arc second resolution product of the SRTM
V2 . SRTM vertical resolution is 16 m in absolute values and 10 m in relative values. The help of a
local expert was required to specify geographical zones with common lamp types and lamp spectral
power distribution (SPD) mix. In fact, these parameters can change from one pixel to another, but, in such
cases, the amount of information can be virtually impossible to gather and manage. MODIS surface
reflectances, LOPs and DMSP-OLS upward radiances are combined using equation 1 to produce the light
radiant power map.
is the spectral radiant power (W),
is the calibration constant for each spectral line considered,
is the DMSP-OLS radiance (Version 4, Elvidge 2008),
is the upward flux fraction,
ground reflectance, and
is the value of the light output pattern toward zenith. We used the zenith
angle even if we know that in many cases, DMSP-OLS is not looking toward nadir. However, we
consider that this weakness of the model is mitigated by the fact that in many cases the reflected light is
dominant in this equation. In a future version of our model, we plan to account for the mean viewing
angle of the satellite and then use the relevant angle instead of 0. The DMSP-OLS dataset is maintained
by the Earth Observation Group of the National Geophysical Data Center (NGDC) which is a part of the
US National Oceanic and Atmospheric Administration (NOAA). DMSP-OLS data are coded from 0 to 63
on a linear, but relative sensitivity scale. Some pixels may be saturated and in that case are set as
numerical value 63. This dataset is a yearly composite were only the cloud screened and stable values are
used. To minimize the potential impact of sensitivity changes from one satellite to another, we used the
data from the same DMSP satellite (satellite F16). We assume that variations in the satellite radiances
from one year to another are directly linked to real ground based variations. Our use of uncalibrated data
is not a limitation in this study, given the entire analysis is based on relative comparisons from one year to
Fig. 5 gives four examples of radiant power maps created with equation 1 for two spectral lines, 546 nm
and 569 nm, associated to specific lamps. The 546 nm line is a mercury line (Hg) produced by MV lamps
and MH lamps. The 569 nm line is a sodium line (Na) produced by HPS. The maximal radiant power in
the 546 nm line appears lower that that in the 569 nm line. We can explain this by the fact that the highest
installed radiant power per square kilometer occurs in cities while MV and MH lamps are more common
in the countryside. According to spectral sky radiance measurements made in 2006 at the MMO by Aubé
and available from the Sky spectral radiance database , the two lines are typically of the same
magnitude. But, of course, the MMO is far from city lights and thus more dependent on countryside
sources. It is easy to notice that the radiant power at 546 nm shows a significant reduction in 2009
compared to 2005 around the MMO (identified by the black star). A lower but significant reduction is also
visible over the city of Sherbrooke located around coordinate 240 W-E and 50 S-N. We also observed a
general increase in the installed radiant power both in the 546 nm and 569 nm lines for all other sites.
Fig. 5. Zoom around the MMO of the radiant power images for 2005 (left) and 2009 (right) at 546 nm ((a)
and (b)) and at 569 nm ((c) and (d)).
Table 1 summarizes the characteristics of the lighting infrastructure both in 2005 and 2009. In this table, it
is possible to find the proportion of each lamp type for different regions. Lamps are distinguished by their
mix of photometric types and their mix of spectral types. The lighting inventory was simplified to retain
only the three most common photometric types. In fact, in southern Québec, most light fixtures can be
associated to one of the three models shown, even if they are not exactly the same. The percentage of
each photometric type was determined using Street View in Google Maps. We first chose a set of random
locations inside each region (MMO, Cookshire and Lac Mégantic, Sherbrooke, and elsewhere in Québec)
and then determined the number of fixtures corresponding to each photometric type present using Street
View. We distinguished urban from rural environments as it is shown in Table 1. For the percentage of
each spectral type, we visited each region at night to count the number of white and yellow lamps
(respectively lamps emitting a 546 nm Hg line and lamps emitting a 569 nm Na line). Resultant LOPs for
each region shown in Table 1 were computed by doing a linear combination of each of the three
individual lamp LOPs with respect to the percentage given in table 1's legend.
Table 1. Lamp angular photometry and spectrum distribution in the zones surrounding the
Mont-Mégantic Observatory (MMO) before and after the establishment of the International
Dark Sky Reserve.
Distribution of light xture types
Before IDSR (2005) After IDSR (2009)
t type Helios Cobrahea
Helios Cobrahead Farm
1: representing 86% sodium (Na) and 14% mercury (Hg)
2: representing 66% sodium (Na) and 34% mercury (Hg)
3: representing 89% sodium (Na) and 11% mercury (Hg)
4: representing 80% sodium (Na) and 20% mercury (Hg)
5: representing 90% sodium (Na) and 10% mercury (Hg)
Finally, the model requires some non gridded parameters that are considered constant over the entire
geographical domain: 1- the wavelength used; 2- the aerosol optical depth (AOD); 3- the aerosol
scattering and absorption cross sections at the given wavelength and relative humidity; 4- the aerosol
scattering phase function (aerosol model) at the given wavelength and relative humidity; 5- the ground
level atmospheric pressure and the average height and distance between subgrid obstacles; 6- the observer
position; and 7- the viewing angles. The values of non gridded parameters used for this study are listed in
Table 2: Non gridded modeling parameter values.
Wavelengths (in nm) 436 (Hg), 498 (Na), 546 (Hg), 569 (Na), 616 (Na)
Aerosol optical depth (no units) 0.05, 0.1, 0.2 (low turbidity, clear sky)
0.5, 1 (high turbidity, hazy sky)
Aerosol model Rural
Relative humidity 70%
Ground level atmospheric pressure 101.3 kPa
Average obstacle height 9 m
Average distance between obstacles 13 m
Observer position MMO
Zenith angles (degrees) 0, 30, 50, 60, 70, 75
Azimuthal angles (degrees) 0 to 345 degrees at 15 degree intervals
3 Results and discussion
3.1 Output data
ILLUMINA generates three different outputs. The first one is the artificial sky radiance calculated for a
given viewing angle and observer position in a given spectral line. The model also generates two gridded
outputs: the contribution map and the sensitivity map. The contribution map indicates from where and in
what proportion the calculated artificial sky radiance originates from. The sensitivity map indicates how
sensitive the artificial sky radiance values are to any change in the installed radiant power at a given
model cell containing at least one light device. Contribution and sensitivity maps for different sites and
periods can be accessed via an interactive web portal developed and maintained by our group .
3.1.1 All sky radiance ratio maps
These maps illustrate changes in artificial sky brightness by taking the ratio of the artificial sky radiances
observed after the light conversion project over the artificial sky radiances observed before the conversion
project. Each plot is composed of 121 ratios for different viewing angles (six zenithal angles (75, 70, 60,
50, 30, 0 degrees) and twenty four azimuthal angles (0 to 345 degrees at 15 degree intervals)). A single all
sky radiance ratio map requires 242 model runs to be accomplished.
3.1.2 Sky radiance contribution map
The contribution maps illustrate how each square kilometer of land contributes to artificial sky brightness
detected at a given viewing angle from the MMO. This map is normalized so that the sum of all pixels
equals 100%. The values are expressed in term of percentage points per square kilometer.
3.1.3 Sky radiance sensitivity map
The sensitivity maps illustrate the impact on artificial sky brightness measured at the MMO of an
hypothetical generic street lamp installed on each square kilometer. The sensitivity map is only calculated
for pixels containing a non-zero radiant power. This map is also normalized so that the sum of all pixels
equals 100%. This map is very useful for local decision-makers as it allows them to identify the places
where intervention is necessary to efficiently reduce sky brightness.
3.2 Sky radiance contribution and sensitivity maps analysis
All the contribution and sensitivity maps are georeferenced and published in a web portal . There is
one contribution and one sensitivity map for each model configuration (121 viewing angles, five
wavelengths, five aerosol optical depths, and two night periods). The maps corresponding to zenith
viewing angle are reported in Figures 6 and 7. Panes (a) and (b) in Fig. 6 show a comparison of summer
zenith contribution maps before and after the creation of the Mont-Mégantic area IDSR under the same
conditions (clear sky and 546 nm Hg line). Artificial sky radiance has decreased significantly around the
MMO. We found a relative reduction of 50% at zenith. The replacement of MV and MH lamps to HPS
Helios lamps in zone 1 was efficient. Panes (a) and (b) of Fig. 6 show that in 2005, the zenith 546 nm Hg
line radiance came mainly from closer and small towns surrounding the MMO. In 2009, the 546 nm Hg
line radiance comes from farther. This change is clearly illustrated by the greater spread in contribution of
pane (b) compared to pane (a). We have integrated all contributions from different IDSR zones and also
for some nearby villages. Table 3 gives a compilation of these integrations. In zone 1 (less than 25 km
from the MMO), the total contribution to the zenith 546 nm Hg line radiance in 2005 was 57.2%; this was
reduced to 12.1% in 2009. In zone 2 (25 km to 50 km from the MMO), the total contribution to the zenith
546 nm Hg line radiance was 23.3% in 2005 and 45.7% in 2009. Interestingly, under clear sky, zone 3
shows a very low contribution to the 546 nm Hg line in 2005 and in 2009 (0.9% and 1.1% respectively;
see Table 3). The same remark stands for the 569 nm Na line for both years (2.3% and 3.3% respectively).
Despite Sherbrooke being the largest city in the region (154 000 inhabitants), it is quite far from the
MMO (60 km away) and. moreover, an important part of the city is masked from view by the MMO by a
hill located on its eastern side.
Fig. 6. Zoom around the Mont-Mégantic Observatory (MMO) of the comparison of zenith summer
contribution map for 546 nm Hg line before (left pane) and after (right pane) the creation of the Mont-
Mégantic area International Dark Sky Reserve (IDSR). Panes (a) and (b) correspond to typical clear sky
conditions for Québec with an aerosol optical depth of 0.2. Panes (c) and (d) correspond to a typical
pollution event in that region with an aerosol optical depth of 1.0. Pane (c) is corresponds to 2005 (before
IDSR) and (d) to 2009 (after IDSR). The white star represents the location of the MMO and thin circles
have been used to identify the following villages or cities surrounding the MMO: 1- Sherbrooke; 2-
Notre-Dame-des-Bois; 3- La Patrie; 4- Scotstown; 5- Milan; 6- Lac-Mégantic.
Fig. 7. Zoom around the Mont-Mégantic Observatory (MMO) of the summer zenith contribution and
sensitivity maps for 569 nm Na line in 2005 (left) and in 2009 (right). The contribution maps are shown in
panes (a) and (b) and the sensitivity maps are shown in panes (c) and (d). All panes correspond to typical
summer, clear sky conditions for Québec province with an aerosol optical depth of 0.2.
Table 3: Integrated contribution to the zenith articial sky brightness for the International Dark
Sky Reserve zones and villages.
Region 546 nm clear 569 nm clear 546 nm hazy
2005 2009 2005 2009 2005 2009
546 nm clear 569 nm clear 546 nm hazy
Zone 1 - 0-25km 57.2% 12.1% 55.0% 27.7% 95.2% 56.4%
Zone 2 - 25-50km 23.3% 45.7% 22.8% 35.4% 3.8% 33.4%
Zone 3 - Sherbrooke 0.9% 1.1% 2.3% 3.3% 0.1% 0.3%
Lac-Mégantic 4.9% 2.9% 4.9% 4.7% 1.4% 4.0%
Notre-Dame-des-Bois 4.1% 0.8% 4.0% 2.4% 8.9% 8.0%
La Patrie 4.7% 1.0% 4.5% 2.9% 7.4% 7.1%
Scotstown 3.2% 0.7% 3.1% 2% 3.0% 3.2%
For polluted atmosphere conditions, we observe that the contribution values are constrained around the
MMO (e.g. in 2005, 95.2% of the 546 nm Hg line radiance was coming from zone 1 (see Table 3)). This
is the effect of the extinction of light with distance. Basically, panes (c ) and (d) of Fig. 6 are showing that
under high turbidity conditions, the light coming from remote light fixtures is not contributing to the local
artificial sky brightness. In other words, the artificial sky brightness is completely dominated by local
sources. According to Fig. 6, we can roughly estimate that the action radius of zenith artificial sky
brightness under an aerosol optical depth of 1.0 is of the order of 20 km while it can reach 50-100 km
when the aerosol optical depth is 0.2.
With contribution maps, it becomes easy to associate part of the artificial sky radiance values with
specific points on the ground. We are then able to identify the places that are most damaging in terms of
their impact on artificial sky brightness. Our results can be used to strategically minimize the impact of
light pollution on the MMO's astronomical activities and to efficiently control the addition of new street
lights in the surroundings.
In addition to the contribution maps in the 569 nm Na line (panes (a) and (b)), Fig. 7 shows the sensitivity
maps (panes (c) and (d)). The highest values in these maps show where we must intervene first in order to
achieve a more efficient sky brightness reduction. Pane (c), which corresponds to 2005, clearly shows that
only the conversion of sources inside zone 1 was likely to reduce sky brightness. The map also shows that
special emphasis had to be put on the sources located nearest to the MMO. In 2009, the extent of the
sensitive zone had increased compared to 2005 and, in the context of a future conversion program, efforts
would have to target a larger zone with an approximate radius of 40 km. In that case also, the most
important sources to check or reduce are those closest to the MMO. In a certain sense, this result is
showing that during the IDSR conversion project, it was more efficient to focus on reducing the radiant
power in zone 1 than making any conversions in zone 2. Here, it is important to understand that
sensitivities are only computed for pixels having non-zero radiant power. This also contributes to the
increased geographical extent of sensitivity from 2005 to 2009. In fact, in 2009, the extent is larger
because light fixtures have been added in places where there were none in 2005.
3.3 All sky radiance ratio maps
The results for the artificial sky brightness ratio data for summer 2009 over summer 2005 are shown in
Fig. 8. The results are reported for the 546 nm Hg and 569 nm Na lines under clear and hazy sky
conditions. The comparison between these lines under clear sky conditions (panes (a) and (b)) shows that
the reduction at the zenith for the 546 nm Hg line is about 50% compared to only about 30% for the 569
nm Na line. These results are explained by the fact that most of the MV and MH lamps were replaced by
HPS lamps in the IDSR, so that the number of final HPS lamps increased. The results under a hazy sky
show a zenith reduction of 90% for the 546 nm Hg line (Fig. 8 (c)). This greater reduction can be
explained by the fact that under hazy conditions, most of the artificial sky brightness is generated by
nearby light sources and the lamp conversion initiative was implemented nearby to the MMO (mainly
inside a radius of 25 km, e.g. zone 1 in Fig. 2). Even if there are no astronomical observations under hazy
conditions because of the extinction of star lights, the sky appears darker after the conversion. In essence,
we can say that in high turbidity conditions, the sky 546 nm Hg line radiance was ~10 times brighter in
2005 compared to 2009.
In southern Québec, the snow cover during winter is quite significant. We decided to take this into
account and performed a modeling experiment using the winter ground reflectance data given by MODIS
(data extracted for February). The all sky 546 nm Hg line ratio is shown in Fig. 8 (d) for clear sky
conditions. The result is very interesting, as we notice that the reduction is almost constant across the
entire sky, contrary to what we observed in summer where the reduction is maximal around zenith. The
uniformity of the winter all sky reduction can be understood by the fact that during winter, most of the
light goes up when it is reflected by the snow (given the high reflectance of the snow). Also, this
reflection is almost lambertian, so that the impact of the angular dependency of the LOP become very
small compared to the reflected flux. One reason to see an artificial sky brightness increase toward the
horizon during summer is that LOPs often show a peak near horizontal emission and this near horizon
light is travelling large distances. During the summer, the lambertian angular dependencies of the
reflection are less important with respect to the global angular emission of a ground element (direct light
and reflected light) because of the low reflectance of the ground. In winter, the reflection becomes
predominant so that the near horizon emission of light is relatively less important, subsequently
decreasing the zenith to horizon variation in artificial sky brightness.
In the North-West direction during summer (Fig. 8 (a)), toward the small village of Scotstown located at
12 km from the MMO, we observed an artificial sky brightness reduction as low as 5% for the 546 nm Hg
line and almost 0% for the 569 nm Na line. The lighting conversion efforts in that village were not very
successful in reducing sky brightness at that viewing angle. This low reduction can be understood by the
fact that, at that angle, an important part of the sky 546 nm Hg line radiance is coming from zone 2, more
specifically, around the Graymont mine near Bishopton and around the village of Weedon. In table 4 we
can see that in 2005, 29.8% of the 546 nm Hg line radiance was coming from zone 2 while 23.9% was
coming from zone 1. This higher contribution of zone 2 is even more evident in 2009. At that time, 36.7%
of the 546 nm Hg line radiance was coming from zone 2 compared to only 5.2% from zone 1.
Table 4: Integrated contribution articial sky brightness toward azimuth 300o and zenith angle
75o for International Dark Sky Reserve zones 1, 2 and 3
Region 546 nm clear
Zone 1 23.9% 5.2%
Zone 2 29.8% 36.7%
Zone 3 1.7% 1.2%
Fig. 8. All sky maps are for ratio data for summer 2009 over summer 2005. Panes (a) and (b) are for
typical clear sky conditions for Québec with an aerosol optical depth of 0.2 for Hg (546nm) and Na
(569nm) lines, respectively. The map in pane (c) presents results for typical hazy sky conditions for
Québec with an aerosol optical depth of 1.0 for the Hg line (546nm). Pane (d) presents results for winter
with typical clear conditions for Québec with an aerosol optical depth of 0.1 for the Na line (569nm). The
direction and the distance (in km) between the main towns and cities and the MMO are displayed around
the map. The ratio is expressed as contour lines.
Our modeling experiment shows that replacing, within a 25 km radius around the MMO, a portion of the
light fixtures with HPS cut-off fixtures and reducing their radiant power to 40% in average of the initial
level are a very efficient ways to reduce sky brightness. Under a clear sky, the reduction of artificial sky
radiance at zenith was ~50% for the 546 nm Hg line and ~30% for the 569 nm Na line. The reduction
achieved is quite high considering that only the nearby lamps were targeted for replacement and only a
portion of them were actually replaced. The level of artificial sky radiance in the 546 nm Hg line came to
be comparable to 1979 levels. For the 569 nm Na line, the target of reducing by 50% was not reached, but
the reduction is nevertheless significant.
The contribution to the sky radiance from large and distant cities is lower than expected. In 2003 , it
was estimated that Sherbrooke was contributing up to ~25% of the artificial sky brightness experienced
by the MMO, but our study suggests that, in 2005, it was of the order of 1% at zenith. On the other hand,
the 2003 estimates for zones 1 (25 km radius from the MMO) and 2 (25 km to 50 km from the MMO)
were quite accurate, 50% for zone 1 and 25% for zone 2, respectively, whereas we report ~56% for zone 1
and ~23% for zone 2, on average for the two lines.
Our results also highlight critical zones where a change in the lighting infrastructure is likely to have a
more significant impact on sky brightness at the MMO. The contribution and sensitivity maps developed
are useful cartographic tools to help authorities manage their lighting infrastructure in such a way to
reduce sky brightness and its adverse effects. The results have been shared with MMO officials and are
being used as a tool to improve sky quality at the observatory.
In 2011, the ASTROLab had to re-launch its light pollution abatement project in order to curb the
installation of non-compliant light fixtures, which deteriorate the quality of the night sky and may
endanger the sustainability of the reserve. The recent massive introduction of Light-Emitting Diodes
(LED), with all of their advantages and disadvantages, represents a major new challenge for the ISDR to
tackle. Projects are currently underway to investigate and monitor this type of lighting, as well as to limit
its potential contribution to sky brightness. Our modeling approach can be adapted to do further research
on this topic to determine whether there are grounds for restricting LED street light installation.
We want to thank the students from the GRAPHYCS group at the Cégep de Sherbrooke for their help and
ideas. We also want to thank Diane Gabay for some of the proof reading assistance. Mathieu Fréchette
was very helpful for his technical assistance and, finally, the staff from ASTROLab for sharing their ideas
and expertise. Part of this research was funded by the Fonds québécois de la recherche sur la nature et la
technologie and by the Cégep de Sherbrooke's CERTEE program.
 Rich C, Longcore T. Ecological consequences of artificial night lighting. Washington DC: Island Press.
2006, 478 p.
 Navara, K. J., Nelson, R. J. The dark side of light at night: physiological, epidemiological, and ecological
consequences. Journal of Pineal Research 2007;43:215-224.
 Haim A, Portnov BA. The Light Pollution as a New Risk Factor for Human Breast and Prostate Cancers,
New York, Springer, 2013, 168p.
 Legris C. Projet pilote de conversion de l’éclairage public et privé dans la region du Mont-Mégantic.
Rapport Hydro Québec, 2008.
 http://astrolab-parc-national-mont-megantic.org/en/welcome.htm Accessed 2013 June 12.
 Giguère S. Le parc national du Mont Mégantic, De la terre aux étoiles, Muséologie In Situ 2012.
 Legris C., Goulet P. Plan d’action : Projet de lutte contre la pollution lumineuse dans la région du Mont
Mégantic, august 2003.
 Aubé M, Franchomme-Fossé L, Robert-Staehler P. Light Pollution Modelling and Detection in a
Heterogeneous Environment: Toward at Night Time Aerosol Optical Depth Retriaval Method. SPIE’S
Annual Meeting Proceedings, San Diego, CA 2005; (paper 5890-40, 1.1 Mb).
 Aubé M. Heterogeneous modeling of artificial sky radiance:
http://www.cegepsherbrooke.qc.ca/~aubema/index.php/Prof/IllumEn Accessed 2013 June 16.
 Aubé M. Illumina Google Code project: https://code.google.com/p/illumina/, Accessed 2014 January 07.
 Aubé M. Light Pollution Modeling and detection in a heterogeneous environment, Proceedings of
Starlight 2007 conference. 2007; La Palma, Spain.
 Aubé M, Kocifaj M. Using two light-pollution models to investigate artificial sky radiances at Canary
Islands observatories, Monthly notice of the royal astronomical society. 2012; 422(1): 819-830.
 Shettle EP, Fenn RW. Models for the Aerosols of the Lower Atmosphere and the Effects of Humidity
Variations on their Optical Properties. 1979; AFGL-TR-79-0214.
 Vermote E F, Vermeulen A. Atmospheric correction algorithm: Spectral reflectances (MOD09), ATBD
version 4.0., NASA contract NAS5-96062, 1999:
http://mod09val.ltdri.org/publications/atbd_mod09.pdf Accessed 2014 January 11.
 Farr T, Rosen PA, Caro E, Crippen R, Duren R, Hensley S, Kobrick M, Paller M, Rodriguez E, Roth L,
Seal D, Shaffer S, Shimada J, Umland J, Werner M, Oskin M, Burbank D, Alsdorf D. The Shuttle Radar
Topography Mission, Reviews of geophysics. 2007; 45:33p.
 Aubé M. Sky spectral radiance database:
http://www.cegepsherbrooke.qc.ca/~graphycs/aubema/recherches/data_usage.html Accessed 2013 August
 Aubé M. Light Pollution Data Server: http://galileo.graphycs.cegepsherbrooke.qc.ca/lpds/
Accessed 2013 August 30.