A Baseline Assessment of Light Pollution Impact

On Anglesey, North Wales, UK.

John Rowlands, FRAS.
Spring 2015
This work, except where otherwise stated, and including all associated image and data files, is the copyright,
all rights reserved, 2015, of John Rowlands FRAS, released under a no commercial use licence to Anglesey
County Council’s Countryside and AONB Office.

AIMS

 To provide the first objective ‘baseline’ assessment of night sky quality ('darkness')
from multiple sites spanning the island.

 To critically assess different potential, simple methods for assessing night sky
quality.

 To develop a ‘standard method’ for such an assessment

 To identify and assess various factors influencing measurements

 Where possible, comparison of output and methods with other areas.

Anglesey only has darker, not dark skies. The island has its own, and is surrounded by, a sea of
light pollution. 2012 Image: NASA/NOAA NGDC Suomi NPP Visible Infrared Imaging Radiometer Suite.
CONTEXT: HOW DARK IS A DARK SKY?

Before we begin, let’s define a very abstract and hopelessly unfriendly unit used
extensively throughout this report. This is magnitudes per square arcsecond (mpsas.)
The arcsecond is a tiny angle of sky, equal to 1/3600th of a degree, and ubiquitous in
astronomy. The term magnitude is simply a measure of relative stellar brightness; the
brighter the star, the lower the magnitude. If we took a star and spread out its light so that
it didn’t appear to come from a point source, but instead from a much more extended area
of the sky, then that is the idea behind this unit of measurement.
In our case, this extended area measures 1 arcsecond by 1 arcsecond. The meter we will
eventually go on to use doesn’t measure such a tiny area. Rather, it ‘looks’ at a very much
larger area of sky, of diameter 42 degrees, and converts that integrated signal into an
average magnitude per square arcsecond. The brighter the sky, the lower the numerical
output from the meter (because this would be equal to a very bright star’s light, spread out
over the square arcsecond.)
Different people have different perceptions of what constitutes a truly dark sky. If one
person moves from the inner parts of Manchester, say, to the immediate surrounding
countryside, he may think the sky to be dark and a good one from which to observe the
cosmos.
A person who has lived in a very remote area of Scotland, on the other hand, will judge the
first person's 'dark' sky to be a badly light polluted one.
Not until one visits an unpopulated area with no light pollution within hundreds of miles or
more can one appreciate just how much modern society has degraded the darkness of the
night sky. From such sites, even dim stars are visible all the way to, and in fact define, the
horizon, giving a distinct and unusual feeling of being aboard 'spaceship Earth'. Otherwise
washed-out features become prominent, detailed and bright. An entirely unpolluted night
sky is something that very few people in the western world have ever experienced, and
those venturing to assess the sky quality must consequently be very careful in what they
perceive as darkness.
When one examines ‘officially-recognised’ dark sky sites within the UK, their night sky
assessments (at the zenith) typically result in an International Dark Sky Association ‘Silver’
award (Sky Quality Meter) output of at least 21 mpsas. However, if one visits such sites, it
becomes readily apparent, even by casual examination, that these are not sites devoid of
all light pollution. Indeed, this is in agreement with the definition of a ‘Silver’ site, which
allows a fairly significant degree of light pollution, certainly enough to be obvious in the
form of light pollution ‘domes’. Publicity photographs from such award sites often contain
evidence of light pollution from more distant sources.
It seems that IDA, like many of us, has - either deliberately or otherwise - settled for a
compromise, in that these ‘Silver’ and ‘Bronze’ sites can be seen as simply the best one
can practically hope to experience from the highly-developed, western world. In other
words, it provides, in the most part, for areas where a ‘better’, but not truly dark sky can be
experienced by the largely urban and suburban population. Whilst ‘Gold’ sites are
certainly extremely dark, even they do not always represent a truly and entirely dark sky
where the only light is that from the cosmos. Light pollution domes may still be evident
from ‘Gold’ sites, albeit weakly and at very significant distances.
 -2
The label on the SQM’s front. The mag arcsec scale is logarithmic, which might lead the unwary to believe
there is little difference between small increments in sky darkness.

This realignment of darkness also holds for the interpretation of light meter (SQM) output.
Upon the face of the SQM unit is printed a simple, linear scale running from 16 mpsas
(polluted) to 21 mpsas (quite dark). This might suggest, to a lay user, that 21 mpsas is
really a very dark sky indeed, and the best that can be achieved. However, this is to
ignore the logarithmic nature of the scale, which is highly unfamiliar to the general public.

Small changes in output indicate large effects in practice. For example, a sky returning a
value ‘just’ 1 mpsas brighter than another is, in reality, two-and-a-half times as light
polluted. Unwary users and policy makers, however, may well believe the ‘small’
difference is inconsequential. It was found during preliminary testing of the SQM-L at
Parys Mountain that low, flat and featureless stratus cloud during an extended period of
high pressure in February 2015 that was obviously lit to the zenith by surrounding
settlements, returned a consistent value of 19.9 mpsas – hardly the ‘quite dark sky’ that
the scale might superficially suggest.

Unfortunately, all this does risk redefining darkness for a modern world. There is a clear
danger of a relatively low magnitudes per square arcsecond output from an instrument
coming to be seen as a definitive indication of a ‘truly’, or ‘acceptably’ dark sky, when a
simple look around the sky with the unaided eye might – and invariably will - suggest
plenty of work yet to be done to tackle sources of stray upward artificial light.

It is for society to decide whether, in the end, it wishes only to have a ‘rough idea’ of what
the night sky looks like, or have much more stringent control of lighting such that it is
capable of seeing the cosmos in all its glory. Much of the glory is, sadly, already lost on
Anglesey.

All this highlights the need for a robust, objective means of assessing the actual degree of
darkness – or, alternatively, the degree of light pollution - of any given site.
MEASURING THE IMPACT: VISUAL ASSESSMENT.

Whilst the human eye and what it sees is certainly not immune from error and inter-
observer variability, it does nevertheless offer a simple means of assessing night sky
quality that, unlike the rather abstract units of the SQM, reflects the real, human
experience of the night sky. Visual assessments can more meaningfully comment on the
existence of light domes, whether street lights are visible in the distance, their nature, and
so on. It is quantifying those impacts, however, that remains the obstacle to objective
visual work.

We would ideally need to have each observer's eyesight tested and any deficiencies
corrected to reach any conclusion about just how reliable the results were. But, one might
conclude that, if we can obtain a number of independent assessments (not groups, who
tend to influence one another) to assess night sky quality visually, then any inter-observer
error will be smoothed out and a 'consensus' view of how dark the sky is can be reached.

There is also a need for caution to ensure pre-formed ideas of how dark a sky might be,
perhaps from a belief concerning the degree of development in the area, do not come to
influence the outcome.

Enhancement techniques sometimes used by amateur astronomers, such as

hyperventilation, are not remotely normal viewing practice for the general population, and
so ought not to be used, not least because deliberate hyperventilation can be very
dangerous.

Two main methods can be employed visually. First is the very long-used stellar limiting
magnitude method, where the dimmest star visible in a specific field of view is recorded.
This does tend to suffer from variability between observers, especially if some are
inexperienced. It is also very time-consuming, necessitating some 40 minutes for full dark
adaptation prior to observation. For this project, where only one person is available for the
work of covering many sites, this becomes highly impractical due to the loss of night
adaptation from car lights when moving from one site to another, trying to keep ahead of
weather changes. Such a method would take months or years to complete, against a
backdrop of ever-changing lighting in the area.

The other visual method is the 9-point Bortle scale, devised as a ready-reckoner to overall
sky quality. The scale assesses the general appearance of a number of different sky
phenomena. It was first published by John E. Bortle in a February 2001 article within the
US-based Sky & Telescope magazine1. As such, it was not intended to be, and was not at
the time, a peer-reviewed methodology text.

Bortle ‘1’ is the darkest, pristine sky; Bortle ‘9’ is that which you would experience in a
brightly-lit, inner city setting.

The Bortle scale, which is admittedly a good and accessible attempt at sky quality
assessment, and created in good faith, does have many problems.
Some of these are comprehensively covered, amongst other matters, within a 2014 paper
by Andrew Crumey2. There is criticism of the very low magnitude limits of 7.6-8.0 for a
Bortle scale ‘1’ (pristine) sky, which the author and, indeed, most experienced observers,
conclude is considerably beyond that detectable (ca. magnitude 6.13) by a typical human,
naked-eye observer looking at real stars under a real sky (as opposed to artificial and
unrealistic arrangements of specially blackened foregrounds and viewing ports.)

The International Dark Sky Association, notably, restricts its own claim of the limiting
magnitude for the very best dark sky sites (‘Gold’ standard) to be not lower than 6.8, again
calling into question the exceptional claims by Bortle.

When using the Bortle scale locally, it is important to detach oneself from the tag given to
various degrees of human development (‘urban’, ‘suburban’, etc.), and use only the
description of astronomical objects for each category. For example, much of Anglesey
might fall into the ‘suburban’ Bortle sky quality category, but that we might, more locally,
resit calling it such, and instead consider the area to be ‘rural’ in nature. Bortle assigned
these tags according to his own experience of the eastern US, which must be taken into
account, and not allow assessments to sway the output one way or another.

Further consideration of the scale leads one to conclude that there is a seasonal bias in
the objects listed, indicating Bortle selected only northern hemisphere summer and early
autumn objects. There are no mid-winter objects, despite this being the darkest period of
the year, when one would be most likely to conduct a night sky survey. The scale is
unusable without fundamental revision for the southern hemisphere, as it contains no
southern hemisphere objects.

Bortle appears to have been resident in New York State at the time of developing his scale,
at a latitude of about 42°. This is some 8° (or, about 500 nautical miles) further south than
the most southerly point of the UK; a similar latitude in Europe would be at the
Spain/France border. This means that whilst twilight conditions that prevent darkness
assessments exist all night during most of the UK summer time, twilight duration and depth
are much different when observing from New York State. This is a fundamental problem of
incompatibility in the Bortle scale, and falls into the frequently-encountered problem of a
US-centric bias.

Limiting magnitude (left) yielding the Bortle scale (right.) The very low magnitude limits cause many to
seriously question whether most humans can detect them under real skies, as opposed to laboratory
conditions.
The Bortle scale requires observation of several sky phenomena and objects over perhaps
a few months, and is not very suited to ‘spot checks’ over many sites on a single night.

Clearly, there is much to do in order to considerably refine and adapt the Bortle scale into
the future and for areas other than the US. This is well beyond the scope of the present
study.

In practice, and within the resources of this particular assessment, several observers for a
limiting magnitude or Bortle survey are simply not at hand to help, more especially as clear
weather may demand observations at very short notice. Moreover, the project resources
do not allow for the very long time it takes to conduct a properly dark-adapted, visual
assessment from so many sites, using several experienced observers, taking perhaps
several years to complete.

MEASURING THE IMPACT: INSTRUMENT-BASED ASSESSMENT.

Photography is one simple and highly effective means of demonstrating light pollution.
However, this typically requires expensive ‘fisheye’ lenses to show the whole sky in one
image. It also requires calibration of the sensor for each of the very many camera types
available, such that an exposure using one camera is not only comparable to an exposure
using another camera, but that the resulting image is broadly representative of the naked-
eye view.

Currently, the more expensive digital SLRs are extremely sensitive to low light, and can
yield skies that appear to hold impressive levels of astronomical detail. Invariably,
however, these images considerably exaggerate the detail that is actually visible to the
unaided eye, whilst also exaggerating the impact of light pollution.

As a result of these problems, and that funding did not extend to the purchase and
calibration of camera equipment, this method was not adopted. However, images were
captured using existing, lower-quality equipment that, whilst uncalibrated, are nevertheless
considered to give a very reasonable indication of the effects of light pollution (but not the
appearance of stars, which are under-indicated) to the naked eye.

Over the past few years, the use of affordable electronic sky quality meters (SQMs) has
become very popular. SQMs are also the de facto standard instrument for assessing night
sky quality for the purpose of gaining an International Dark Sky Association award at one
of three levels. Only a zenith (overhead) measurement is required by IDA, but this can be
inadequate, as will be discussed later.

Despite their popularity, SQMs are not without potential problems.

 The SQM-L (‘lensed’ version)

The meters employ a TAOS TSL237 (side-looking format) silicon photodiode sensor with a
Hoya CM-500 infrared blocking filter to assess sky brightness within the visual spectrum,
yielding an eventual output in magnitudes per square arcsecond (an arcsecond is 1/3600 th
of a degree.) A clear plastic protective panel sits over the detector and blocking filter. The
characteristics of this panel are unknown.

A bright sky results in a smaller numerical magnitude, a dark sky a higher magnitude
value. The scale is logarithmic, such that that a difference of one unit equates to a
brightness ratio of 2.512; a difference of 5 magnitudes thus equates to a brightness ratio of
100 (2.5125.)

The instrument used for this study was the Unihedron SQM-L, manufactured in Canada.
This ‘lensed’ version reduces the field of view of the unlensed version from near all-sky to
a narrower cone of diameter 42°, with a Full Width Half Maximum sensitivity stated by
Unihedron (following Cinzano4) as “~20°.” The Half Width Half Maximum is stated as
“~10°.” This is stated as yielding a sensitivity to a point source at 19° off-axis being 10
times lower than on-axis. The un-lensed version has almost a fisheye, whole hemisphere
coverage, leading to potential errors from direct artificial light sources on and near the
horizon (nearby settlements, cars, etc.)
 4
Angular response of Unihedron SQM-L according to Cinzano .

Approximate field of view for SQM-L, as projected near the constellation Orion.
As with any electronic device, the output of a superficially accurate number onto a screen
does not necessarily mean that such accuracy exists; calibration on a continuous basis is
ideally necessary, although periodic recalibration is widely accepted as adequate in most
avenues of scientific and industrial activity.

Unihedron has somewhat nebulous terms for defining the performance of their meters,
stating:

"Each SQM-L is calibrated using a NIST-traceable light meter. The absolute

precision of each meter is believed to be +/- 10% (+/- 0.10 mpsas.)
The difference in zeropoint between each calibrated SQM-L is typically
+/-10% (+/- 0.1 mpsas.)"

Terminology is everything, and it is clear that the above carries no statement about the
accuracy; it merely indicates the precision (and a believed, rather than demonstrated
precision at that.) Precision is the repeatability of the output when measuring the same
brightness - which is a very different matter to accuracy, which is how close to the true
value the output is.

Unihedron, whilst asserting “NIST-traceable calibration” for each instrument produced,

nevertheless also assert in their literature that no calibration certificate is supplied. The
end-users, and those then relying on the SQM output, thus have no means of readily
knowing how accurate the meter is. Precision can be determined through use, but
calibration data would be very useful. There appears to be no reason why Unihedron could
not quote the calibration output for accuracy and precision for "each instrument" they claim
to determine, together with the details of that calibration device. Indeed, it would provide
considerable and necessary assurance to the increasing number of individuals and
organisations (including the International Dark Sky Association) using and relying upon the
instrument to inform policy and gain ‘dark sky’ recognition.
 SQM-L Weighted response with increasing zenith distance4.

A study by Cinzano4, which appears to be an internal, non-peer reviewed draft containing

only references to previous studies by the author himself and published on the Unihedron
web site, does highlight inter-unit calibration variability amongst three tested units; one unit
was deemed to "likely be out of calibration."

The units tested by Cinzano were within seven serial production numbers of one another,
so variability over production time and age of the units cannot be assessed. We can only
say from the Cinzano report that an out-of-calibration rate of 33%, if it extends to
production in general, ought to be a cause for concern.
 Transmission characteristics of the HOYA CM-500 IR-blocking filter (source: HOYA)

Concern must be expressed that, whilst the Cinzano study may be reliable it has, as a
non-peer reviewed, draft study been cited in a conference presentation published in the
Journal of Physics, Conference Series5 as indicative of the accuracy and performance of
the SQM units. Similarly, Cinzano is cited as providing assurance of the reliability of the
SQMs for a Master's thesis6 thus:

"All the SQM data are comparable, and that is assured by a calibration process
done during the production phases."

One can only refer the present reader to the caution about the potential limitations of the
Cinzano study before accepting such statements as meaningful, and the fact that, despite
these studies, no calibration claimed to be undertaken on each unit is made available to
those purchasing them. That can only be said to be unsatisfactory.

These points were put to the co-designer of the SQM, Mr. Andrew Tekatch, in an e-mail
exchange commencing 19/2/2015, who was extremely open and honest in his very prompt
response. Mr. Tekatch did not challenge the view that Mr. Cinzano’s reports were non-
peer reviewed. He also did not challenge the view that one out of three SQMs had been
found to be out of calibration in the Cinzano study, and that this, as a very limited study,
was the only inter-unit calibration information available.
He did point me to a useful paper that compared nine individual SQMs and found inter-unit
variability (in output reading) of -7% and +9% after intercalibration. Mr. Tekatch accepted
that better calibration information should be available to the end-user of each unit, but that
no affordable means of providing this had yet been found. He had not been advised of
significant outliers, which the present study’s unit very infrequently produced (and were
discounted from the results.) I requested further clarification on what aspect of each SQM
was, in fact, calibrated at the manufacturing stage. The response from Mr. Tekatch
explained that each unit is tested at a testing instrument-dictated 8 magnitudes per square
arcsecond source, and that it is this source that is NIST-traceable calibrated. Each unit
does, therefore, undergo calibration, albeit only at one, bright datum. A second calibration
point is effectively available by each end-user measuring the dark current, although in
practice, very few do so.

A point of current interest arising from the spectral response curve for the SQM indicated
by Cinzano is the lower response to bluer light. This may be significant for sky brightness
assessments more recently and into the future, due to the increasingly widespread
installation of white-light LED street lamps, with strong emission around 450nm. The SQM
has, according to Cinzano, a response of ca. 0.75 at 450nm. There appear to be few
other assessments of SQM spectral response, but Shariff et al7 find the spectral response
at 450nm to be significantly lower, ca. 0.6. Overall, the SQM-L meets the criterion of a V-
band (visual spectrum) response with a 551nm midpoint.

Because SQMs are often used as 'point and shoot' devices, with no real appreciation for
careful use by prospective developers and lay operators, there is plenty to suggest caution
when data from SQM sky assessments are used, more especially when this involves
'public outreach' campaigns where further inter-operator weaknesses are very likely to
affect the output. It is not always stated which SQM was used, nor what conditions existed
during the assessments. In many cases, operator-based errors are likely to be significant.

As this is a baseline study, it is necessary to appreciate that future sky quality meters may
and can be expected to utilise different, possibly better detectors with perhaps different
acceptance ('viewing') angles, different spectral responses, etc. All these need to be taken
into consideration when comparing the sky quality of tomorrow with the sky quality of
today.

Overall, one would have to treat with great caution any claim that the current SQMs are of
known accuracy calibration to the end-user, although those few assessments of the
instruments in formal studies and user experiences suggests that they are not wildly
inaccurate or imprecise. However, no user should, in this author’s view, quote a digital
output from an SQM are being definitive and beyond challenge; SQMs are very much a
work in progress and the skill, or lack thereof, of the user always has to be taken into
account. It would strongly appear that the astronomical community must immediately
subject SQMs to widespread, rigorous, independent and peer-reviewed testing, and that
Unihedron should provide unique calibration data with each unit. Standard methods
beyond zenith-only assessments also need to be developed.
METHOD DEVELOPED AND ADOPTED IN THIS STUDY.
Despite the identified problems, we must make the best of things as they stand today to
obtain a reasonably reliable baseline assessment, providing sufficient information to future
assessors so that meaningful comparisons can be made.

Of the few methods available, the SQM-L was considered the most objective and
repeatable means of assessing Anglesey’s night sky. The ability to characterise the
SQM’s underlying architecture in detail is of fundamental importance for present and future
studies, and is fortuitously made possible by ample technical specifications published by
the detector’s manufacturer, and that of the IR blocking filter. The overall SQM circuit is
very simple, and there is no reason to think that detectors in the foreseeable future will be
based on a fundamentally different design.

An SQM-L, serial number 8606, was purchased in the UK (First Light Optics, Ltd.) Dark
current (a ‘blank’) was obtained by taking seven readings in a darkroom with the detector
also covered by a dense, folded matt black cloth. After a median of 65 seconds, the meter
yielded a result of 24.00, +/-0.01 mpsas, which is in essentially perfect agreement with true
darks taken by the few others who have thought to do so (Exmoor: 24.00; Brecon: 24.02.)
Normal sky readings under reasonably dark skies take only about 4 seconds. The dark
current reading can be taken as a known end-user calibration for accuracy, which
suggests very good accuracy overall.

Weather must be taken into account before any night sky assessment takes place, and
this was found to be much more complex than expected. The direction of tropospheric
wind flow is a critical consideration. Easterly winds tend to bring significant particulate
pollution from, principally, lignite-fired electricity generating stations in Germany and
Poland to Anglesey, rendering clear skies extremely hazy due to enhanced light scattering
from lights at ground level. Temperature inversions can simultaneously enhance or reduce
local light pollution, depending on whether the observer is on low-lying land, or elevated
ground above the trapped pollution and mists, respectively. Changes in humidity levels
can also modify the degree of light pollution, but no particular value of humidity can
realistically be chosen if one wants to proceed in a timely fashion with a limited-resource
study.

Indeed, such are the number of possible variables that may influence the exact outcome of
sky conditions from night to night, that this warrants its own, separate study.

Winds from the west to north quadrant usually bring the clearest weather with lower
humidity, and skies behind recently –passed cold fronts can bring particularly clear air.
However, a careful examination of surface wind flow against sky transparency during this
study confirmed that it is the source of the air, and not simply the direction of the local
wind, that is of critical importance in determining any given night’s sky quality.

For example, a NE wind on Mar 14 had a source in the eastern Baltic. On the morning of
the 14th, the surface visibility was a moderately good 30km. This flow bypassed central
Europe, flowing directly from the Baltic, across Denmark and the North Sea to north
Wales. Accordingly, the wind flow picked up almost no particulate pollution from the
heavily-industrialised nations of central Europe and was very clear. During the following
two days, the wind direction remained much the same NE – E, but the source of the wind
shifted slightly to encompass all of Germany, with a convergence of winds over Belgium
that led to a concentrated haze over much of the UK – including Anglesey (8am surface
visibility 16/3/2015 = 9km (author’s weather station data.))

It is tempting to limit the study to only the clearest airflow conditions, but one must ask
whether or not this is a representative approach. As the present study received funding
only a month or so ahead of the vernal equinox, and that the Moon precluded observations
for days on end during this short period, time was very much of the essence in gathering
data. Using baseline data gathered at the start of this work, a sky quality assessment
session commenced whenever there was no cloud, a light wind or stronger (to avoid
mists), and the SQM-L reading at Parys Mountain (zenith) was within -0.05 mpsas of 21.25
mpsas, this being the typical value obtained under good, clear westerly flows, although
under exceptional conditions, the Parys zenith value could darken by a further ~0.15
mpsas. Whilst this darkening was likely real, it nevertheless remains broadly within the
expected performance limits of the instrument.

A very limited, indicative assessment of the effects of a more easterly wind – in this case,
NE - where the source of the airflow was southern European Russia and northern central
Europe, was undertaken during the night of 13/03/15. This yielded surprising results in
that the impact of light pollution was seen to be affected strongly according to direction.
Light pollution from Amlwch, ~ 2km to the north of the sampling site, increased by 42% at
45 degrees zenith distance. This was clearly visible to the unaided eye as a column of
strong light pollution, rendering details, such as water pools on the ground, clearly visible.
However, there was a significant decrease in light pollution levels to the south – a drop of
13%, with the zenith value also decreasing in brightness by 16%. Values to the east and
west remained unchanged at the expected precision level of the instrument, although they
might be expected to have increased in brightness. This could indicate that the extinction
due to haze at and near the surface leads to lower light pollution from more distant
sources, whilst nearby sources are significantly enhanced in their impact. Considerable
additional work is needed to more carefully examine what is happening during these haze
events.

On the night of 20/21 March 2015, an anticyclonic circulation with a small, localised centre
in the Atlantic east of Iceland, yielded a light northerly wind (~5 kts maximum), which drew
air directly from the far Arctic. This produced the best dark sky conditions seen during the
entirety of the study, where darkness at the zenith increased by ~16%. Again, the likely
instrument performance limits of ~ +/- 10% must be borne in mind.

This highlights a problem: should we wait for the best night possible before conducting an
assessment? When could we conclude that this condition had, in fact, been reached such
that no further improvement in darkness could be realised? Would this be simply a bias
similar to never measuring under particulate-induced hazy yet otherwise clear skies?
From the author’s experience, it is likely that the 20/21 March 2015 event is quite rare, and
is at or very close to the best that will ever be realised on Anglesey at this point in its
development. A very limited study of a small number of sites could give a ‘best case’
summary of the island’s light pollution condition. But it would not represent the ‘typical’
experience of Anglesey’s night sky, which remains that arising under a generally westerly
flow, with a source out in the mid-Atlantic.

The study took care to examine the impact of the Moon prior to moonrise/after moonset.
Even at 100% phase, the Moon started brightening the zenith only when about 2.5
degrees (~5 lunar diameters) below the horizon prior to computed (atmospheric refraction
not included) upper-limb moonrise. This was doubled for 'safety', measurements being
taken at all Moon depression angles lower than 5 degrees (~10 lunar diameters.) The Sun
always had to be below 16 degrees (it being the definition of the end of astronomical
twilight.) In practice, it was found necessary to allow the Sun to sink to about -18 degrees
to be sure of no twilight effect caused by normal refraction and abnormal thermal delays,
compared to software computations not including these effects.

As Anglesey lies close to several commercial air corridors, care was taken to ensure no
aircraft navigation and strobe lights passed through the field of view of the instrument
during measurements. Care was taken to ensure no satellite flares entered the field of
view. All this was easily achieved because only ~6 seconds was required to secure each
reading in dark sites, and less than 1second in brighter areas.

One must incidentally note that in all dark sky assessments seen by the present author,
none make reference to the actual depression angles of the Sun and Moon at the time of
measurement, and many lack any indication of time or even date of each measurement.
In some assessments, cloud or mist has been present, potentially interfering with the
output. It is clear from this that critical factors are not always considered by those
assessing the night sky, rendering some of them suspect, and a few potentially worthless.

At each site, a total of 35 readings were taken, comprising seven zenith readings of the
sky, together with a further seven readings at each cardinal point at 45 degrees elevation,
giving an indication of direction of main polluting sources and increase in pollution levels
with increasing zenith distance. The median value was obtained for each point, thus
reducing the effect of any large outliers, though there were few of these in practice. The
overall, zenith plus cardinal points sky quality was determined by an aggregate arithmetic
mean of all 35 measurements at each site, thus allowing the zenith ‘outliers’ (being much
darker than the 45 degree values), to influence the outcome.

For all readings, the SQM was tripod-mounted using a simple mount purpose-built for the
study, ensuring stability and rapid, reliable repeatability in zenith and 45-degree pointing
through the use of a pre-aligned, double-hinged cradle. A handheld device is not
considered adequately stable to provide good, repeatable results, as it generally fails to
provide reasonable accuracy in pointing in either dimension. In particular, there is a
tendency, increasing with age, to point at an area significantly away from the true
overhead point when simply holding the unit aloft. Attempting to guess any other angle by
hand leads to greater pointing errors. Cardinal points (‘true headings’) were established
simply by aligning a white index mark on the mount with Polaris, and then other points
positioned by use of a spirit-levelled turntable – a recycled polarising microscope stage –
graduated in degrees.

Measuring points (illustrative only): Zenith, and 45 degrees elevation at each of cardinal points.

A recycled polarising microscope stage mounted on a tripod head, for simple, repeatable and accurate
azimuth and altitude pointing.
The meter was placed in a car to cool to broadly ambient temperatures at least 30 minutes
before an observing session began. In practice, it was found that this cooling period was
not at all critical, but in transit between sites, in-car heating was kept to a minimum. The
pre-aligned nature of the mount was very successful and allowed these readings to be
completed very quickly and with little potential for error in alignment due to working in the
dark and being tired. The above image and comment was posted to the International Dark
Sky Association's Facebook page in early March 2015, with positive comment from IDA.

The observer operated the instrument wearing very dark grey clothing such that the
potential for any reflected light was minimised. The observer also remained behind the
instrument, and so entirely out of its field of view during measurements.

Observing sites were such that there was no high ground, trees or hedges to obstruct or
remotely call into question that the field of view of the instrument was anything other than
entirely clear.

The potential influence of body heat was assessed by standing underneath the
instrument’s viewing port during several measurements, but no significant effect was
noted. The battery life of the instrument was also extremely good, with no deterioration
after the majority of sites had been visited once and some 2000 point measurements
taken. A new battery was installed for the evening session of 21/3/2015, which yielded no
apparent change in performance characteristics.

During discussions with NRW, a comment was made than an entirely objective study
would also assess readings of night sky darkness under cloud cover. However, this study,
which has the promotion of tranquility tourism as one aim, is concerned only with the
question of how dark the night sky is when there is zero cloud cover and when visual
observers would in fact observe the stars. The assessment is not invalidated by being
limited only to cloudless conditions. However, from Parys Mountain, low stratus cloud
during high pressure, zero wind conditions and no Moon, yielded and indicative median of
19.9 magnitudes per square arcsecond.

Finally, digital images were taken at some illustrative sites, with an exposure that
approximated the visual experience, using a ‘fisheye’ lens on an old Nikon 4500 compact
digital camera of 4 mega pixel resolution. This was to obtain some indication of the
distribution and general source of light pollution only, as calibration of images to be truly
representative of the naked eye view was well beyond the resources of this project. For
this camera and lens combination, an exposure time of 8 seconds at f/2.6 was found to be
acceptably close to the actual human experience of the impact of light pollution, albeit with
fairly high sensor noise levels. The images are likely of adequate quality, and certainly
representative enough to use in an IDA submission.

It was noted that some submissions for awards to IDA use images that considerably
exaggerate, through very sensitive digital imaging and long exposures, the degree to
which stars (and any light pollution) are visible to the human eye. This should be avoided.
OTHER FACTORS FOR FUTURE ASSESSMENTS.
It is well-established that solar activity, which follows a roughly 11-year cycle, influences
the brightness of the background sky through varying ionisation levels of the upper
atmosphere, and so ‘airglow’. Auroral light is also a not-insignificant factor, even from
Wales. At the time of this initial assessment, the sun is at Cycle 24, a weak cycle showing
a double peak maximum, very near the end of the second peak. Auroral glows have been
evident on many nights, just prior to the study, to the north from Anglesey. Future
assessments will need to ensure there is no aurora or airglow present, to ensure solar
effects on sky brightness do no complicate comparisons with the past (and future.) These
effects can fall and rise within minutes, so an awareness of potentially rapid changes to
sky conditions should be present. During this study, no airglow or aurorae were evident,
and geomagnetic conditions were quiet. On one evening, during a repeat assessment of
Lynas, weak auroral glow was evident both from measurements and visually. This did not
affect readings in other directions.
Results.
The SQM is a very good, practical and relatively easy-to-use instrument. But it did show
peculiarities that can, unless used with caution, skew the resulting data. It was noted that
the first two to three, sometimes more readings, as the manufacturer acknowledges,
tended to have wide variation, but then settled down to extremely good precision that was
at its best when seven readings could be taken in rapid succession, presumably as this
ensured a near-constant temperature within the circuit.

The output of the SQM-L for the zenith and all-sky, across the island, are presented below.
The results are not surprising, showing that the Menai Hub, A5, Llangefni and Holyhead
areas, together with the combined impact of smaller, well-scattered settlements, yield a sky
that is everywhere affected by light pollution to at least a moderate degree. The darker
sites should not be interpreted as defining a truly dark sky, and indeed, they remain sites
affected quite obviously by artificial light.

The maps are accompanied by distribution plots of sky quality ranges both at the zenith
and for all-sky assessments.

The zenith-only sky quality distribution plot shows that 45 of

the 60 sites (75%) are of International Dark Sky Association
‘Silver’ standard. 15 of the 60 sites (25%) are of IDA ‘Bronze’
standard. The majority of the island thus has ‘Silver’ standard
skies, which are of fairly low light pollution impact.

The all-sky assessments reveal a very different outcome. Here,

only 16 of the 60 sites (27%) meet an IDA ‘Silver’ standard, with
44 sites (73%) being of ‘Bronze’ standard. Under this
assessment regime, most of Anglesey has ‘Bronze’, moderately
light polluted skies. The IDA scheme does not formally extend
to non-zenith values, but the overall outcome is not invalid.
Zenith values from the SQM-L in magnitudes per square arcsecond. The decimal point in each value approximates the
location of the assessment site. Below: distribution of zenith values in ranges. Orange is IDA ‘Bronze’ standard, grey is
IDA ‘Silver’ standard.

18
Zenith Distribution/mags per sq arcsec

16
Orange=IDA
'Bronze'
14
Grey = IDA
'Silver'
12

10

8
n

SQM-L Output Range

Colour-aided interpretative map, as earlier, for zenith-only assessments. Black does not indicate a properly
dark sky, only that here, light pollution is relatively lower. Boundary limits are different from the all-sky
values.
All-sky quality assessment based on zenith plus 45 degree elevation at each cardinal point. Values in magnitudes per
square arcsecond. Below: distribution of all-sky values in ranges. Orange is IDA ‘Bronze’ standard, grey is IDA ‘Silver’
standard (this standard is not formally applicable to non-zenith values.)

All-Sky Quality Distribution/ mags per sq arcsec

16

14

12

10

n
6

SQM-L Output Range

 Colour-aided interpretative map, as earlier, for all-sky assessments. Black does not indicate a
properly dark sky, only that here, light pollution is relatively lower. Boundary limits are different
from the zenith values.

After a short period of being switched off, it was found that somewhat higher variability in
the output usually returned, and the cycle of discarding the first few readings had to be
repeated each time in order to reach good stability. Occasional outliers of more than 0.5
magnitude arcsec-2 in either direction would be produced at random, which might indicate
some issue other than temperature instability. The only study examining thermal stability
found the SQM to remain stable even under extreme and rapidly-fluctuating conditions8.

Such anomalous outliers were also noted in data obtained from Snowdonia’s survey,
where the downstream use of arithmetic mean, rather than the median value, has allowed
them to exert a modest skew on the data. The present study discarded the few wide
outliers from the recorded data as being anomalous. The justification for discarding rather
than including them was based on the fact that the unit was measuring the exact-same
area of sky, that they occurred only rarely, and that such outliers did not repeat during a
sequential run of measurements. The manufacturer has not had similar reports of such
outliers previously, and the issue remains an unexplained but probably temperature-
related matter for future study.
Once the unit had reached (presumably thermal) stability, the precision was normally very
good indeed – typically about 0.03 mag arcsec-2, or 3%, or better. This is to be set against
the predicted precision given by the manufacturer as 10%. The accuracy, of course, is
entirely unknown owing to the lack of traceable calibration available to the end user, but
later contact with the manufacturer indicated a single point calibration at ca. 8 mpsas is
conducted on each device, which provides some reassurance on accuracy, especially in
tandem with the user-determinable ‘dark current’ calibration point at the other extreme of
the range, which in this case showed a perfect agreement with dark current results
measured by a few other studies.

There were no statistically-significant differences in output measurements when the

instrument was pointed at the zenith in one direction, and then at 90 degrees to this first
direction. Thus, the detector has no detectable directional bias. However, the presence of
the Milky Way is known to contribute about 0.1 mpsas to the sky brightness, so its
absence can also reduce sky brightness to the same degree. This is probably evident in
the data obtained. No correction has been made, because the Milky Way is a part of the
night sky, and covers different parts of the sky as the year progresses.

No statistically-significant effect of body heat was found on precision, even when body
heat was escaping directly from below and into the path of the SQM measuring a 45
degree point.

Checks were carried out to determine whether lights near the horizon influenced the
outcome of the zenith measurements. There was an indication (Cors Tygai, south of
Llangefni) that, when tilted at 45 degrees, the SQM-L output was influenced by strong,
direct light sources along the horizon. This informed the project to take screened
measurements from the few locations where direct lights were significant, using bushes,
walls, etc. but of a low height such that the sky was otherwise entirely clear of
obstructions. Care must be taken to extinguish hand and especially head-worn torches,
and avoid very close streetlight and car lights. Car lights can cause localised sky-reflected
light pollution even when at some distance, and care has to be taken to avoid these.

In the case of Cors Tygai, the first assessment was discarded owing to the large, 0.98
mpsas difference between it and the second assessment, showing the problem to have
been direct light, exacerbated by the sensor’s parabolic reflector, affecting the outcome of
the first. Despite this, there was an unexpectedly large difference in the easterly value,
which is unexplained, other than to suspect instrument drift, for the sky was unremarkable.

A small number of further sites were resampled to assess the degree of impact on sky
condition reporting due to a combination of instrument variability, sky condition changes,
humidity, etc. Resampling was reinforced by measurements taken at each clear night at
Parys Mountain, which was much easier to accomplish due to there being no need to
travel to site. 21.25 mpsas was taken to be the most ‘typical’ darkness level at Parys
Mountain, but could increase in darkness by up to 0.15 mpsas. The expected
performance limits of the instrument are ~ +/-10%.
Discussion.
How do Anglesey skies fare when measured by a machine against the IDA standard, and
what do they look like to a real person standing under them?

As expected, zenith darkness was always better than the 45 degree points, except for the
case of Llanlleiana, where the northern 45 degree reading matched that of the zenith. This
shows both the much reduced light pollution due to the Irish Sea (although ships and gas
flaring in Liverpool Bay are transient issues), and the effect of fairly evenly-distributed,
small settlements across Anglesey.

Aberffraw dunes. The typical story for Anglesey skies is that of relatively good quality, but marred by
extensive light pollution ‘domes’ from centres of population and scattered, smaller settlements.

Indeed, it is a peculiarity of Anglesey that, especially from several NW, N and far SE
coastal areas, the sky can be quite badly light polluted in one hemisphere, whilst directions
out over the sea are almost free of the problem. This highlights the crucial importance of
evaluating the sky in many directions, not just ‘taking the best bits’ and quoting this as the
sky quality. However, atmospheric variability does make this complex and unpredictable.

IDA’s ‘Gold’ standard requires an SQM reading – it is not stated whether it must be the
lensed or unlensed version, but is usually taken to mean the former – of 21.75 mpsas or
better. This is considered by the present author to properly define a sky with extremely
low light pollution levels, but not the lowest possible in the absolute absence of light
pollution.
None of the sites measured on Anglesey met the ‘Gold’ degree of darkness at the zenith
(and not an aggregate of all-sky measurements.) Even when the sky transparency was at
its very clearest, the best Anglesey zenith measurement (Newborough forest) was 0.3
mpsas (32%) brighter than that required for an IDA ‘Gold’ award. Highly localised light
pollution control will not address this; it needs regional lighting control.

The next, ‘Silver’ standard, requires an SQM reading of 21.00-21.74 mpsas. When
considering zenith measurements only – which is the current IDA ‘standard method’,
almost all sites on Anglesey achieve this level of darkness, although often only just so.
Even so, the presence of light pollution is obvious even under the most casual look around
the sky. That is why for this study, an integrated mean of zenith plus cardinal point, 45
degree measurements was considered much more appropriate and indicative of the ‘real’
sky quality. It is artificial in the extreme to imply we only ever look at the sky directly
overhead, as normal human experience takes in the whole sky, and indeed the sky in
relation to the terrestrial or aquatic horizon.

When the integrated, ‘all sky’ mean (allowing the zenith to influence the results) is
considered, then the Anglesey night sky quality reduces, in most cases, to a ‘Bronze’
(20.0-20.99 mpsas) standard. This is fully in accord with the experience of those who
observe the night sky regularly from various parts of the island. However, extending the
IDA system to non-zenith values is merely indicative; it is not approved by IDA.

The ‘Bronze’ standard, however, merely indicates a sky darker than urban settings, and is
in fact a moderately light polluted sky under typical, clear westerly flows, but which might
rarely improve somewhat in darkness under flows directly from the Arctic, or where other
patterns lead to unusually reduced airborne particulates.

Data was obtained from award submissions to the International Dark Sky Association, from
Sark, Galloway Forest, Exmoor, Northumberland and Kerry. Snowdonia National Park
provided SQM data under a Freedom of Information Act 2000 request. This data allowed
a very simplistic comparison of darkness between areas. It was not within the remit of this
present study to critically analyse others’ methods and output in depth; in most cases,
detailed methodology and assessment of the SQM, akin to this study, is not presented.
However, certain problems do require discussion.

Anglesey’s zenith range is 20.70 magnitudes arcsec-2 (brightest) to 21.45 mpsas (darkest).
Newborough Forest car park had the darkest zenith value (21.45 mpsas.) The close
outskirts of Llangefni had the brightest, very closely followed by the outskirts of Holyhead.

Exmoor has a zenith range (mpsas) of 21.08 – 21.80, Galloway 21.95 – 22.72,
Northumberland 20.6 – 22.0, Sark 21.30 – 21.53, Kerry 21.4 – 21.85, Snowdonia 21.32 –
21.67 (lit areas excluded), and Brecon Beacons 21.29 – 21.56 (urban areas excluded).
 Anglesey thus has a brightest sky similar to
Northumberland’s brightest, and a darkest similar to
Sark’s and the Brecon Beacons’ darkest. However, these
darker sites on Anglesey are extremely limited in
geographical extent.

It is critical to note that our position relative to other places, especially those considered
‘remote’ such as Sark, does not mean Anglesey has skies that are unpolluted and that
cannot be much improved whilst preserving and improving amenity.

Sark and Northumberland, like Anglesey, suffer from moderate internal and imported light
pollution from various and sometimes considerable distances amongst the surrounding
landscape. Visiting the darkest sites of, say, Newborough forest and Carmel Head with the
expectation of experiencing a majestically dark sky will inevitably result in disappointment;
the direction of Holyhead, Llangefni and the Menai hub are very easy to deduce from here.

In terms of methodology across those areas that have assessed their night sky quality,
there are variations in approach, and not all are as informative as others.
As for the majority of submissions, Snowdonia’s survey, for example, engaged a mix of
members of the public (often seen by authorities as a very positive public engagement
outcome) and professional consultants. Whilst the SQM is very easy to operate, it does
have characteristics which might lead to significant inter-operator variability in outcome.
None of the submissions seen acknowledged this, although certain problems with
repeatability of results in the case of Exmoor did hint at this problem being realised,
although weather variability was also a factor there. In quite a few cases, SQM data is
simply accepted at face value and averaged to oblivion, with little regard for whether the
input and/or output values might require attention. Some studies suggested initial,
unstable values from the SQM had not been discarded as the manufacturer - and
operating experience - shows is certainly necessary.
 Eglwys Cwyfan, Aberffraw (looking SE), receives plenty of official attention as part of our cultural heritage.
Another component of our shared heritage - the night sky - has been largely ignored, such that our children
th
can no longer experience the night sky as it was when this church was first built, ca. 12 century. It need not
be that way. Image © Syed Zaidi, www.photographyrush.co.uk reproduced by permission.

It was surprising to discover that several studies used an unacceptably low number of
replicates (typically three), and in some cases, only one measurement upon which to
assess a given site. This may have been a result of the ‘point and it gives me a result’
attitude that can come to hold sway when deploying members of the public and a digital
readout to conduct such work. As all studies have been guided by IDA information, none
appeared to have used a combination of zenith and lower-elevation measurements.
This author is clear that the IDA regime of requiring only zenith measurements to assess
night sky quality is inadequate, because it ignores the effect of light pollution at lower
elevation angles, and suggests real humans observing the sky only ever look directly
upward – an unrealistic proposition. No other reports have commented upon this fact, nor
indeed appear to have recognised it as a weakness. The situation is loosely akin to saying
there is no train coming down the track because one is looking in the opposite direction!
IDA have, as part of this project's contact with them (March 2015), accepted that the
zenith-only assessment is a weakness, but have not indicated any intention to change the
guidelines as yet.
The strength in the use of zenith-only measurements is that it is less prone to changes in
how the local atmosphere affects light pollution levels, and so tends to vary less than
measurements taken at greater zenith distances as a result.
Having performed a careful first assessment across the island, a number of sites were
reassessed under similar sky conditions as determined by the zenith value at Parys
Mountain, used as a standard zenith assessment site. It became very clear that, even
where repeat assessments yielded very similar outcomes (+/- 0.05 mpsas) at the zenith,
the 45-degree directional values could change by up to a maximum of ~ 0.58 mpsas, or
71%. This did not always apply at other sites, making the situation very complex.

Does this large variability in sky quality at non-zenith elevations make the study worthless?

No. IDA requires only zenith values for the award of dark sky status, if deserved.
Normally, this simply involves going out on any clear night of unspecified transparency
relative to any other night, and submitting that value. IDA requires “between three and
five” readings only. This study took seven.

The author considers this study has taken great care to draw on lengthy experience of
observing the night sky, and to ensure various possible factors affecting the results have
been fully considered within the limits of available resources. No other study seen as part
of the research into this work has taken this approach, and it must be considered a ‘first’ in
that regard.

By taking this approach, the study confirms that the zenith data is less variable between
evenings, but that it also fails to identify surrounding light pollution – which is always very
obvious visually - and the direction from which it comes.

However, additional data from the greater zenith angles brings more complexity. It can
sometimes, but not always, be extremely difficult to reach agreement between repeat
assessments. This simply reflects variability in dust, pollution and humidity between air
masses, although to the observer, such variability is usually undetectable by visual
appearance. There may be room to consider whether instrument drift is a factor.

This means that observers in the past, as in the present study, have gone out on a very
clear night that, even with long experience, seems to be perfectly acceptable for
conducting sky quality assessments. This may not lead to significant problems with zenith
assessments, but can lead to difficulties at greater zenith angles.

Taking a standard zenith assessment each night, before a wider session of assessment,
helps to reduce the incidence of venturing out when the sky is in fact of a somewhat
decreased transparency. It may provide a potential correction value for subsequent
measurements, although in the case of non-zenith values, this may not be reliable. Such
correction may also be suspect where the standard site has a different overall lit
environment to that currently being assessed.

The present author concludes that the zenith values ought to be cited as the most reliable
‘spot check’ assessment of sky quality, but that, despite potentially higher variability, the
non-zenith values provide valuable information on the extent and, more particularly,
direction of more significant light pollution sources at any given site.
The saving grace is found by limiting ourselves not to fractions of mpsas, but to the IDA
standard limits for sky quality. These are perfectly sensible and would seem to
acknowledge the difficulty posed by inter-evening variability at the zenith.

In that regard, we can reliably report that Anglesey skies are overwhelmingly of a ‘Silver’
standard as defined by IDA – being 21mpsas or more at the zenith. Apart from highly-
localised, non-developed areas to the SW, NW and N of the island, and a further few sites
around Cors Erddreiniog, surrounding light pollution reduces the ‘all sky’ outcome to
‘Bronze’ (if that were a valid measure.)

The lights of Llangefni, 4.5km distant, have a large impact at Eglwys Llanddyfnan. The precise value of this
pollution varies greatly between evenings. However, the general extent and need for reduction in light
pollution is always self-evident by even a simple visual assessment.

‘Spot check’ assessments will always be prone to other evenings yielding a

somewhat different result. For that reason, it is worth considering the
purchase and deployment of datalogging SQM-Ls, at least for the darker and
brighter sites, in order to gather a true average and an indication of quality
variability over the year. Fully assessing night sky quality to the highest
standard is not a simple ‘point and shoot’ exercise.

It is relevant to note at this point that, if one is to gather a second line of evidence for night
sky quality that relies on photography, caution is also required. For example, the
photograph very kindly provided by Syed Zaidi, above, was not taken for the purposes of
faithful and scientific representation of the naked-eye view of the night sky. It was an
‘artistic’ image. As such, it shows much more sky detail than is visible to the eye, whilst
also exaggerating the degree to which light pollution is evident. Clearly, the detail is there
to be detected, but the human eye is not a comparable detector to an electronic sensor.
Care, therefore, has to be taken to ensure photographic evidence has a reasonable fidelity
with the naked eye experience of the sky. Several submissions to IDA showed that this
issue was often not recognised, although in such cases as Kerry, overexposure merely
underlined the almost complete lack of light pollution at many sites.

What’s in a number?

In considering the numerical output of an electronic sensor, one can become distracted by
a need to obtain the best precision and explain certain small differences in the values
obtained. Those very few studies who have conducted repeat measurements at some
sites have found it frustrating, some even worrying, that they have been unable to
precisely replicate a value previously obtained.

Whilst this is understandable, it is not necessarily an issue for concern.

The three, ‘Gold’, ‘Silver’ and ‘Bronze’ categories used by IDA are, mindful of these l
variations, very sensible. In encompassing a range of 1 mpsas each (‘Gold’ is open-
ended) they, in effect, acknowledge small and largely inconsequential sky condition
variations do occur, and that it is not very helpful to endlessly worry about whether the
exact and ‘correct’ value is, say, 21.1, or 21.3 mpsas. Remember that, in the case of the
alternative, naked-eye assessment systems, observers assessing the dimmest stars
visible are typically in disagreement by considerable fractions of a magnitude or more.
The SQM disagrees with itself by a factor of about 1.1. Between nights, even allowing for
real darkness changes, it varies by only a factor of about 1.3. Whichever way we look at it,
the SQM is preferable to the alternative. The IDA categories do appear to be sensibly
chosen, and are practical as a means of summarising site conditions, without endlessly
worrying about occasional variations from night to night.

At least part of the aim of this study is to inform season-extending tourism activity. It is
doubtful whether it is fair to say to potential visitors: “Look! We measured Anglesey’s skies
only during the most super-dark conditions that occur only on a handful of nights a year,
and the numbers prove they are superb!” The tourist then spends his/her money on a visit,
only to find ‘typical’ conditions have returned for the vast majority of clear nights, and will
be moderately more light polluted that this.

It is reasonable, however, to say we have measured under typical and most commonly-
occurring conditions, acknowledge that they can vary slightly, and have a numerical
evidence base on which to make our claims. That claim should currently be that, away
from the immediate surroundings of settlements, skies are of an IDA (zenith only) ‘Silver’
standard in most places. They do deteriorate to ‘Bronze’ if one is to take all-sky values
into account, but that is not a valid IDA standard at present; it is merely an informal, but
probably not entirely invalid, extension of the concept.
Problem Sites
This project involved a substantial degree of travel across the island, allowing a picture to
be built up of current problem sources of light pollution, other than the obvious centres of
population.
Comment on any given location is on the clear understanding that private landowners,
unless where they have been proven to cause a statutory nuisance or threaten transport
safety, are acting entirely lawfully.
Given this is a publicly-funded project, it was very disappointing to confirm that almost all
Local Authority buildings, most notably, but not exceptionally schools, have over-intensive,
poorly-controlled lighting units across the island. Some newer buildings (Llangefni Police
station, Ysgol y Bont) have much better installations. Given the current angst over funding
cuts, it is surprising that energy is being used so profligately and without question.
Wylfa power station has a large number of poorly designed and badly controlled lighting
that dramatically reduce the sky quality across the northern third of the island. It amounts
to a brightly lit settlement in itself. The impact on the RSPB reserve at Cemlyn and the
otherwise remote-feeling Carmel Head area is unacceptable, given the simplicity and lack
of expense with which down-shining lights could be installed. The author decided to add a
number of sample sites to the northern part of the island, to provide evidence for any
deterioration – or indeed improvement – as a result of decommissioning of Wylfa ‘A’, and
the likely building of Wylfa ‘B’. Horizon has provided a written undertaking to carefully
control lighting in all activities. The authorities ought to ensure this is adhered to over the
coming years. Horizon’s EIA section dealing with light pollution was, at the time of drafting,
inadequate, and considers only direct, and not indirect light pollution.
A number of farms, but certainly not all, use high-intensity yard lights, which have a high
impact because the farms are located in otherwise darker areas of the island. Advice and
possibly provision of subsidised ‘sky-friendly’ lighting to farmers would seem to be a good
idea, more especially as the use of poor lights is usually the result of using what is readily
on offer, and what is cheap to buy. The cost of running such lights does not seem to be a
consideration or, indeed, a concern. The lighting of farms and rural properties is
increasing.
At the time of writing, a large lorry park is nearing completion at Holyhead Business Park.
It incorporates high-intensity lighting atop tall masts, and is likely to lead to a further
deterioration in light pollution from the Holyhead Area. The Local Authority asked for
lighting details to be submitted, but there are, at the time of writing, concerns as to which
units have in fact been installed, and the needless use of ‘tilt’ of the lamps. This has been
raised by the author with the Planning department. The impact from the imminent ‘Land
and Lakes’ development also needs close attention, given the proximity of this and the
lorry park to the spectacular yet often overlooked AONB near the Stanley Embankment.
A number of garages, bus/lorry depots around the island, and notably one at Llanfaethlu,
are sources of serious light pollution in otherwise dark areas. Some sites clearly have a
potential to affect road safety, but action by the Local Authority seems to require a specific
complaint and, even then, very rarely occurs.
CADW produces considerable light pollution, with its strong upward illumination of
Beaumaris castle. CADW ought to take as much care of the night sky as it does of the
historical built environment, and the author encourages it to do so.
Whilst Holyhead Port lies within a larger conurbation with high light pollution, the port itself
is a major contributor to that pollution. Stena have, for a long time, made much of their
care for the environment, and so there is some considerable scope for them, and other
operators, to demonstrate this through improved lighting control.
The Mona light business park has at least one source of particularly poorly-directed, high-
intensity lighting that has an impact over a wide area around Mona. The Gwyndy quarry
also has fairly poor control of its all-night lighting, badly affecting the central and largely
unlit region of the island.
Whilst the red colour of navigation safety lights atop the large wind turbines at Ysgellog are
of a colour and, after much local complaint, lower intensity such that their impact is
relatively low, there is concern at the potential proliferation of such lights as smaller
turbines are increasingly replaced with fewer, larger ones. The seascape of the north
coast has been under threat from similar navigation lights due to the ‘Rhiannon’ offshore
wind farm, which has been abandoned, at least for now.
The lights of various units within the Gaerwen Industrial Estate cause serious deterioration
in sky quality out to significant distances from source.
It was, however, interesting and gratifying to see very significant reductions in light
pollution from the retrofitting with ‘full cut-off’ (down-shining) highway lights along the A55,
notably near Abergwyngregyn, and again at Pont Britannia across the Menai Strait. The
encouragement of continued and accelerated retrofitting with low light pollution units
should be an obvious and immediate starting point in reducing light pollution further.
Estimation of Bortle classification.

As already noted, there are some significant problems with the Bortle scale, and using it
objectively is not as straightforward as many believe.

Nonetheless, it is a useful approach to assessing overall night sky quality, and is not
restricted, as in the present case of IDA-approved zenith-only assessments, to just one
area of the sky. An attempt to apply it to Anglesey was therefore useful.

Bortle developed his scale according to certain phenomena that were or were not visible.
The following paragraph takes this same approach, using somewhat different phenomena,
when considering Anglesey skies.

From all the more rural locations, away from settlements on Anglesey, the
Milky Way is easily visible in some detail overhead, but that much of that
detail is lost within, typically, 20 – 30 degrees of the horizon. The zodiacal
light can be seen fairly weakly on the best nights, but not to very high
elevations – 30 to 40 degrees is about the maximum ever seen by the
author over many years of observing. Light pollution domes from
settlements are very obvious in several directions from all sites, and
extend to about 30 degrees above the horizon, but increasing in extent
with proximity to larger settlement centres. The limiting magnitude from
sites away from Holyhead, Llangefni and the Menai Hub, is about 6.
Clouds of any type and altitude (including cirrus) are obviously lit from
nearby sources, and from sources out to great distances (Liverpool,
Dublin, Belfast, Douglas). Clouds remain strongly lit overhead. The
auroral glow is often evident along the northern horizon from darker parts
of northern Anglesey, and is spectacular under stronger geomagnetic
disturbances, but only because of the presence of the sea. Airglow
emission is weakly visible on rare occasions, but only from the darkest
sites.

There are elements of the Bortle scale that apply to conditions on Anglesey, away from
settlements, that give a possible span of Bortle classification from 4 to 6, with Bortle 5
(termed ‘suburban sky’ by Bortle himself) having the most features that match our skies. A
car’s number plate could be read easily at a distance of about 1 metre by light pollution
alone from all sites on Anglesey. The presence of the sea, and thus absence of almost all
development, contributes significantly to the overall outcome for several sites at or near
the coast.
 Based on a modified application of the Bortle scale,
and drawing on years of experience of observing
under Anglesey skies, the island, away from centres
of population, generally has Bortle class 5 skies. The
sky quality reduces rapidly with reducing distance
from settlements.

On the immediate outskirts of the larger towns, the Bortle classification worsens to a 7, but
this is still directional to a degree, mostly because of the sea having no light pollution.

Because this project sadly did not extend to travelling to remote parts of the northern
hemisphere to develop the darker and brighter sides of the revised scale further, we have
to be content with a broad alignment as given above.

Groundtruthing Satellite Data.

Starting in the mid-1990s, the US Department of Defense started releasing night time
images of the Earth’s surface, revealing to a mass audience, for the first time, the extent of
light pollution across the world. Resolution was fairly low, and extensive debate broke out
about how representative of actual, ground-experienced light pollution the images were.
The classified nature of the satellites hindered full assessment of detectors, etc.

Many years later, the images being released by other satellite operators are of much better
resolution and quality. Light pollution models based on satellite data and published online
at http://www.avex-asso.org/dossiers/wordpress/?page_id=127XXXX, suggest Anglesey
has moderate light pollution levels centred on larger centres of population, with darkness
across the island interrupted by fairly evenly-scattered, small settlements. It suggests that
light pollution affects the whole island to a moderate level, with no areas entirely free of the
problem.

It is clear from the ground, SQM-L based measurements taken by this study, together with
the human experience of the sky, that the satellite-based modelling of light pollution across
Anglesey, even from the earliest release data, is very reliable, but not perfect.

Notably, the Avex model fails, completely, to identify the fairly extensive light pollution from
the Wylfa power station site and the overall impact of R.A.F. Valley’s apron lights. This
was not missed by the much earlier data. Whilst it cannot (yet) provide an absolute value
for light pollution, it does correctly predict, noted errors notwithstanding, the general extent
and impact of the problem with a remarkable degree of fidelity.
Recent Developments.
For the International Year of Light 2015, a review9 of satellite light pollution data was
released on March 25, 2015 that claimed a 28% reduction in light pollution within Wales
since 1992. The methodology is difficult to find, and there would appear to be scope for
questioning how well satellite data from 1992 can be compared with that from later
satellites. However, the mass change to low light pollution LED lighting for a large part of
the highway network is likely to have brought some reduction in impact.

Anglesey, however, remains very undeveloped economically when compared to much of

the UK. This leaves significant room, unless managed with coherent and enforced lighting
plans, for substantial increases in light pollution in the future.

Future Developments.

Satellite-derived light pollution information is likely to improve in terms of accuracy and

frequency of release to the general public over the coming few years.

Handheld ‘tablet’ computers and ‘smart’ phones are now ubiquitous in the western world,
and already, several global light pollution monitoring programmes using these devices are
being used to assess sky quality. The ‘Globe at Night’ campaign, and associated ‘Loss of
the Night’ application uses paper or computer-guided star identification to select areas of
the sky which the observer then uses to assess sky quality. The systems have automated
corrections for weather, age and any eye glasses worn. Results are uploaded to a central,
open-access web site.

Another application called ‘Dark Sky Meter’, measures the sky in much the same way as
the SQM used in this study, via the device’s integral camera. To ensure proper inter-
device calibration, only Apple iPhones are able to make use of this application at present.

A very limited indicative study of the DSM app was undertaken under cloudy but moonlit
conditions, and also under controlled indoor, very low illumination conditions. These tests
showed that the iPhone consistently overstated the brightness of the sky relative to the
Unihedron SQM-L by between 1 and 2 mpsas. If a SQM-L is to hand, then the iPhone app
allows a correction value to be input to bring it into good agreement. But the general
public will not have this capability, and nor is it in the spirit of attempting to replace an
expensive, specialist piece of equipment with a very inexpensive app.

However, with further development, it is probable that these applications will, in time,
remove the need for specialist equipment like the Unihedron SQM, whilst also greatly
increasing the coverage of sky quality assessments, thanks to mass public participation.

As part of an informal extension of this project a press release, inviting the public to
measure their local skies with an iPhone, was published in the Daily Post and associated
newspapers on 15/4/2015.
Going for Gold: putting the results to positive use.
Anglesey’s darkest zenith-only value (Newborough forest car park) is some 30% brighter
than that required for the IDA ‘Gold’ standard. With the island’s scattered settlements,
coupled to Holyhead, Llangefni and the impacts of Bangor, Caernarfon and the A55
corridor, achieving ‘Gold’ will not be an easy or short-term task.

In the short term, it may be useful to submit an application to IDA on the now
demonstrable basis that, for the most part, Anglesey enjoys ‘Silver’ standard skies (zenith
only.) This should draw attention to the issue, and allow positive marketing on the basis of
darker, if not entirely dark skies. However, this is not an ‘easy option’; it requires a lot of
work.

The present author does not favour the creation of ‘ghettoised’, small dark sky parks. This
is because it tends to create the belief that sky quality can be preserved simply by drawing
boundaries around relatively small areas of land. Those darkest areas of Anglesey are
already largely devoid of local lighting; it is light pollution from somewhat further afield –
and, often, much further afield – that must be tackled in order to properly reduce light
pollution.

However, it is clear that, as a starting point, a small number of sites on Anglesey are
worthy of consideration for urgent protection from potential development.

Because Newborough forest and Llanddwyn beach are extremely

popular tranquility venues for tourist visits at all times of the year,
and due to the very good access for both able and disabled
persons (now free during the evenings) this site deserves particular
consideration for immediate nomination, at the very least, as a
‘Dark Sky Discovery Site’. With the likely extension of electricity
supply within the forest area, now is a critical time to contemplate
very seriously the potential for any lighting introduced to seriously
damage the dark sky quality of this important site. Lights used in
the past at the gated entrance to Newborough Forest have been
excessively bright, of a very low design and installation quality,
creating significant light pollution in an otherwise dark area.

The AONB sites of Carmel Head and Dulas Bay areas should also receive immediate
attention in terms of stemming increases in light pollution, and moving forward to reducing
the pollution. The SSSI sites around Cors Erddreiniog and Bodafon also yield dark skies,
and should be protected from further increases in light pollution. However, light pollution
at all sites comes in from several, often distant sites, and tackling this requires a wider
response.
To move closer to ‘Gold’ status or, indeed, simply to reduce light pollution impact,
Anglesey will need to implement the following in a dedicated and integrated manner:

 Extend the replacement of sodium street lights with low light pollution LED units,
with an emphasis on Holyhead and Llangefni. The Highways department has
already made modest inroads and realised significant reductions in impact in some
areas. Reductions in lighting units, and dimming of others after 11pm (as in the Isle
of Man), also needs attention.

 Work with Gwynedd Council to refit existing lights in the Bangor – Caernarfon areas
with low light pollution LED units. Gwynedd CC was one of the first to take the
issue of light pollution seriously, but the momentum has not been kept up as it
might.

 Ensure the planning system is consistently robust in requiring lighting to be

minimised and, where shown necessary, in accordance with low light pollution
standards. The stated rationale behind requiring low light pollution units should be
more clearly environmentally-based, not simply ‘amenity’, which is nebulous.

 Commence work, probably alongside nationwide agencies, to examine how best to

encourage private landowners, farmers and business owners to fit low light pollution
lights.

 Examine existing lighting control plans submitted by applications for dark sky status
in other areas, and how they might be applied/modified for use on Anglesey.

 Ensure that a mentality of project/funding-driven ‘ghettoising’ of dark skies to very

small ‘parks’ or other areas does not come to rule. This will simply create a short-
lived ‘feel good factor’, whilst lighting remains entirely unaddressed outside these
areas.

 The Local Authority ought to lead by example by ensuring its own buildings are not
covered in poorly-installed and over-bright ‘security’ lights, as is the present
condition. The government has accepted for several years that lighting does not, of
itself, reduce crime; the Police locally, as elsewhere, should stop presenting an
alternative and untrue claim about lighting, which is often used as a ‘be seen to do
something’ solution. Police statistics in Bristol have shown a 20% reduction in
crime following the removal of lights in certain areas of the city in 2010/11.

 Present the public with well-prepared education about the growing environmental
and health impact of light pollution, the loss of the night sky, and the potential
economic benefits. For an island that markets itself heavily on the basis of being
‘unspoilt’, this should be a natural and easy extension. Drawing on the experiences
of others, possibly by some form of short secondment/sabbatical arrangement.

 Caution is always needed to avoid a negative response from the public if they
wrongly come to believe the work involves mass switch-offs of street lights, or
 increased interference with their rights as private owners. However, there is an
opportunity to widen the message that lighting, of itself, does not reduce crime.

The Study Sites

A total of 60 sites, detailed in Sheet ‘A’, were selected on the basis of, primarily, being
away from centres of population and direct sources of light (e.g. nearby streetlights), and
sampling as much of the island as possible within the resources, notably of time, available.

Initially, sample points were overwhelmingly located on statutory public footpaths or other
generally-accessible sites. However, it was found to be very time consuming and
impractical to walk a long way with equipment to many of them. The sites chosen then
became as close to the original sites as reasonably possible, but on public roads and
tracks where the equipment could be deployed from a car with minimal delay. None of
these roads were lit, no lit cars were present during measurements (the time of night this
study took measurements made this very easy) and all points were well away from any
nearby external lights. Any direct lights visible were shielded behind convenient hedges or
a notepad. The difference between the darkness of, say, the centre of Aberffraw dunes
and a layby on the road cutting through the same feature is insignificant or none at all –
provided care is taken not to measure when lit cars are passing anywhere nearby.

Settlements and other areas with a high level of nearby street lighting, from which few
would choose to observe the night sky in any event, were avoided.

Associated Digital Files.

The native files were created within the Open Office suite. In the event that the Microsoft,
xls files do not format correctly, Open Office may be downloaded for free from:

https://www.openoffice.org/

Raw data of sky assessments: raw_data.ods (also in xls format)

Summary site outcomes, with ranked sites: summaries_and_ranks.ods (also in xls)

Distribution data of zenith and all-sky outcomes: distribution_data.ods (also in xls)

Distribution plots: zenith_distribution_graphic.jpg and allsky_distribution_graphic.jpg

Map of zenith-only assessments: zenith_map.jpg

Map of all-sky assessments: allsky_map.jpg

Coloured interpretation map, zenith only: zenith_col_interpmap.jpg

Coloured interpretation map, all-sky: allsky_col_interpmap.jpg

SQM sensitivity to moonrise test: moonrise_test.ods (also in xls)

References

1. Bortle, J. E. 2001. http://www.skyandtelescope.com/astronomy-resources/light-

pollution-and-astronomy-the-bortle-dark-sky-scale/ First published in Sky & Telescope
magazine (print) February 2001, then published online July 18, 2006. Accessed 2014
November 27

2. Crumey, A. 2014. "Human Contrast Threshold and Astronomical Visibility." Posted to

Arxiv May 2014, accessed 2014 November 27. Accessed 2015 January 13

3. Weaver, H.F., 1947. “The Visibility of Stars Without Optical Aid.” Publications of the
Astronomical Society of the Pacific, Vol. 59, No. 350, p.232. Accessed 2015 January 13.

4. Cinzano, P. 2007. "Report on Sky Quality Meter, version L preliminary draft, ISTIL
Internal Report n. -, v.0.9 2007" Istituto di Scienza e Tecnologia dell’Inquinamento
Luminoso, Via Roma 13, I-36106 Thiene, Italy. Accessed 2015 January 03

5. Kollath, Z. 2010. "Measuring and modelling light pollution at the Zselic starry sky park."
In Journal of Physics: Conference Series, volume 218, page 012001. IOP Publishing.
Accessed 2014 November 27.

6. Conci, A. Masters Thesis "SQM-PI: helping astronomers to measure light pollution.",

Academic Year 2012/13, Universita Degli Studi Di Trento, at

Accessed 2014 November 27.

7. Shariff et al. 2012. "The Application of Sky Quality Meter at Twilight for Islamic Prayer
Time". International Journal of Applied Physics and Mathematics, Vol. 2, No. 3, May 2012
143-145. Accessed 2014 January 03

8. Schnitt, et al, 2013. “Temperature stability of the Sky Quality Meter”. MDPI Sensors,
Volume 13, Issue 9, September 2013. Accessed 2015 February 18.

9. http://www.hillarys.co.uk/skyglow/ Accessed 2015 April 16.

About the Author.
John Rowlands took up an interest in astronomy at the age of 11, helped by the subject
matter of this study: relatively dark Anglesey skies. From 1989 until the present, he has
been a contributor of feature articles, news items, book and equipment reviews to
Astronomy Now, Popular Astronomy and Wiley Science Publishing.

John wrote the first UK, popular article about light pollution in 1989, which has since
developed into a mainstream topic of environmental concern. He has been an active
campaigner since then. He gained the first changes in UK lighting around 1990 (from
Gwynedd CC) to reduce light pollution. Such fittings are now standard installation choices
across the UK and much of the western world. John wrote a paper on developments with
reducing light pollution for the Astronomical Journal of the Atlantic in 2004.

John took part in a hugely successful night sky-related public outreach production on
Radio 4’s Material World during 2010, and has since appeared on NHK (Japan), Falkland
Islands radio, and other radio and TV productions.

The BBC work developed into a long-term collaboration with NERC’s MST radar facility at
Capel Dewi, Aberystwyth, resulting in poster presentations on the relationships between
PMSE and NLC at MST conferences in Germany and Brazil.

John is a co-author of a 2013 paper with Peter Dalin, IRF Kiruna (Sweden) in the Journal
of Geophysical Sciences: Atmospheres, examining an artificially-induced, Soyuz rocket
launch noctilucent cloud formation over Russia and Europe. John was the first to both
photograph and identify ablated dust from the large Chelyabinsk bolide entry in February
2013, and worked briefly with the Australian Antarctic Division to model – using NOAA
HySplit trajectories - and confirm the origin and transport within the stratosphere of bolide
dust around the northern hemisphere. This work predated, by several months, similar
work produced by NASA.

John was principal author of a feature article in Popular Astronomy with Melissa Saentz,
then at Caltech, bringing the first ever, visual-audio synaesthesia-based explanation,
backed-up by experimental evidence, for sounds long-claimed by some to be heard during
aurorae.

John holds a first at degree level in Astronomy and Planetary Science from the Open
University, and was elected a Fellow of the Royal Astronomical Society in 2012. For
several years, he has run a popular Facebook page aimed at outreach activities related to
noctilucent cloud science and observing.

John is the author of a guide to noctilucent cloud science and observing for the general
public, and the author of several other technical books. He successfully nominated Bob
Mizon, the Campaign for Dark Skies’ Coordinator for very many years, for recognition in
the Queen’s Birthday Honours List. Bob was awarded an MBE for “services to astronomy
and the environment” at Windsor Castle in 2010.
Acknowledgements.
The author wishes to thank Efan Milner, Anglesey County Council’s Countryside and
AONB Service for his support in obtaining funding to make this report possible.

Thanks also to John Ratcliffe of Natural Resources Wales, and Tim Bowie, Bodorgan
Estate Manager, and the Estate itself, for their valued support in furthering the cause of
tackling light pollution.

The author appreciates greatly the kind persmission granted by Syed Zaidi to reproduce
his image within this report.