Contributions of artificial lighting sources on light pollution in Hong Kong measured through a night sky brightness monitoring network Chun Shing Jason Pun1 4 [email protected] 1 0 2 Chu Wing So1 n [email protected] a J 4 Wai Yan Leung1 ] [email protected] M and I . h p Chung Fai Wong - o r t s a [ ABSTRACT 1 Light pollution is a form of environmental degradation in which excessive artificial outdoor v lighting, such as street lamps, neon signs, and illuminated signboards, affects the natural 0 environment and the ecosystem. Poorly designed outdoor lighting not only wastes energy, 0 money, and valuable Earth resources, but also robs us of our beautiful night sky. Effects of light 2 pollution on the night sky can be evaluated by the skyglow caused by these artificial lighting 1 . sources, through measurements of the night sky brightness (NSB). The Hong Kong Night Sky 1 Brightness Monitoring Network (NSN) was established to monitor in detail the conditions of 0 4 light pollution in Hong Kong. Monitoring stations were set up throughout the city covering a 1 wide range of urban and rural settings to continuously measure the variations of the NSB. Over : 4.6 million night sky measurements were collected from 18 distinct locations between May 2010 v i and March 2013. This huge dataset, over two thousand times larger than our previous survey X (Pun & So 2012), forms the backbone for studies of the temporal and geographical variations r of this environmental parameter and its correlation with various natural and artificial factors. a The concepts and methodology of the NSN were presented here, together with an analysis of the overall night sky conditions in Hong Kong. The average NSB in Hong Kong, excluding data affected by the Moon, was 16.8 mag arcsec−2, or 82 times brighter than the dark site standard established by the International Astronomical Union (IAU) (Smith 1979). The urban night sky was on average 15 times brighter than that in a rural location, firmly establishing the effects of artificial lighting sources on the night sky. Subject headings: light pollution, night sky brightness, skyglow, Moon radiation, urban lighting 1 1. Introduction ferred to as NSB hereafter). The NSB is a com- bination of the natural sky glow due to celestial Outdoorlightingisanindispensableelementof objects (Sun, Moon, planets, stars, Milky Way, modern civilized societies for safety, recreation, galaxies, etc.), and the sky glow due to direct and decorating purposes. However, poorly de- or reflected light from artificial lighting sources. signedoutdoorlightingsystemsandexcessiveillu- Apart from the effects of the Sun and the Moon, mination levels can lead to light pollution (Smith the chief contributor to the NSB for a highly 2009;Mar´ın&Jafari2007). Thescatteringofarti- populated metropolitan city is the artificial light- ficiallightbycloud,aerosol,andpollutantssuchas ing source. The level of the NSB had mostly suspended particulates in the atmosphere spread been studied at professional astronomical obser- the effects to distances beyond the position of the vatories using the traditional astronomical tech- lighting source and can brighten the entire night nique of photometry — the amount of light de- sky (Benn & Ellison 1998). Light pollution is a tected in star-free regions on the CCD images form of environmental degradation in which ex- was extracted to estimate the NSB at different cessive artificial outdoor lighting affect the natu- wavelengthbands(Pedani2009;Stalinetal.2008; ral environment and the ecosystem. It not only Patat 2008; Sanchez et al. 2007; Krisciunas et al. represents a waste of energy, money, and valuable 2007;Tayloretal.2004;Liuetal.2003). Thiskind Earth resources, but also indirectly contributes to ofobservationhelpedastronomerstorevealpoten- the global environmental problems. Last but not tial light pollution threats to the observatory and least, the skyglow due to these artificial lighting aidintheirsearchforanewpotentialdarkobserv- sources leads to the degradation of the quality of ing location. These observations would normally night sky and reduces the number of observable require skilled personnels using delicate experi- stars, robbing us the beautiful night sky on dark mental setup (telescope/lens/camera/mounting). nights when it would otherwise be visible. Thegeographicalandtemporalcoveragesofthose The energy wasted by artificial lighting sources studies were usually limited (limited to a single can be directly monitored by night-time images site, and only a relatively few observations per of the Earth. Satellite imagery taken from the night). US Air Force Defence Meteorological Satellite Large-scale surveys of the sky conditions had Program (DMSP) Operational Linescan System also been carried out by campaigns which re- (OLS)1 sampledlargeareasoflandmassatamod- cruited ordinary people to conduct visual studies erate spatial resolution (∼ 1 km), making studies of the night sky following simple procedures with of the extent and degree of night time radiation minimal technical expertise required, such as the over a metropolis or even globally possible (Kyba GLOBEatNight3 andtheGreatWorldWideStar et al. 2013; Small & Elvidge 2013; Cinzano & Count4 projects. Without the need for special- Elvidge 2004; Cinzano et al. 2001). The night- ized or expensive equipment, this kind of citizen- time photographs taken by astronauts onboard scienceprojectencourageda largenumberof peo- the International Space Station (ISS)2 provided ple from around the world to take part in studies evenhigherspatialresolution(at∼6mperpixel) of light pollution, yielding results with broad ge- for selected locations on Earth. While the light ographical coverage and high spatial resolution if intensityontheISSphotographofHongKongwas the number of observers were large enough. Fur- found to be positively correlated with population thermore,manyobservationswereconductednear densityingeneral(Liuetal.2011), thescarcityof places where the participants lived, such as city these photographs was insufficient for studies of centersorsuburbanregions,oratplacesthatthey nightly variations of light pollution in the city. could gain access to, such as country parks. Re- The extent of light pollution can also be mon- cent comparison of a subset of the GLOBE at itored by studying the night sky brightness (re- Night results with the night-time satellite images 1DepartmentofPhysics,TheUniversityofHongKong, 3http://www.globeatnight.org/ PokfulamRoad,HongKong 4http://www.windows2universe.org/citizen_science/ 1http://ngdc.noaa.gov/eog/ starcount/ 2http://eol.jsc.nasa.gov/ 2 illustratedthatthiskindofstudyhashugepoten- light pollution. However, the dataset collected tialforglobal scalestudiesof thenightsky (Kyba was limited by its geographic distribution (volun- et al. 2013). teers made measurements usually within or near Hong Kong is a populous metropolitan city urban population areas), temporal resolution (a (mid-2012 population 7,154,600 (Census and majority of volunteers made measurements usu- Statistics Department 2013)) with a high pop- ally once or twice every several nights), a short ulation density (6,620 per km2), known for its monitoring time (volunteers were swapped every spectacular night lights. The mountainous city few months to allow for more participation), and has a complex geographical landscape, leading to possible human-related errors (volunteers might a short supply of habitable land. A photograph make mistakes during measurements and/or data taken from the ISS in March 2003 (Figure 1) re- reporting). vealed the huge amount of upward shooting light WelaunchedthesucceedingprojectHongKong atnight. Asseeninthefigure,thereweresubstan- Night Sky Brightness Monitoring Network (re- tial variations in the level of night-time emission ferredtoasNSNhereafter)in2010tocomprehen- from the city, mostly due to differences in pop- sively study the properties of the NSB in Hong ulation and land utilization. It seems obvious Kong and its dependence on time, location, and that the degree of light pollution in Hong Kong variousatmosphericandmeteorologicalconditions should be strongly dependent on human activity, with the support of the Environment and Con- in particular, how and where people use external servation Fund of the Hong Kong SAR govern- lighting. ment. The range, depth, and accuracy of data The availability of low-cost light sensors orig- collection were optimized by setting up automatic inally targeted for the astronomical community NSB measurement stations in 18 distinct urban allowed for detailed and comprehensive studies and rural locations around Hong Kong. All these of the NSB. Between 2007 and 2009, we con- stations were designed so that on-site NSB mon- ducted a citizen-science survey of light pollution itoring for over a year was possible after secur- inHongKongbyinvitingstudents,astronomyen- ing long-term commitments from our collaborat- thusiasts, and campsite employers to measure the ing partners (refer to acknowledgment section for NSBusingonesuchdevice,theSkyQualityMeter the full list). The temporal resolution of data (SQM) (Pun & So 2012). Compared to projects collection was vastly improved and human errors such as GLOBE at Night, the use of a standard were eliminated through the use of the ethernet measuring device in this study reduced the uncer- version of the SQM. In Section 2, details of the tainties in night sky measurements due to varia- survey are discussed, including the setup of the tions in observers’ eyesight and experience. From automatic NSB measuring stations, site selection, the over 2,000 measurements taken at almost 200 data selection, data quality control, and calibra- locations by over 170 volunteers, it was concluded tion of the measurement devices. The resulting that light pollution in Hong Kong is severe, with database forms the backbone for a study of the largebrightnesscontrastbetweentheobservedur- overall light pollution conditions in Hong Kong. ban versus rural locations. Moreover, later night Results of the analyses can be found in Section 3, skies (at 23:30, local time (UTC+8) hereafter) focusing on the contributions of artificial lighting were generally darker than at an earlier times (at sourcesontheobservedNSBinanurbanenviron- 21:30), which could be attributed to some public ment. Discussion of the results from this study andcommerciallightsourcesbeingturnedofflate and possible potentials of this NSB database are at night. presented in Section 4. This survey not only provided the first glimpse of the light pollution situation in Hong Kong, but also spread the message of dark sky conservation and energy saving among students and the gen- eralpublicthroughparticipationsinthehands-on sky brightness measurements and first-hand ob- servations of the environmental consequences of 3 Fig. 1.— Image taken from the International Space Station (ISS) when it flew over Hong Kong in orbit at altitude of ∼ 300 km. The image was taken at 00:51 (UTC+8) on 11 March 2003 local time. Landmass boundaries and locations of NSN monitoring stations obtained from Google Map are overlaid on the im- age. Refer to Table 1 for station codes and Section 2.1.2 for details. (credit: Image Science and Analysis Laboratory, NASA-Johnson Space Center, The Gateway to Astronaut Photography of Earth) 2. Methodology mag arcsec−2. The SQM-LE is manufactured by the Canadian company Unihedron6 and each unit 2.1. Data collection was calibrated by the manufacturer before ship- ment. The light sensor of SQM-LE is the TAOS 2.1.1. Measurement instrumentation TSL237High-SensitivityLight-to-FrequencyCon- In the current study, the lensed version of the verter covered by a near-infrared blocking filter SQM, the Sky Quality Meter - Lens Ethernet Hoya CM-500. The combined filter-sensor sys- (SQM-LE), was used to collect the night sky ob- servationdata. BoththeSQMandSQM-LEmea- thecelestialsphere. 1”=1/3,600◦. Supposethemeasured sure the brightness of the night sky in units5 of NSBatsiteAis20.0magarcsec−2,thenthebrightnessof the sky is equivalent to a celestial object of 20 mag filling up a patch of sky of area 1 arcsec × arcsec. Suppose the 5Magnitude (mag)isalogarithmscaleinternationalunitto measuredNSBatsiteBis19.0magarcsec−2,thenthesky measure the brightness of astronomical objects. A differ- atsiteBis2.512timesbrighterthanthatatsiteA. enceof1magreferstoanobservedlightfluxratioof100.4 = 2.512. Arc second (arcsec or ”) is the unit of length on 6http://unihedron.com/projects/sqm-le/ 4 tem has a central wavelength of 540 nm and a SQM-LE,theconfigurationandappearanceofthe Full Width Half Maximum (FWHM) of 240 nm setup were shown in Figure 2. The enclosure for (Cinzano2013a),andisdesignedtohaveasimilar eachmodulealsoincludedapowersupplyadapter sensitivity to that of the human eye. Compared and a network device for data transmission (to be to the earlier version of the SQM, the SQM-LE discussedindetailinSection2.1.3). ExceptSQM- has a narrower field of view of ∼ 20◦ in FWHM. LE, almost all components of the observing mod- The detector sensitivity drops sharply away from ule employed readily available commercial prod- the zenith direction, with sensitivities at factors uctstoreducethecostforconstructionandmain- 15 and 100 respectively, lower than that at the tenance. The entire housing was supported on zenith at 20◦ and 40◦ off-axis (Cinzano 2013b). a tailor-made stainless steel frame mounted on a The SQM-LE was preferred over the original un- poleattachedusuallytotherailingsontherooftop lensedversionoftheSQMforaNSBstudyinHong of buildings. The entire housing and frame could Kong because the measurements would be less af- easily be dismounted so that the entire unit could fectedbyeffectsofstraylightingdirectlyreaching be collected from an observing site to our labora- the light sensor in this dense urban environment. tory for regular checking and maintenance. The The stated accuracy of the SQM-LE by the man- modulardesignalsoallowedforswappingofhard- ufactureris0.1magarcsec−2,orroughly±10%of ware between stations. the brightness of the sky in terms of emitted flux. The summer temperature of Hong Kong can While the convenience and the durability of reach a high of 35◦C and thus temperature con- SQM-LEfavoredtheirusageforalong-termnight trolisimportantforprotectionofthevariouselec- skymonitoringproject, afewproblemsofthisde- tronic devices, including the SQM-LE (maximum vice had been identified. First, dropped trans- operating temperature 85◦C). Even though the parency of the filter could occur due to frosting unit’s temperature compensation algorithm is ad- developedwhenoperatedinahumidenvironment equateundertheexpectedoutdooroperatingcon- foraprolongedperiod. Afilterwithanti-moisture ditions in Hong Kong (Schnitt et al. 2013), we in- coatinghadbeenprovidedbythemanufacturerfor stalled ventilation fans at the bottom of the hous- replacement(cfSection2.3). Second,theunithad ing and glued silver reflective linings to reduce not been calibrated to accurately measure a night the heat accumulation in the interior of the unit. sky darker than 23 mag arcsec−2. However, this We placed the linings inside the housing instead is a non-factor for our study as the night sky in of outside to ensure that they would not be de- HongKongwouldneverreachthatdarklevel. Fi- tached after long-term outdoor exposure, and it nally, in earlier versions of the SQM-LE firmware had proved to serve its purpose. In each mod- (before version 19), the precision of the meter ule, an electronic timer switch limited the power for readings brighter than ∼ 17.0 mag arcsec−2 supply to 15:50 − 09:00 each day for both energy suffered from digitization problems, causing the savingandforreducingheatgeneration. Thedaily units to round off measurements between 15.2 power reset also occasionally helped to overcome and 17.0 mag arcsec−2 to the nearest 0.15 mag systemfailuresduringthenight,andrecoveredop- arcsec−2, and measurements brighter (smaller in erationofthemoduleinthenextdaytoreducethe NSB) than 15.2 mag arcsec−2 to the nearest 0.18 number of required manned on-site service main- mag arcsec−2. As this problem had been fixed by tenance. asubsequentfirmwareupgrade, only∼ 8%ofthe Through regular maintenance of the observing total data collected at 13 different stations from modulescollectedfromthemonitoringstations, it 14 individual units in the first 20 months of the wasdiscoveredthatthetransparentcoverwindow survey could be affected by this error. of the polycarbonate housing suffered from obvi- The SQM-LE was not originally designed for ous aging and turned visually yellow after ∼ 12 permanent outdoor installations. To protect it months of outdoor operations. To prevent fur- from the wear and tear of the outdoor environ- ther degradation in data quality, a small portion ments,eachunitwasfittedintoawaterproofpoly- ofthepolycarbonatehousingrightontopofSQM- carbonate housing which has a transparent cover LE was replaced with a 25.4 mm in diameter and for light to reach the SQM-LE sensor. A unit of 3 mm thick high efficiency circular glass window 5 Fig. 2.— A unit of SQM-LE is shown in (a) (credit: Unihedron). Its semi-conductor light sensor is located under the filter in greenish blue. The view in the bottom shows the power inlet, the ethernet inlet, and the switch for calibration purpose. The dimension of a SQM-LE unit is about 91 × 66 × 28 mm. The internal hardware configuration and appearance of the iObs NSN monitoring station are shown in (b) and (c) respectively. The dimension of this model of housing is 278 × 278 × 100 mm. As discussed in the main text, a small piece of the polycarbonate surface where the SQM-LE points towards the zenith had been replaced by a piece of a thin glass 25.4 mm in diameter. For each complete module, hardware costed about USD770in2010, weights∼5kg, andrequireslessthan∼16WACpowersupplyduringnormaloperation. In the outdoor photograph, in addition to the night sky measurement module, two cloud sensors (narrow cylinders mounted on the right side of the rectangular housing) were also mounted on the left-side pole, while an automatic weather station installed by the Hong Kong Space Museum on the right-side pole was also shown. withanti-reflectioncoating(EdmundOpticsStock Apart from the demographic and geographic No. #46-0987). The glass was glued directly considerations described above, each observing above the SQM-LE sensor by drilling a small hole module should have a wide field-of-view (FOV) on the housing cover. Procedures for corrections to the sky not obstructed by any lighting source. of the data that suffered from the aging problem Thiscouldposeabigchallengeforoururbanmon- will be described in detail in Section 2.3. itoring site selection due to the high density of buildingsinthecityresidentialareas. Inalmostall 2.1.2. Monitoring stations cases the measurement module was placed on the rooftop of a building for the widest FOV. More- From our previous study of the light pollu- over,theambientenvironmentneareachmonitor- tion condition in Hong Kong (Pun & So 2012), ing station was checked so that the top floor of it was clear that the level of external lighting was the nearest building was at least 40◦ away from the dominating factor of the NSB level observed. zenith. As discussed in Section 2.1.1, the sensitiv- With a highly mixed land utilization within a ity of the SQM-LE at that angle was only a fac- small landmass, along with a complex landscape tor of 0.01 of that at the zenith. All monitoring and diverse forms of human activities in different stations satisfy this criteria, except in the WTS areas in Hong Kong, it was desirable to strategi- where the rooftops of a handful of buildings were callychoosemonitoringstationstoformanetwork visible at an zenith angle of ∼ 30◦. Other fac- coveringaswideageographicalcoverage, rangeof tors for consideration included the availability of population density, landscape type, and land use power, and more importantly, steady and strong as possible within the budget constraint. mobileInternetnetworkattheobservinglocations (cf Section 2.1.3). These could be demanding re- 7http://www.edmundoptics.com/ 6 quirements for some of the rural locations in our Kong Observatory (HKO). Situated on the survey. summit of a small hill at the middle of With a hilly and complex terrain, the popula- Kowloon, it offers a location with a rela- tion density of Hong Kong is highly uneven. Re- tively wide FOV in the middle of a densely gions of the highest population density lie on ei- populatedmixedresidentialandcommercial thersideoftheVictoriaHarborbetweentheHong area. A large variety of surface meteorologi- Kong Island and Kowloon. Additional popula- calmeasurementsareconducteddailybythe tioncentersspreadthroughouttherestofthecity many equipment operated by the HKO at (known as the New Territories region), observable this station8. as clusters of light in the ISS image in Figure 1. TC This urban station was located in a satellite On the other hand, large portions of the territory residential town adjacent to the Hong Kong are allocated as reserved park areas and remain InternationalAirport. TheNSNmodulewas sparselypopulated. Inanidealsituation,themon- installed at the rooftop of a two-storey gov- itoring network should cover as wide a geograph- ernment building in the center of the town. ical distribution and population density as possi- In addition to shopping malls and high-rise ble. However,withlimitationsofmanpower,fund- residential buildings nearby, the major ar- ing, and accessibility to suitable sites, we ended tificial light source is the Airport passenger up selecting 18 locations for construction of NSN terminal at about 3 km away. stations. Details of these stations were listed in Table 1. In Figure 1, the station locations were HKn This urban station was located near the overlaid on the night-time image taken from ISS astronomical observatory on the rooftop of in 2003. Station codes in Table 1 will be used an environmental education center complex to refer to specific stations hereafter. While this on the foothill of Tai Mo Shan, the high- projectwasconductedbetween2010and2013,al- est peak in Hong Kong. While it has the mostadecadeaftertheISSimageinFigure1was highest elevation (∼ 140 m above sea level) taken, it is expected that the overall pattern or amongallstationsinthenetwork,itwasstill distributionofpopulation, henceforththeamount affected by external lighting of even taller of external lighting used, did not change by much highrisesnearbyandfromtheadjacenttown foramajorityoftheregionsduringtheintervening center. The education center held regular period. school group and public visits which could Unlike many cities worldwide, the wide spread occasionally affect the data taken. of population centers in Hong Kong made the ex- HKU This urban station was installed near the act classification of each monitoring locations as teaching astronomical observatory on the urban, suburban, and rural to be ambiguous. On rooftop of the Department of Physics build- the other hand, the ISS image did seem to justify ing at the University of Hong Kong (HKU). theclassificationofonlytwocategories,urbanand HKU has an urban campus with high-rise rural. The classifications were listed in Table 1 residentialbuildingsinthesurrounding. Me- and shown in Figure 1. With this rough system, teorological equipment including two cloud a total of 10 monitoring stations were classified as sensors, an all sky camera, and an auto- urbanand6asrural. Intwoinstances,thestation matic weather station were installed near was neither classified as urban nor rural because the night sky monitoring module for addi- of their unique locations near the Hong Kong In- tionalmeasurementsoftheatmosphericcon- ternational Airport (GFS) and the River Trade ditions. Occasional educational events were Terminal (TMD). held at the observatory which could affect Briefdescriptionsofallthemonitoringstations the operation of the monitoring station. listed in Table 1, in their order of operation start dates, are provided below. iObs This rural station was installed on the rooftop of the building housing the robotic KP This urban station was located at the King’s Park Meteorological Station of the Hong 8http://www.hko.gov.hk/wxinfo/aws/kpinfo.htm 7 observatory of the Hong Kong Space Mu- observationsduringclearweathernightsand seum inside a rural country park. The ob- theseoperationscouldbrieflyaffectthedata servatoryispartofacomplexofbuildingsin collected. a public camping village. The main sources Cap Thisruralstationwaslocatedontherooftop of artificial lighting include the external fix- of the marine science research center oper- tures of the camp cottages and recreational atedbyHKUatoneofthesoutheasterntips facilities within the complex. Moreover, of the Hong Kong Island. The building lies scattered lights from a town centre about on the shore of a marine reserve area with 6 km away in southwest (Sai Kung) can be minimalamountofartificiallightingfixtures noticed from the observatory. Similar to intheambientenvironmenttoallowforeco- the HKU station, meteorological equipment logical studies of the region. The main light including two cloud sensors, an all sky cam- sourceisalighthouse(ataheight20mabove era, and an automatic weather station were the observing module) located at ∼ 200 m installednearthenightskymonitoringmod- away. ule (the latter two equipment were installed by the Hong Kong Space Museum). As part AP This rural station was located in the As- of a public educational facility, stargazing tropark, a designated stargazing park de- events were regularly held as part of the signed by the Hong Kong Space Museum. Museum and village programs. The AP station was located ∼ 3 km deeper into thesamecountry park comparedto the TST This urban station was installed on the iObs station described earlier. No exter- rooftop of the Hong Kong Space Museum. nal lighting operates within the park, but The museum is located in the waterfront of scattered light from a water sports center at the Victoria Harbor, which is the central 100mawaycouldaffectmeasurementstaken commercial area of the city clustered with a at the station. Since its opening in 2010, large number of hotels, shopping malls, and the park has been one of the most popular touristattractions. TheMuseumbuildingis stargazing venues in Hong Kong for astron- relatively low in height (∼10 m) compared omy enthusiasts. to office buildings and hotel towers nearby that can reach ∼100 m. In addition to the TMD This station was located in a temporary spilledlightfromtheinteriorsofthesebuild- observatory setup near the Hong Kong Sci- ings, other lighting sources include the large ence Museum’s depot in western New Ter- andbrightilluminatedbillboardsdecorating ritories. While the nearest residential town the building exterior, particularly near the was relatively distant at ∼ 4 km away, the rooftopofthebuildings. SimilartotheiObs station’s lighting environment was mainly station,stargazingandeducationalactivities affected by the River Trade Terminal at at the rooftop could occasionally affect the about 0.5 km away. This container terminal measurements. for river barges has 49 berths and a large number of lighting fixtures remained turned SH This rural station was located on the rooftop onthroughoutthenight. Thereforethissta- of the building housing the observatory of tion is classified as neither an urban nor a theHongKongAstronomicalSociety, which rural site. was operated by local amateur astronomy enthusiasts in a rural village setting, with MWo This rural station was installed on the onlyasmallnumberofhousesinthevicinity. rooftop of a campsite at about 1 km away While the Airport is located about 10 km from a ferry pier, which was also the loca- away to the North, this site is not expected tionofaruraltowncenter. Thisstationwas to be strongly affected by its strong lighting closed permanently in August 2011 due to due to the shielding effect from the Lantau the closure of the campsite for major reno- Peak(altitude934m)inbetween. Members vation work. of the Society might use the observatory for 8 TPo This urban station was installed on the ST This urban station was installed on the rooftop of a government office building lo- rooftop of a school located in a population catedwithinapopulationcenterintheNew center in southern New Territories. This Territories. Public facilities and residential school is also surrounded by taller residen- estates in form of high-rise apartments as tialbuildings,withtheclosestblockatabout closeas30mawaycouldaffectthemeasure- 80 m away. ments collected. GFS This station was installed within the con- TSW Thisurbanlocationwasestablishedwithin fines of the Hong Kong International Air- the Hong Kong Wetland Park which is a port. This busy airport operates 24 hours conservation,education,andeco-tourismfa- a day and most of the external lighting for cility. The park comprises of a visitor cen- air traffic remains turned on throughout the ter and 60 hectares of outdoor wetland pro- night. This station was installed on the tected areas. The station was installed in a rooftop of a facility building on the south- location inside the park that was not open western corner of the airport. The South to the pubic. As an ecologically important runway and the airport passenger terminal area, usage of outdoor lighting within the are located about 500 m in North and 3 km park is kept to a minimum. On the other innortheast. BoththeHKOandtheairport hand, a major population center is located operator(AirportAuthorityHongKong)in- about 500 m away from our detector and stalledsuitesofadvancedmeteorologicaland could contribute to the observed NSB. air quality monitoring systems at different locations of the airport for air traffic opera- WTS This urban station was installed on the tionsandforenvironmentalmonitoring. We rooftop of a school located in a densely also installed cloud sensors adjacent to the populated district with residential estates, SQM-LE for further studies. shopping malls, and transportation hubs in Kowloon. At only nine storeys high, our de- Thelightingenvironmentneareachmonitoring tector was practically surrounded by taller stationwererecordedatthebeginningofthestudy buildings in all directions, with the closest andmonitoredregularlythroughourmaintenance one at about 70 m away extending up to trips to ensure that the FOV of all the SQM-LE ∼ 30◦ in zenith angle. werenotblockedbyanyobstaclessuchastreesor artificialstructures. Moreover,weobservednoev- TpM This rural station was installed on the idence of drastic change of the lighting settings in rooftop of a Police Post on a sparsely pop- our monitoring stations throughout up to 3 years ulated (population ∼ 100) island on the of duration of survey. northeastern corner of Hong Kong. The entireislandwasinsideacountrypark,with 2.1.3. Data flow limited artificial lighting sources apart from public facilities. The remoteness of the site The NSB measurements taken from the SQM- also led to our selection of a different data LE could be transferred by Internet directly via acquisition mode for this station due to the a TCP connection to the survey computer server limited mobile Internet strength (cf Sec- (the main server). With a majority of monitoring tion 2.1.3). stationslocatedatthird-partyproperties,themo- bile3GnetworkwasusedforInternetconnections SSh This urban station was installed on the toallthemonitoringstationstominimizethecost rooftop of a school located in a population and labor required. centerinnorthernNewTerritories. Situated For all but one of the monitoring stations, we near the edge of town center, this observing adopted the active mode of data transfer. Each location has relatively wide FOV. This sta- SQM-LE unit acted as a TCP client, linking first tionwasthenorthernmostinourstudy, and to a commercial 3G router, and then a USB 3G its data could also be affected from the city modem for Internet connection. Each unit would of Shenzhen, China, at only 4 km away. 9 ateY) DY 0000000111 000001 01 nM- -1-1-1-1-1-1-1-1-1-1 -1-1-1-1-1-1 -1-1 o 5889922333 991113 15 ati-M 6-01-08-01-05-00-11-15-02-00-0 2-07-01-12-16-12-0 6-11-0 erD 2120112123 021122 13 pD O( ) el v e nl oa vatiese 653014010010255501020 8015552515 1510 ev Elbo a m ( E e 3508567521 933117 21 d 2389148648 250630 30 u 7403760927 216306 41 git 4.13.94.14.14.14.14.04.14.14.1 4.33.94.24.34.04.3 3.93.9 n 1111111111 111111 11 o 1111111111 111111 11 L N 0863209254 319663 91 de 12888383945166459266 082207767371 6796 u 3232244343 422324 32 atit 22.22.22.22.22.22.22.22.22.22. 22.22.22.22.22.22. 22.22. L d nt fie detailsoftheNSNmonitoringstations. Facility UrbanParkMeteorologicalStation,HongKongObservatoryMonitoringStation,EnvironmentalProtectionDepartmeKoonNatureEducationcumAstronomicalCentreObservatoryDome,HKUHongKongSpaceMuseumonitoringStation,EnvironmentalProtectionDepartmentHongKongWetlandParkOurLady’sCollegentiaCollege(SponsoredByEducationConvergence)POHChanKaiMemorialCollege RuraliObservatory,HongKongSpaceMuseumHauObservatory,HongKongAstronomicalSocietyTheSwireInstituteofMarine,HKUAstropark,HongKongSpaceMuseumHKPASilvermineBayCampMonitoringStation,EnvironmentalProtectionDepartment NotclassiMunGovernmentDepot,HongKongScienceMuseummentFlyingService,HongKongInternationalAirport graphical King’sngChungHo TaiPoM Elega Shui apMun TuenGovern o u T Ge T able1: Code KPTCHKnHKUTSTTPoTSWWTSSShST iObsSHCapAPMWoTpM TMDGFS T 10