Spatial and temporal analysis of artificial light pollution of the city night sky. A case study from Toruń

Several degrading factors contribute to the deterioration of the modern urban ecosystem, including those that have already been determined and those that are still at the stage of identifying the importance and extent of their impact. Changes in the quality of the natural environment, their assessment and the determination of their direction are the subject of frequent interdisciplinary studies undertaken by teams of experts. In these undertakings, it has become necessary not only to search for new, more efficient recording methods for conducting measurements as such, but also to increase interest in implementing targeted and remote monitoring of factors negatively affecting the environment. As a result, devices and integrated technologies are being developed that allow not only the measurement of a given parameter, but also data processing, visualisation of the results and their effective archiving. The current state of knowledge of environmental degradation, combined with modern Industry 4.0 technology, enables the optimisation of the measurement process in such a way that it is possible to better understand the scale of the phenomenon and more accurately determine its variability and range, as well as contributing factors. This interaction contributes to better targeting of research and automation of the measurement process itself, resulting in new information and ultimately a more complete knowledge of the impact of a given factor on the health and life of plants and animals, as well as the functioning of man himself.

Nowadays, one of the increasingly thoroughly studied phenomenon with a major impact on the human ecosystem is light pollution (Falchi et al. 2016; Jechow et al. 2017; Linares et al. 2020), which is defined as an excessive emission of artificial light at night over an extended period of time (ed. Szlachetko 2022). The phenomenon is caused primarily by incompetently or incorrectly designed or installed outdoor lighting, and is compounded by the presence of glare from LED advertisements and excessive illumination of architectural features. These factors cause excessive light emission at night, resulting not only in higher economic costs, but also in illuminating a space in the upward direction that should not be exposed to this process (Kołomański et al. 2015). Many scientific studies prove that prolonged residence in a light-polluted area causes many detrimental consequences for the health and life of plants, animals and, of course, humans (Adams et al. 2019; Garcia-Saenz et al. 2018; Jechow & Hölker 2019; Lacoeuilhe et al. 2014; Longcore et al. 2017; Macgregor et al. 2017). Research conducted on this phenomenon in urban areas also shows that light pollution further intensifies during adverse weather conditions, such as fog, cloud cover and increased presence of particulates (PM) of anthropogenic origin in the troposphere (Karpińska & Kunz 2023a; Kocifaj & Bará 2020; Ribas et al. 2016; Ściężor et al. 2010; Ściężor 2020).

The pollution of the night sky by artificial light is now a global phenomenon, which, with the development of industry and increasing urbanisation in recent decades, has steadily increased the spatial extent of its impact (Falchi et al. 2016). The human right to enjoy the dark sky is unquestionable, so in order to improve its quality and consequently human health and functioning at night, numerous governmental and non-governmental organisations, scientific and social entities have set themselves the goal for the coming years of providing adequate education to people of all age groups (International Dark-Sky Association; Ministry of the Environment of the Czech Republic 2022; ed. Szlachetko 2022). Another initiative involves the establishment of dark sky protection sites, recognising that they are one of the most natural elements of the world around us and therefore worth emphasising and promoting (International Dark-Sky Association; Karpińska & Kunz 2020b). Some European countries have also begun to introduce formal regulations on outdoor artificial light emission (Ministry of the Environment of the Czech Republic, 2022; ed. Szlachetko 2022), while several others, such as Croatia, Germany and France, have enacted laws to protect the environment from the effects of light pollution. These are, however, just few exceptions. The vast majority of countries do not have any legislation governing this issue and little attention is paid to publicising the problem of modern cities and to providing adequate education from an early age.

This paper discusses the results obtained from more than two years of continuous and targeted measurements of light pollution in the night sky over the area of Toruń. The city of Toruń is an example of a medium-sized settlement unit (with an area of about 116 km² and a population of up to 200,000), which so far has paid little attention to the operational status of municipal lighting. All processed data were acquired automatically by in-house developed measurement devices operating wirelessly in the monitoring network (Karpińska & Kunz 2022). Testing of the system under natural urban conditions was carried out in 2020, while fully operational operation of the light pollution monitoring network started at the beginning of 2021, and since then the network has been successively expanded with new measuring devices, with duration of the continuous observation series increasing month by month.

Initially, research into the phenomenon of light pollution was carried out by single researchers, mainly those interested in the deteriorating conditions for observation of the night sky. However, when the increasing impact of excessive light on the natural environment and living organisms was recognised, several research groups from around the world started targeted measurements to understand its variability and specificity, and to try to determine its consequences (Falchi et al. 2016; Hänel et al. 2017; Jechow & Hölker 2019; Kocifaj & Bará 2020; Kolláth 2010; Kołomański et al. 2015; Pun et al. 2014; Ribas et al. 2016; Ściężor et al. 2010; Ściężor, 2020). These teams use several different measurement methods, both highly complex and simple ones, available to amateurs. The most commonly used are measurements with a specialist photometer, which is available in manual (hand-held), portable, and stationary versions (Karpińska & Kunz 2019; Karpińska & Kunz 2022; Mander et al. 2023; Pun et al. 2014; Ściężor 2020). Measurements, e.g. with SQM and TESS devices, are carried out in various places around the world, both through the implementation of single measurements and longer measurement sessions (Bará et al. 2019; Cavazzani et al. 2020; Globe at Night; Pun et al. 2014; Ściężor et al. 2010; Zamorano et al. 2016). Another frequently used method is to take pictures of the night sky using wide-angle or all-sky lenses and then process them using professional software (Jechow et al. 2017; Karpińska & Kunz 2022; Kolláth 2010; Mander et al. 2023). As an alternative to ground-based surveys, remote sensing methods are becoming increasingly widespread, allowing the observation of excess light emissions using satellite imagery and aerial photographs (Ściężor, 2021). This enables a larger area of interest to be surveyed simultaneously and these acquisitions to be prepared periodically (Cinzano et al. 2000; Elvidge et al. 2013; Levin et al. 2020; Zhang et al. 2019).

In the area of Toruń, pioneering measurements of night sky light pollution were carried out in 2017–2018 using hand-held Unihedron SQM-L photometers (Karpińska & Kunz 2019; Karpińska & Kunz 2020a). The result of this work was, among other things, the first insight into the spatial distribution of light pollution in the city and a compilation of mean values obtained over 12 consecutive months. Figure 1 shows the first documented spatial distribution of light pollution collated for Toruń.

Figure 1.

The experience gained during these manual logging operations led to the preparation of another research project, which involved the establishment of an automatic monitoring network. Its main objective was long-term systematic recording at permanent monitoring sites carried out without human presence. In this way, it is possible to gain a better understanding of the temporal and spatial variability of light smog and to identify the main factors contributing to the deterioration of the night sky. The studied phenomenon is characterised by variability depending not only on the existing artificial light sources, but also on local atmospheric conditions and the phase of the Moon at the time of recording. The results obtained on cloudless, moonless nights differ significantly from the values obtained on cloudy nights, particularly in urban areas. During nights with changing cloud cover, the results obtained at different monitoring sites, even in relatively close proximity to each other, can vary considerably if taken at different times during one night. It is therefore very important that measurements are taken simultaneously at all sites set up in the area under study.

According to the assumptions made, the monitoring network was to be wireless, unmanned and cost-effective to set up and maintain. A battery-powered device of our own design was connected to the non-commercial wireless transmission network. Another assumption was to perform measurements of very low light intensity at night with spectral characteristics similar to commercially available SQM photometers. During the analysis of available optoelectronic components, it was decided to choose a light sensor with high accuracy, whose calibration was performed at the factory production stage (Adafruit 2019). The spectral response of the selected light sensor is similar to the human eye. Thanks to the use of two diodes (VIS+IR and IR) and their compensation, there is no need to use additional filters to eliminate part of the spectrum. It was assumed that the measurement session would be performed daily at 15-minute intervals, starting at 21:00 and ending at 06:00 CEST the following day. Data from the measuring devices will be transmitted via LoRaWAN technology (Mikhaylov et al. 2018; Semtech Corporation 2015) to a server and stored in the cloud disk. The selected technology enables efficient use of battery power, which improves the mobility of the devices. The aforementioned features of the system under development are in line with the strategy of Industry 4.0 and the spectrum of solutions used in the development and implementation of the Smart City concept.

The first stage of testing the monitoring system was to check its correct operation based on the case of two devices mounted on the roofs of the buildings of the Toruń Regional Development Agency located in the left-bank part of Toruń (Karpińska & Kunz 2022). The measurement of the night sky brightness and the communication between the devices and the network proved to be correct, whereby only the length of the message, the so-called frame size, was changed after analysis of the data sent to the server. An identified problem of the constructed recorders was the operating time of only a few days; this, was extended to more than nine months after important hardware and software improvements were made. Following these measures, the operation of the entire monitoring network of 40 loggers was tested. For this purpose, all devices were placed on the roof of the Observation Platform of the Faculty of Earth Sciences and Spatial Management (Nicolaus Copernicus University) in the vicinity of the operating factory-made SQM device and the results obtained in weekly measurement sessions were compared. In addition, measurements of the LoRaWAN network signal quality were carried out on the campus of Nicolaus Copernicus University (NCU) in Toruń and its vicinity to verify the capabilities of the LoRa network. This field experiment confirmed the spatial range of the selected data transmission technology of up to 4 km in a built-up area, i.e. in a typically urban landscape consisting of buildings of different sizes and heights, further diversified by tall vegetation (Karpińska & Kunz 2021). Measurements of the vertical variability of the night sky surface brightness were also carried out using UAVs, the purpose of which was to determine the limit heights above ground level for the placement of devices at the monitoring sites. These tests showed that above street infrastructure, the measured brightness of the night sky does not change, which makes it possible to place devices at any height above street lighting away from point light sources (Karpińska & Kunz 2023).

After the tests, field experiments as well as additional and complementary studies, the construction of a monitoring network based on measuring devices developed in-house was launched. The location of the measuring devices and access gateways of the LoRa network based on the status at the end of February 2023 is presented in Figure 1.

Measurement data recorded between the beginning of 2021 and the end of December 2022 were analysed in detail. During this period, the monitoring network was systematically expanded with new measuring devices. In 2023, further expansion is planned in the area of Toruń, especially in the central part of the city, both with new measuring devices (up to about 40 field stations) and access gateways (up to six devices in total). As a result, the city will be covered by a complete monitoring network, with a density of data loggers unmatched by any other settlement unit in the world.

The data collected over the two years of operation of the monitoring network has already allowed several elements to be identified, including the seasonal variability of the results, the relationship between the brightness of the night sky and meteorological conditions, the representation of the spatial distribution of the light smog phenomenon, and the comparison of the data collected with the results obtained at locations far from the influence of the urban agglomeration.

The measurement results are presented in mag/arcsec² units and are based on the astronomical unit: magnitude. When analyzing the results, it is worth remembering that it is a logarithmic and inverse unit. Results above 20 mag/arcsec² will represent dark skies, while results between 14 and 17 mag/arcsec² or less will represent bright, urban skies.

To check for the differences between measurements of the night sky surface brightness in relation to the astronomical seasons, all collected data were divided with respect to the observation season. The arithmetic mean of the registration results, the maximum value and the minimum value were calculated for each of them. The results thus obtained are presented in Table 1. The mean values over the analysis period were calculated on the basis of more than 46,000 recorded data, with an mean of 1,700 measurements per location. In the summer season, the mean value for all locations was obtained on the basis of more than 9,900 collected data, in the autumn season – 12,000, in the winter season – 12,200 and in the spring season – 15,500. The difference in these figures results from different dates on which individual measuring devices, systematically added to the monitoring network, started operating.

Figure 2.

The obtained mean, maximum and minimum values of the surface brightness of the night sky for Toruń were also presented graphically, as documented in Figure 3. The analysis was prepared using ArcGIS (Esri) software and geostatistical interpolation using the Empirical Bayesian Kriging method. The latter method is also used for small sets of input data. Spatial distributions of mean values in Toruń by astronomical seasons are also presented using the same tool (Fig. 4).

Figure 3.

Figure 4.

The results of the interpolation analyses presented in Figs 3 and 4 indicate that the highest mean values of surface brightness of the night sky coincide with the largest housing estates in Toruń. The eastern part of the city is characterised by dense multi-family housing, interspersed with a network of streets equipped with abundant municipal lighting. High light pollution is also found in the western part of Toruń, where older housing estates are located, characterised by compact and low-rise buildings. The presentation of the mean value of the data by astronomical seasons clearly shows that the highest light pollution is observed in winter, while the lowest in summer. According to previous studies, the brightness of the night sky is greater when various scatterers are present in the troposphere, such as particulate matter or fog. A significant increase in the recorded measurement values occurs mainly in the autumn and winter seasons. In addition, light pollution is further exacerbated when cloud cover is present, which in Poland is more common in winter. The impact of this element on sky brightness measurements is documented by the analyses described in the next section.

Comparing the two interpolations shown in Figures 1 and 3, which come from different measurement periods (2017–2018 and 2021–2022) and were obtained using different data acquisition methods (commercial, factory-made SQMs vs. devices of our own design and construction), we can see a strong similarity in the spatial distribution of the phenomenon across the city – there is an overlap between areas with the brightest sky, i.e. the most polluted ones, and those with relatively better night sky quality. The difference can only be seen in the north-eastern part of the city, where an inappropriate site was selected during the first measurements, as it was located too close to an intensive source of outdoor lighting. Consequently, this element significantly overstated the measurement in its vicinity. The results obtained in the last two years with our measuring devices operating in the monitoring network are more accurate, mutually comparable and obtained simultaneously at identical time intervals, and thus not burdened with errors of non-simultaneity of measurements. The identified flaws and imperfections of the manual data acquisition process carried out in 2017–2018 during the first pioneering light pollution studies in Toruń were thus improved and automated.

Meteorological conditions prevailing at the time of logging affect the level of values recorded by photometers. With a long and continuous observation series, one can try to determine the relationship between these variables. Therefore, the relationship between the measured value of the surface brightness of the night sky and the atmospheric factors was analysed. The following were used to determine the linear correlation: cloud base height expressed in metres above the ground, cloud cover in oktas, where 0 means clear sky, while 8 means a completely cloudy sky (completely overcast) and 9 is used when it is impossible to determine the degree of cloud cover (sky obstructed from view) due to e.g. fog, and visibility expressed on a scale from 0 to 9, where 0 indicates the worst visibility and 9 indicates perfect visibility.

The meteorological data used as a basis for the analysis were recorded at the Meteorological Station of the Institute of Meteorology and Water Management (IMGW) Toruń-Wrzosy in the period 2021–2022. Until the end of March 2022, data were collected 24h a day on an hourly basis, but from April 2022 only automatic measurements were recorded at night. Total cloud cover is determined manually, so no night-time measurements of this element are available from April 2022. Furthermore, the summer season is characterised by the absence of an astronomical night, so only measurements taken on the hour hour between 22:00 and 01:00 CEST were included in the data comparison, and during the summer solstice this time window was further narrowed to the hours between 23:00 and 00:00 CEST due to the noticeable influence of sunlight.

Due to the large number of cloudy days in Toruń, it is necessary to take meteorological conditions into account in the analysis of night sky brightness measurements. According to the data from the IMGW Toruń-Wrzosy Meteorological Station (Table 2) for the period 2019–2022, as many as 30% of nights were completely cloudy (overcast skies), 45% – partly cloudy to varying degrees, and about 25% – cloudless (clear skies).

The linear correlation between the selected factors and the surface brightness of the night sky is presented in Table 3. The table shows the correlation results from each device related to both the entire period of analysis and the individual astronomical seasons.

The reduced number of comparisons for the total cloud cover parameter results from the aforementioned unavailability of these data, starting from April 2022 onwards. However, even with this number of analyses, a high correlation between this parameter and the measurements made is evident. The correlation value for almost all locations in each distinguished season and on an annual basis is above 0.5. The correlation results are negative due to the selected mag/arcsec² unit for the presentation of the phenomenon – the more clouds present the lower the result.

The results of the correlation analysis between the height of the cloud base and the values of the night sky brightness recorded by measuring devices, carried out both for the whole year and individual seasons, are also presented graphically in spatial terms. Figure 5, prepared using the point cartogram method, shows the value of correlation relative to the location of a monitoring site on the map of Toruń.

Figure 5.

The results presented in Figure 5 indicate the highest correlation for locations with considerable light pollution resulting from the presence of a large number of street lamps whose light is reflected by clouds and derivatively increases the values of the recorded parameters. The correlation between these variables varies throughout the year. Most of the monitoring locations in each season showed a high correlation between the measurements and the cloud base height and total cloud cover. However, such a correlation is not observed for the visibility. Most values above a correlation of 0.5 were obtained in autumn, followed by summer and winter, while the weakest correlation was determined in spring. The underestimated correlation for this period may be caused by additional factors, such as numerous light scatterers in the form of particulate matter or water particles in the atmosphere, but also by the decreasing duration of the night. Even though only the hours between 23:00 and 00:00 were selected for data logging during the summer solstice, this may have affected the results and the smaller number of comparisons for correlation.

The analysis was further extended by comparing the night sky pollution values recorded in the urban area of Toruń with locations away from larger human settlements. To this end, it was necessary to use factory-made SQM photometers, as the LoRaWAN network signal could not be used outside the city limits of Toruń. A total of three such devices further enriched the observations, and they were deployed in a decreasing gradient of human impact. One of the devices was deployed at the Integrated Environmental Monitoring Station in Koniczynka, located 10 km north-east of Toruń’s boundaries, while the second was installed in the forest hamlet of Klaniny in Tuchola Forest (about 100 km north of Toruń), away from any influence of light sources. The third SQM photometer is located on the Observation Platform of the Faculty of Earth Sciences and Spatial Management NCU and has already been used for testing all of the proprietary devices at the technology demonstrator stage. The results from the three locations were collated and compared with each other in the context of increasing distance from a large city and a decreasing gradient of impact exerted by the urban fabric. In addition, the results obtained by the three devices operating in the urban monitoring network located in areas with identified significant light pollution were added to the comparison (Fig. 6).

Figure 6.

Analysis of Figure 6 shows a significant difference between the measurements obtained at these selected locations. The data recorded in Tuchola Forest (KLAN) show the highest values, which translates into the lowest light pollution in this area, while the measurements made at the station in Koniczynka (KON) show the influence of the urban light island of Toruń. This graph also shows the previously proven effect of cloudy nights on the level of measured values. When comparing these measurements made over several months, one can also see the effect of the full moon on the recorded value. This is very apparent from the data obtained in the village of Klaniny. This effect, however, is not observed when juxtaposing the phases of the Moon with measurements made in urban areas. This means that the presence of street infrastructure has a significantly greater impact on the measured values in urban areas than the natural light of the Moon in inhabited areas.

A similar relationship is presented in Figure 7, which shows the results of measurements carried out over several months at two selected stations (device No. 3 located on the premises of Nicolaus Copernicus University and No. 23 located closer to the city centre). A moving mean has been superimposed on the graph, making the seasonal variation in the measurements mentioned earlier more pronounced. In addition, the time of the full moon (in this case marked with grey dots at the top of the graph) and the cloud cover occurring at that time, presented on the oktans scale, were added to the graph. The presented data show an increase in the measured brightness of the night sky at the time of the full moon. This, however, is not an absolute rule, but a certain regularity. A greater effect of variation in the surface brightness of the night sky is observed when the sky is completely overcast. A stronger correlation of the obtained results is demonstrated by device No. 3, located further from the city centre, which confirms the previous results of the analysis shown in Figure 6. The data cited suggest that it is necessary to further investigate the influence of various meteorological factors on the increase in recorded values of the surface brightness of the night sky, so that in future it will be possible to gain an even better understanding of this phenomenon and to develop a more accurate model of light pollution taking into account different atmospheric conditions.

Figure 7.

Pollution of the night sky by artificial light is becoming an increasingly understood and observed (not only by specialists) phenomenon that degrades our environment in which we function and live. To better understand it, interdisciplinary research is being conducted to determine both its extent, scale and severity, as well as its nature and impact on individual living organisms. The growing interest in the subject is evidenced by the increasing number of scientific articles and conference presentations published in recent years (Levin et al. 2020; Mander et al. 2023). This study presents the occurrence of this phenomenon in the Toruń area, its spatial distribution and magnitude, as well as its seasonal variability and correlation with selected atmospheric conditions. The presented results show the complexity of the measurement process and describe a comprehensive analysis of light pollution carried out in an urban area. In the process, the role and impact of external factors on the recording of measurement values, resulting in the increased emission of artificial light at night, were also estimated. The winter period shows the highest light pollution further aggravated by the presence of numerous scatterers of anthropogenic origin in the troposphere, while in summer the effect is considerably smaller. Presentation of the annual data correlation with typical meteorological elements indicates their effect on the increase in surface brightness of the night sky in urban areas. During the continuous two-year research in Toruń, no significant effect of the full moon on the brightness of the night sky was observed. This effect is dominated by the presence of mainly street lights, LED advertisements and the illumination of architectural objects. This is the opposite relationship to that observed in areas far from city lights, where the full moon significantly affects measurements of the night sky brightness. The automatic network for monitoring night sky pollution, designed and operated in Toruń for the purposes of the project, proved to be fully effective and indispensable for continuous data acquisition aimed mainly at better understanding of the phenomenon and its accurate spatial mapping. The correctness of the operation of the low-cost measuring devices designed from scratch and the selection of the locations of the loggers to cover full and comprehensive acquisition in an urban area of this surface size and type of development was also operationally verified. The developed method of data acquisition in an urban area, which uses Industry 4.0 technologies and smart transmission grids, appears to be a good complement to the methods already in use for monitoring not only the pollution of the night sky by artificial light, but also other environmental parameters that will require monitoring in the near future. The obtained results provide an excellent basis for planning further analysis and research into light pollution in urban areas. Toruń is one of the few cities in the world for which targeted studies of light pollution in the night sky have been carried out in such a dense scattered measurement network and which has such a long series of simultaneous and comparable ground-based observations.

The next stage of the work should be a complex analysis of the relationship between the brightness of the night sky and all elements that can affect its value, such as the presence of atmospheric particulate matter of various sizes, fog, cloud cover, snow cover or technological solutions used for outdoor lighting in terms of of the colour of the light, the temperature and operating time.