Varying Performance of Low-Cost Sensors During Seasonal Smog Events in Moravian-Silesian Region

1. Introduction

The Moravian-Silesian Region in the Czech Republic, particularly the city of Ostrava, is recognized as a significant air pollution hot spot in Europe. This situation arises historically from a combination of industrial activity, high population density, and geographical factors that worsen air quality levels. The topography of the Moravian-Silesian Region contributes to the accumulation of pollutants, especially during winter months when temperature inversions are common. This leads to poor dispersion, resulting in elevated concentrations of air pollutants. Additionally, air pollution from neighboring areas, particularly from the Silesian Voivodeship (Poland), also plays a significant role in the given context [1,2,3].

Recently, "Polish smog" was proposed as a specific type of air pollution that occurs in Poland, particularly in the winter months. It is characterized by high concentrations of particulate matter (PM), such as PM_2.5 and PM₁₀, as well as polycyclic aromatic hydrocarbons like benzo(a)pyrene [4]. This type of smog is particularly prevalent in Eastern Europe, where it arises from the burning of coal and other solid fuels for heating purposes, especially during winter months. When compared with photo-chemical smog found in industrialized urban areas, which is driven mainly by volatile organic compounds (VOCs) and nitrogen oxides leading to high ozone levels, Polish smog is more closely linked to residential heating practices and industrial emissions.

Yet, another type of smog event, which is further considered through this paper is caused by particulate matter originating from Sahara desert. Dust storms can transport particles over thousands of kilometers, affecting air quality far from their source. The transport of Saharan dust poses significant challenges for air quality management and public health in Europe. Although this phenomenon is more common in the Mediterranean region, it can occasionally cause a significant deterioration in air quality for Central European countries.

Low-cost sensor (LCS) networks have emerged as a promising solution for monitoring air pollution and providing smog alerts as it can supplement data from regulatory-grade reference instruments [5]. By filling in spatial and temporal gaps in air quality monitoring such information can provides a more comprehensive understanding of pollution patterns both at local and regional level. Citizen science projects involving the public in sensor deployment can further expand the reach of these networks [6,7,8]. However proper calibration of LCS is crucial to ensure data accuracy and reliability [9].

In principle, uncertainties of factory calibrated LCS response are studied by experiments in controlled (laboratory) or uncontrolled (field) ambient conditions [10,11,12]. Such calibration procedure is followed by selection of suitable numerical correction method and estimation of parameters using reference datasets for model training or testing purposes [13]. Co-location of the LCS node with reference instruments in real outdoor atmosphere is usually performed over the period of several weeks in order to accumulate appropriately large datasets required for the reliable outputs of the calibration process. Prolonged period of co-location enables to investigate seasonal variability of LCS performance across a wide range of environmental conditions relevant for the target locality.

Following questions are mainly addressed through the paper: Which information (e.g. in relevance to public smog alerts) can be extracted from data provided by LCS and CAMS model predictions and what is their case-specific reliability compared to reference data from air quality monitoring (AQM) station?

2. Materials and Methods

Our prototype LCS sensor node (more details described in Appendix A) was mounted of the roof of AQM station, see Figure 1. This setup allows for direct comparison between the outputs of the LCS node and data from reference-grade instruments.

2.1. Selection of Low-Cost Sensors

Following our goals, LCS node is equipped with a set of low-cost sensors suitable for monitoring of target air pollutants for the specific area, which is particulate matter (PM), carbon monoxide (CO). Selection of the gas and PM sensors was mainly based on the extensive literature survey and experience of previous investigators. Availability of the sensors (distribution in EU) were also taken into account as well as affordability of the entire LCS node setup and level of complexity relevant to requirements for its integration and further development.

For particulate matter (PM) measurement, the node utilizes two sensors: Sensirion SPS30 and Alphasense OPC-N3. The SPS30 is a laser-based optical sensor well-suited for measuring fine particulate matter (especially PM₁) mass concentration based on the principle of light scattering. The OPC-N3 also uses optical particle counting, when detecting a wider range of particle sizes, from 0.35 µm to 40 µm, across 24 size bins enabling particle size distribution to be determined, which is critical for understanding the composition of atmospheric aerosols and determination of their origin in relevance to source apportionment.

For carbon monoxide monitoring, the LCS node incorporates the Alphasense CO-B4 electrochemical sensor. This sensor can detect CO concentrations between 0 to 1,000 ppm with a resolution of 0.1 ppm. Its sensitivity ranges from 55 to 85 nA/ppm, providing accurate detection of small changes in CO levels.

Knowledge of atmospheric pressure, temperature and humidity is essential when aiming at corrections of LCS response in various ambient conditions. For this purpose, digital module including Bosch Sensortech BME280 sensor was integrated in LCS node. This sensor should operate with pressure accuracy of $\pm 1$ hPa, temperature accuracy of $\pm 1^{\circ }C$, and humidity accuracy of $\pm 3\%$ RH.

2.2. Co-Location Site and Reference Instruments

The co-location measurements of the LCS node were conducted at an air quality monitoring (AQM) station of Health Institute located in the municipal area of the city of Ostrava, which is close to various industrial sites (e.g. metalurgical, chemical, etc.). The AQM station provides the information on meteorological conditions, i.e. wind speed, wind direction as well as reference air quality data in hourly intervals. Atmospheric pressure, temperature, and humidity are measured using the COMET T3113D sensor for temperature and humidity, and the NXP Semiconductor MPX4115A for pressure.

For CO measurement, the HORIBA APMA-370 analyzer is used, which operates on the nondispersive infrared (NDIR) spectroscopy principle. The TEOM 1400 analyzer was utilized as PM reference during entire winter evaluation period including S1 episode. This instrument measures PM₁₀ by drawing air through a filter mounted on an oscillating balance, with the mass of collected particles altering the oscillation frequency to calculate PM concentration. At the end of March 2024, the TEOM 1400 was replaced by Palas FIDAS 200 analyzer, which employs optical scattering technology for continuous real-time reference measurement of size distribution and following quantification of PM₁, PM_2.5, and PM₁₀.

2.3. Overview of Co-Location Measurement

Low-cost sensors are often collocated with reference instruments in laboratory or field conditions for period of several weeks in order to improve their performance. In our case, an evaluation measurement campaign lasting for three months (from mid-November 2023 to mid-February 2024) was initially planned to be carried out in order to verify and validate the performance of individual LCS and their variation for seasonal meteorological conditions typical in the Moravian-Silesian region.

Accidentally, smog alert events, which occurred in Ostrava during December 2023 and April 2024 were also recorded during extended period of co-location. These datasets enabled us to focus our attention on performance of LCS and CAMS data versus reference measurements during the episodes of serious air quality deterioration, see Table 1.

It is worth to note that atmospheric conditions for S1 episode is typical for above mentioned "Polish smog". In the given case the highest concentrations of particulate matter are generally recorded at low temperatures, specifically between -10°C and 0°C. Additionally, higher atmospheric pressure correlates with increased PM concentrations, as stable air masses inhibit vertical mixing and allow pollutants to accumulate near the surface. [4]. Evolution of above described meteorological situation can be identified from Figure based on reference data from AQM station.

Different atmospheric conditions for S2 episode corresponds to seasonal Saharan dust storm over Central Europe1. Moderate speed of wind blowing from south-west direction can be clearly identified during PM₁₀ maxima from the Figure 2.

2.4. Datasets and Preparatory Analysis

Reference 1-hour averaged data were extracted from monthly datasets provided by the Health Institute (AQM station) after verification involving and replacement of values of measurement below limit of detection ($LoD$) by the value equal to $LoD/2$.

Similarly, co-location site-specific timeseries of concentrations for selected pollutant (CO, O₃, NO₂, PM₁₀, PM_2.5 and Dust) based on the forcast of the CAMS ENSAMBLE model [14] with 11 km spatial resolution were downloaded in the form of comma-separated value (CSV) files from Open Meteo webpage. Monthly datasets extracted from LCS node (10-minute averages of sensor readings available from SOASENSE project webpage were converted into hourly timeseries using pandas (Python library) resample method and ordered with reference and CAMS model data according to relevant GMT timestamps. Multilinear fitting (MLR) of the carbon monoxide LCS voltage were converted into concentration using linear model form scikit-learn (Python library). Entire dataset of winter measurements, i.e. from November 2023 to February 2024 was assumed as representative for the given step. Temperature reading form LCS node was converted into Kelvins in order to avoid numerical issues relevant to negative values. Entire dataset was splitted into training and testing subsets having 1232 and 2392 datapoints respectively. Finally, MLR model having high coefficient of determination, i.e. $R^2=0.89$ and acceptable Mean Average Error ($MAE<50$$\mu$g/m³) for predicted CO concentration was estimated asuming only T[K] and CO-B4 sensor voltage values as predictors with presumed parametrization of the Equation 1:

$${\left[CO\right]}_{LCS}\left(\tau \right)=c_1\times [U_{WE}\left(\tau \right)-U_{AE}\left(\tau \right)]+c_2\times T\left(\tau \right)+c_3$$

(1)

where $U_{WE}\left(\tau \right)$ is a working electrode voltage [V], $U_{AE}\left(\tau \right)$ is an auxillary electrode voltage [V] at a given Geenwich (i.e. GMT) time $\tau$ and case-specific values of relevant MLR coefficients are as follows: $c_1=4394$, $c_2=-0.693$, $c_3=0$.

Exploratory data analysis including Correlation Matrix and Kernel Density Estimation (KDE) of dataset pairs was performed using seaborn (Python library). Subsequently, simple linear regression (SLR) and plots of diurnal variations of air pollutant concentrations were performed employing relevant methods implemented in atmospy (Python library). Particle size distribution measured by Alphasense OPC-N3 sensor were analyzed using the smps-py (Python library).

3. Results

3.1. Key Findings from Winter Evaluation Period

In winter, the time series for key air pollutants are expected to be significantly correlated, which was confirmed from analysis of the reference data (PM₁₀, CO, NO₂, O₃). Carbon monoxide shows strong linear correlation with PM₁₀ having a Pearson coefficient value of $r=0.85$, see Figure 3.

At the same time, the NO₂ concentration values are also correlated with CO and PM₁₀, which can be attributed to a similar source of these pollutants or to a related mechanism. This observation is in line with an assumption, that incomplete combustion of fossil fuels and subsequent atmospheric dispersion of smoke plumes is dominant contributor to winter air pollution in the given location. As can be expected, due to the lower solar (UV) radiation intensity, ozone concentrations only rarely reached elevated values (above 75 $\mu$g/m³) during this period. However, the values of Pearson coefficient for the negative correlation of O₃ with other pollutants suggest that the ozone reactivity influence atmospheric chemistry even in the colder part of the year, mostly in the case of NO₂.

Most prominent qualitative features of the LCS response and selected CAMS model predictions can be recognised from the plots of selected hourly data (Figure 4) when compared to reference measurements. Firstly, we can spot excellent fit of [CO]_LCS to [CO]_REF together with systematic shift of [CO]_CAMS data towards slightly higher values. On the contrary, PM₁₀ response provided by SPS30 sensor is considerably overestimated (especially for PM₁₀>50 $\mu$g/m³) compared to PM₁₀ prediction by CAMS model which is very close to reference data.

Negative correlation of major pollutant concentration ambient temperature is also evident from the plot depicted above. Numerous spikes in CO and PM₁₀ timeseries follows sharp decrease of ambient temperature (mostly below $0^{\circ }C$). We can also conclude excellent agreement of temperature from LCS node with reference, which is important due to requirement of correcting CO-B4 sensor according to Equation 1.

The overall quality of the LCS node data and the qualitative indicators of the CAMS model compared to the reference measurements are also illustrated by the plots shown in Figure 5. The average reference CO concentration over the evaluation period shows the same trend of diurnal variation as PM₁₀.

In the case of carbon monoxide LCS data shows nearly perfect coincidence with the reference diurnal trend. There is obvious increase of CO and PM₁₀ during morning rush hours (at about 8 AM) and second peak around 8 PM. While for the morning peak the [PM₁₀]_CANS value (predicted by CAMS) coincides with the reference measurement of PM₁₀ ([PM₁₀]_REF).

As in the case of the CO diurnal variation plot, the peak PM₁₀ concentration observed in the late evening is significantly overestimated by the CAMS model, which may be related to uncertainty in emission factors or data related to industrial and domestic heating, which is dominant in the winter. For example, the Czech Republic is currently undergoing technological improvements based on government support for the replacement of domestic heating systems, which may have already been effective but has not yet been reflected in the model inputs. The plot of diurnal SPS30 response, i.e., [PM₁₀]_LCS, shows a large overestimation with a more pronounced deviation during the night hours.

In the following sections of this paper, we anticipate that this change is due to a combination of the specific sensitivity of the SPS30 sensor and the changing ratio of fine to coarse particles in the air over the course of the day and year due to different intensities of domestic and industrial combustion and smoke dispersion, see Section 4.

Prior to the start of this study, we assumed that the Alphasense OPC-N3 sensor integrated in the LCS node would be used to accurately determine PM concentrations as complementary measurement to the TEOM reference. However, it has been found that OPC-N3 response do not show adequate agreement with the reference instrument ($R^2<0.3$ for PM₁₀ from the evaluation period). Nevertheless, the particle size distribution was continuously measured by this sensor throughout the co-location period. Thus we used these data to determine the particle size-resolved spectra. Particle volume concentrations for each of 24 bins were determined from measured values (i.e. number of particles counted per sensor bin) and their daily mean values were recalculated. Despite considerable noise and uncertainty in the response of few bins (e.g. between 3-4 $\mu$m), some seasonal trend of fine and coarse particle concentration is evident from data depicted in the Figure 6.

3.2. LCS Performace During Polish Smog

Figure 7 shows highly linear response of the CO-B4 sensor during S1 episode. The slope parameter is nearly equal to its value obtained for entire evaluation period. This fact can be expected due to presence of high concentrations datapoints recorded during the smog episode. However, we can find close agreement of the data with initial [CO]_LCS calibration according to Equation 1. Note that data points recorded, when the values measured by the reference instrument were below the detection limit (i.e. [CO]_REF < 200$\mu$g/m³) were removed before the SLR analysis.

Systematically over-predicted PM₁₀ response was obtained from SPS30 sensor compared to TEOM reference instrument during S1 episode. Nevertheless, similar value of slope parameter a was obtained from SLR fit (where a is the slope of $y=ax+b$ regression line) for the entire winter evaluation period, see Figure 8). These results indicate consistency of datasets obtained from the SPS30 sensor response together with relatively high coefficient of determination ($R^2>0.8$) for PM₁₀ over entire co-location period.

3.3. LCS Performace During Saharan Dust Storm

Dramatic difference in SPS30 response to coarse particles was observed during the S2 episode. Nevertheless, thanks to availability of innovated reference instruments (FIDAS) we were able to determine simultaneously SLR fit for fine particulate matter. Nearly perfect agreement of SPS30 response was found for PM₁, see Figure 9. It it worth to note that the concentrations of carbon monoxide during S2 episode were steadily below the limit of detection for reference instrument ([CO]_REF<200$\mu$g/m³).

4. Discussion

We first discuss our results regarding the response of the CO-B4 sensor and compare them with the observations of previous researchers. In the work of Camprodon et al. [8], a very high correlation ($R^2>0.8$) and low error ($RMSE<0.1$ ppm) of CO measurements were observed during more than two months of CO-B4 sensor deployment. The sensor was found to behave linearly with respect to the CO concentrations and its decrease during the co-location period was negligible, which is quite consistent with our measurements. Our data obtained during the winter evaluation period (3 months) show slightly higher coefficient of determination $R^2\approx 0.9$ when re-calibration according to the Equation 1 is evaluated and compared with the reference data. Similar performance of this sensor is reported in Han et al. [11], evaluating an almost identical season with similar ranges of air pollutants but with temperatures in the range 0-20°C. Our co-location was carried out at much lower temperatures, while temperature correction following Equation 1 seems to be less effective at extremely low ambient temperatures (below -10°C) and high CO levels, see Figure 10. Conversely, slightly overestimated values of [CO]_LCS were observed during warmer days (with $t>$5 °C). In the given case, we can attribute biased [CO]_LCS values to direct temperature effect on sensing mechanism, i.e. reduced rate of (electro-)chemical reactions, and corresponding non-linearities.

On the other hand, as far as the influence of temperature on the response of the SPS30 sensor in our local conditions is concerned, we anticipate rather an indirect effect consisting in the change of particle size distribution due to the increased need for domestic and industrial heating at lower ambient temperatures.

This hypothesis is consistent with a number of previous publications, e.g. [15,16], mentioning in particular the work of Zareba et al. [16], which show a negative correlation between ambient temperature and the air pollution in an area close to our co-location site. Their study confirms that in moderate climate zones with coal burning as the primary source of air pollution, temperature is the most significant factor influencing monthly average PM₁₀ concentrations.

As in the case of our co-location site, many AQM stations in the region covered by this paper are not yet equipped with reference instruments for measuring fine PM fractions. Moreover, the air quality criteria recommended by WHO, EU, or local authorities for declaring a smog alert situation usually take into account PM₁₀ concentrations or rarely PM_2.5. Therefore, our main motivation was to find a solution to reliably determine PM₁₀ values based on LCS data.

Below we briefly summarize some of the findings from previous studies on the performance of SPS30 sensors, in particular on their reliability in measuring fine and coarse PM concentration.

In a study of Roberts et al.[17] co-locating the SPS30 with regulatory methods, it achieved an average bias adjusted $R^2=0.75$ for 24-hour averages and 0.57 for 1-hour averages, suggesting reasonable accuracy in real-time monitoring. The mean bias error was minimal, indicating that the SPS30 provides reliable data for PM_2.5 levels.

According to Kuula et al. [18] the SPS30 sensor is suitable to be used for measuring PM₁ particles when $R^2=0.91$, indicating high accuracy and consistency. Whereas for PM_2.5 particles this value was 0.83 and for PM₁₀ particles it was 0.12, which is characterized by low measurement reliability and this sensor is not suitable for larger particle sizes.

Vogt et al. [19] also confirm that the SPS30 sensor is the mostly accurate and reliable for PM₁ particles with $R^2=0.94$. For PM_2.5 particles the $R^2$ value was around 0.73. The results for PM₁₀ particles indicate a higher value ($R^2=0.46$) compared to the results of Kuula et al. [18], yet still not suitable for practical AQM applications.

Molino Ruada et al. [20] confirmed the trend of the SPS30 sensor being able to measure PM₁ particles with a high accuracy of $R^2=0.93$. As particle size increases, the accuracy of the sensor decreases yielding $R^2=0.72$ for PM_2.5 and $R^2=0.23$ for PM₁₀, respectively.

The physical explanation for the unreliable measurement of larger particles is related to design of optically based LCS and principle of their operation (i.e. light-scattering). Above all, shortened viewing angles, losses occuring during particle intake and also differences in particle shape and refractive index needs to be taken into account as well as the effect of humidity and sensor aging, when these LCS are exposed to realistic outdoor conditions.

Considering these findings together with the results of our LCS node measurements against the reference data, we can conclude that the SPS30 provides a reliable response to fine dust particles, especially PM₁, even under Saharan dust storm conditions. The PM₁₀ readings from the SPS30 sensor according to its original calibration (i.e. factory setting) are burdened with a systematic bias whose trend (negative or positive) depends on the type of smog situation. Therefore, to conclude this discussion, let us take a closer look at the size-resolved histogram of the PM volume concentration distribution obtained from the OPC-N3 sensor on days with maximum PM₁₀ concentration in the case of S1 and S2 episodes, see Figure 11. The difference in particle size resolution is noticeable, with both data showing significant bimodality. In the case of the Polish smog (S1), the total volume is clearly dominated by PM₁. On the other hand, in the case of the Saharan dust storm (S2), particles with aerodynamic diameter $D_p\approx 4\mu$m have the highest volume concentration from the total PM₁₀ found in the size-resolved distribution.

In analogy to recent work by Kaur and Kelly [21], we propose a strategy to derive PM₁₀ concentrations from the biased PM-LCS response based on correction factors obtained from the OPC-N3 sensor working in concert. Further assume an Equation 2 and Equation 3, which can be used to adjust the slope $a_{SLR,S1}$ and $a_{SLR,S2}$ respectively to ideal value $a_{COR}\approx 1$. Then we can use the inverse estimation in order to determine that the calibration of SPS30 is presumably carried out with the aerosol mixture having [PM₁/PM₁₀]_SPS,calibr≈0.2 ± 0.05, which corresponds to a common (traffic-related) air pollution in urban areas.

$${\displaystyle \frac{{[P{M}_1/P{M}_{10}]}_{OPC,S1}}{a_{SPS,S1}}}={\displaystyle \frac{0.5}{2.1}}=0.23\approx {[P{M}_1/P{M}_{10}]}_{SPS,calibr}$$

(2)

$${\displaystyle \frac{{[P{M}_1/P{M}_{10}]}_{OPC,S2}}{a_{SPS,S2}}}={\displaystyle \frac{0.03}{0.19}}=0.16\approx {[P{M}_1/P{M}_{10}]}_{SPS,calibr}$$

(3)

Therefore, the PM₁₀ values measured by SPS30 are systematically biased if the actual [PM₁/PM₁₀] values differs significantly from the [PM₁/PM₁₀]_SPS,calibr. In other words, it was proved, that the biased SPS30 reading of PM₁₀ can be roughly corrected using [PM₁/PM₁₀]_OPC divided by a factor ($0.2\pm 0.05$). More precise corrections will only be possible after further analysis and experimentation.

4.1. Practical applicability

This work represents a significant step towards strengthening the role of citizen science and democratizing environmental data in AQM, and demonstrates the importance of academic support for these efforts as the current state of knowledge and technology is still rather prohibitive to the straightforward deployment of commercially available LCS systems in their default (factory calibrated) setup. Therefore, careful evaluation of LCS performance (in the form of co-location measurements) and consideration of specific conditions of their deployment at local and regional level before their practical application is inevitable.

In the framework of this work, we have been able to explain the seasonal variability of the Sensirion SPS30 sensor response and a correction method increasing the reliability of its PM₁₀ response have been established. According to our findings, we can exploit the strengths of the SPS30 sensor and overcome its previously reported limitations. Correction of biased response can be expressed based on fine-to-coarse particle ratio, e.g. as PM₁/PM₁₀ evaluated from OPC-N3 sensor. It was also found that additional temperature correction needs to be estimated for CO-B4 sensor to account for biased response at extremely low temperatures.

Our future aim is to enhanced reliablity of the regional AQM data when combining CAMS model predictions and LCS response by the means of machine-learning approaches employing parameterized (e.g. MLR or HDMR [22]) or non-parameterized methods [23].

4.2. Limitations

This study has several limitations, mainly due to seasonal character and the influence of weather conditions relevant to the location and the winter season. It also specifically focuses only on the response of selected LCS systems integrated into a prototype node that is still under development. In our study, only individual pieces of the selected LCSs were tested and evaluated, thus not including the influence of inter-unit variability. Due to the duration of the co-location measurement, LCS aging factors were neglected.

5. Conclusions

Prototype LCS node was successfully developed and co-located at AQM station in Ostrava-Mariánské Hory for the period of several months. Important features of LCS response was observed during seasonal smog episodes when searching for appropriate calibration procedure and methods. It was found that the key question in this context is not how to perform sensor calibration (i.e. which mathematical method is applied) but when an where the co-location measurement is carried out (with case-specific circumstances). Our results demonstrate the potential of LCS-based networks for reliable monitoring of major pollutants (CO and PM₁₀) during smog situations that occur during the winter months in the Moravian-Silesian region.

Appendix A.1. Hardware Description

The LCS sensor node is designed as a modular system with three interconnected printed circuit boards (PCBs), each serving a specific function: control and communication, sensor interface, and power management.

The Control and Communication Board houses the LilyGo TTGO LoRa32 T3 v1.6 with 868MHz microcontroller module, featuring the ESP32. This board manages data collection and communication, utilizing Wi-Fi (employed within this work) and LoRaWAN (optional). It operates primarily at 3.3V, with voltage regulators and DC-DC converters ensuring stable power supply for consistent performance.

The Power Management Board distributes power across the system and supports multiple power input options, including DC from the grid (employed within this work), lithium-ion batteries, and 6V lead-acid batteries (optional). This board includes DC-DC converters and voltage regulators to step down input voltages to the necessary levels. It also integrates a slot for NEO-6M GPS module for geolocation, providing real-time position data alongside environmental measurements.

The Sensor Interface Board integrates individual sensors, including particulate matter (Alphasense OPC-N3 and Sensirion SPS30). It also features an Analog-to-Digital Converter ADS1115 to process signals from analog Alphasense type B sensors. Stable voltage of 3.5V is provided for Individual Sensor Boards as required for measurement accuracy.

The interconnection of these PCBs ensures seamless communication and power distribution, contributing to the mechanical stability of the sensor node and simplifying assembly and future modifications.

Additionally, a custom-designed enclosure for the sensor node was developed and manufactured using MDF 3D printing technology. The material chosen for the enclosure is ASA (Acrylonitrile Styrene Acrylate), which offers excellent weather resistance and durability. The enclosure not only protects the sensors from external elements but also ensures proper airflow for accurate measurements. A 3D model of the sensor node’s enclosure is shown in Figure A1.

Figure A1. Rendered 3D model of the LCS node enclosure.

Figure A1. Rendered 3D model of the LCS node enclosure.

Appendix A.2. Datalogging

Each sensor node is equipped with an ESP32 microcontroller, which gathers data at regular intervals. This data is transmitted using Wi-Fi via the MQTT protocol, specifically through the Mosquitto MQTT broker, which publishes the data to designated topics. Telegraf, configured to listen to these MQTT topics, subscribes to the relevant data streams and stores the incoming information in an InfluxDB database. To make this data accessible and interpretable, Grafana connects to the InfluxDB database and visualizes the data on customizable dashboards. Selected datasets from co-location measurement are resampled with 10-minutes temporal resolution and automatically transferred to CSV format in monthly intervals. These data can be accessed from publicly available web interface SOASENSE.