Sage Journals: Discover world-class research

Abstract

Air temperature variability in an urban environment is considered a vital element for understanding urban climate. The advancement in sensor design and data collection methods have made it possible to install and collect cost-effective urban climate data. The user-friendly applicability, sustainability, expandability and acceptance in the scientific community of the All-In-One (AIO) and Low-Cost (LC) sensors have ensured their popularity in an urban environment. The selection of air temperature sensors based on sensor accuracy, cost and various selection methods are being applied in scientific research. In this study we present a comparative assessment of air temperature sensors through a two-stage selection method. First, we selected the sensors (METSENS 600, ClimaVUE50, WXT536 from AIO and nMETOS180, Netatmo, SenseBox Home, iButton, HOBO-U23 from LC sensors) based on commercial availability, cost, accuracy, sensor weight, power supply and consumption, literature review and user-friendly data management system. Second, we tested the sensors applying climate chamber (CC) and in-situ testing. We found that HOBO-U23 had the lowest Mean Absolute Deviation as compared to the reference sensor in the daily cycle test, nMETOS180 in the extreme range test and in stable temperature testing during CC testing. The time series and statistical results for AIO sensors indicated that ClimaVUE50 had the lowest Mean Absolute Deviation (MAD), Root mean square deviation (RMSD), Mean Square Deviation (MSD), and highest coefficient of determination (R²) values among the AIO sensors. Among the LC sensors, nMETOS180 showed lowest MAD, MSD, RMSD and highest R² values while Netatmo had the highest MAD, MSD, RMSD and lowest R² values. Deviations were mainly higher during the day and lower during the night due to radiation shield type variabilities. AIO sensors responded more slowly to changing temperatures due to thermal inertia and compactness as compared to most LC sensors. The study highlights clear deviations between AIO and LC sensors. Such deviations between sensors need in-depth investigation when comparing results between different sensors and different urban meteorological networks.

Keywords

Air temperature sensors urban environment All-In-One (AIO)Low-Cost (LC) sensors comparison sensor selection process

Introduction

The increasing trend in urbanization in the last few decades has made urban climate observation important for human safety and climate-resilient city planning (Bai et al., 2024; Ramírez-Moreno et al., 2021; Zhou et al., 2025). Air temperature variability in an urban environment is considered to be a vital element for understanding urban climate (Fenner et al., 2017; Meier et al., 2017). In order to understand the variabilities and changes in urban climatic parameters, it is important to have a well-established network of meteorological observations stations (Chapman et al., 2015; Gubler et al., 2021). Urban meteorological networks (UMNs) can be classified on the defined parameters of urban scale, spatial and temporal resolution, meteorological variables and long term project continuity (Hart and Martinez, 2006; Lelovics et al., 2014; Muller et al., 2013; WMO, 2024). The Birmingham Urban Climate Laboratory operates a network of 25 automatic weather stations and 101 low-cost air temperature sensors distributed on the basis of urban surface types across the city mounted in schools and on street light poles (Chapman et al., 2015). The ‘Monitoring the City’s Climate and Atmosphere’ network consists of six automatic weather stations that are spread across the city of Ghent based on Local Climate Zones (LCZ) distribution design (Caluwaerts et al., 2020). The ‘Urban Climate Bern’ network (Gubler et al., 2021) focuses on microclimate variabilities through an urban observation network of 79 low-cost sensors selected due to their high energy efficiency and large storage capacity and which are distributed based on land specific classification involving LCZ, land cover type and building heights along with three automated weather stations (AWS) as reference stations. HiSAN network (Chen et al., 2019) consists of 100 air temperature and relative humidity sensors located throughout the city of Tainan based on land use and land cover features. The establishment of urban climate observation networks were found to be based on limited number of expensive but highly accurate sensors or large number of low-cost sensors with lower accuracies (Caluwaerts et al., 2020). Low-Cost (LC) sensors have become a notable part of environmental observation networks in the last two decades (Mao et al., 2019). LC sensors are not only lower in cost but are also used for specific measuring parameters with inexpensive production, maintenance and user-friendly operational features as compared to other sensors with similar application functions (Karagulian et al., 2019; Nalakurthi et al., 2024). All-In-One (AIO) compact weather stations are considered to be a single unit compact designed weather station containing multiple sensors along with manufacturer reported sensor specifications and calibration details (Dombrowski et al., 2021; Warne, 2017). The advancement in sensor design and data collection methods have made it possible to install and collect urban climate data (Chan et al., 2021; Ioannou et al., 2021) but AIO sensors need to be cost-effective for their selection in research projects (Chen et al., 2022; Dombrowski et al., 2021). Crowdsourced data, specially Netatmo sensors, have recently been applied in many studies (Büchau, 2018; Chan et al., 2021; Lussana et al., 2023; Meier et al., 2017; Peerlings et al., 2024). The debate in the scientific community over sensor type utility has intensified over the critical aspect of data quality and accuracy reliability (Chapman et al., 2023; Fenner et al., 2021). Criteria such as robustness, compact design, low power consumption, low cost, maintenance, compatibility with loggers, wireless communication (Rashid and Rehmani, 2016), user-friendly applicability, sustainability, expandability and acceptance in the scientific community have given the AIO and LC sensors priority over the traditional meteorological observation instruments in many applied and urban sensing projects (Burt and Podesta, 2020; Chapman et al., 2023; Chodorek et al., 2022; Fenner et al., 2021).

These days various companies such as Campbell Scientific, METER, Kipp and Zonen, YOUNG, LUFT, Vaisala, GILL, Pessl, Netatmo or Onset offer AIO and LC sensors for professional meteorological observations (Apogee Instruments, Inc., 2022; Burt and Podesta, 2020; Campbell Scientific inc., 2022a, 2022c; Chapman et al., 2015; Dombrowski et al., 2021; iButton, 2022; Netatmo, 2022; Onset Computer Corporation, 2022; Pessl Instruments, 2022; senseBox, 2022; Ulpiani et al., 2022; Vaisala, 2022). Very few studies have been published that undertake the comparison of these sensors regarding their usability and accuracy. The response time of air temperature sensor to changing temperature were analyzed by Burt and Podesta (2020) and Chodorek et al. (2022) who explored the impact of ventilation, sensor manufactured design and material variabilities on response times. Some studies concluded that there is development of certain uncertainty levels in the sensors after transportation (Aranda and Merlone, 2024) and long periods of usage (Chan et al., 2021; Chapman et al., 2023). Other studies have identified the specification of sensors like short term observations capacity (Whiteman et al., 2000) and spatial coverage limits like micro climate observations (Vasic et al., 2022). Air temperature sensors are selected on the basis of a single or multiple parameters such as accuracy levels, response time and long term stability (Chapman et al., 2023; Chodorek et al., 2022; Quanz et al., 2018). Finding a suitable sensor has been a matter of individual selection based on the targeted application and manufacturer specifications e.g. on sensor accuracy, usability, dimensions and costs etc. (Dumas et al., 2021). Kowal et al. (2020) applied heat exposure and humidity as environmental conditions on two temperature sensors (HMP 155 and Vaisala’s HMP 45AC) and concluded that time, high humidity, high and low temperatures and rapid temperature changes show increasing deviation on the repeated testing results. The temperature range of -20°C to +40°C has a much larger impact on the uncertainties of conventional sensors such as thermocouples and resistance temperature detectors (RTD) which have limitations in terms of precision and stability as compared to AIO sensors (Oukaira et al., 2024). Chodorek et al. (2022) applied response time criteria in a controlled environmental setup for the selection of LC temperature sensors that would be more applicable in mobile weather stations for urban climate coverage. Dombrowski et al. (2021) compared ATMOS41 AIO compact sensors with Vaisala WXT520 and reference weather station data to identify the uncertainties in air temperature observations during the nighttime and declared that an analysis with mean values sometimes underestimates the importance of lowest temperature observations especially in rainy seasons. Freitag et al. (2016) compared air temperature sensors (MP-101, HygroClip-2 from Rotronic and HMP 155 from Vaisala) mostly used for tropical weather analysis with reference weather station data with observations in the range of 11°C to 35°C in to the test configuration. The authors concluded that air temperature and humidity differences were smaller during the nighttime than daylight observation hours. Only three studies from the literature were identified as having done a comparative analysis of the sensors applied in this study. 1) Chapman et al., (2023): ClimaVUE50 – Netatmo outdoor module, 2) Dombrowski et al., (2021): WXT536 – ATMOS41 (similar to ClimaVUE50) and 3) Young et al., (2014): Thermochron iButton - HOBO-U23. Chapman et al. (2015) identified the high costs of maintaining and running such UMNs even with low-cost sensors. Caluwaerts et al. (2020) highlighted the difficulties of installing some sensors, like anemometers, at the height of 10m as recommended by WMO (2006), instead of placing them at 2m from ground level for practical reasons. Chan et al. (2021) emphasized the testing and calibration of sensors before their applications in the field. Muller et al. (2013) indicated the specific urban scale validation for selection of a number of sensors to be used for a dense network development. Keeping in mind the above-mentioned literature references about air temperature sensors applied for urban climate observations, it is evident that there is a lack of transparent criteria for air temperature sensor selection and for independently conducted sensor comparison as a sound basis for the selection of air temperature sensors for the urban climate. Here, we provide a method that could be applied for the selection of air temperature sensors based on defined criteria. This method will provide a common base for urban climate observation network sensor selection that could be applied across the globe.

To contribute to the understanding of the capabilities and selection of different AIO and LC sensors to be used particularly in the urban context, our research study provides a comparison of selected AIO and LC sensors regarding air temperature. The objectives of the study are to compare AIO and LC sensors for air temperature observations in a controlled environment (climate chamber) and in an open environment (in-situ) in the context of an urban environment. This will help the scientific community save time and resources when selecting air temperature sensors particularly in the urban environment. This will also contribute to clarifying the ambiguity between AIO and LC sensors with regards to accuracy and cost-effectiveness.

Materials and methods

We conducted two main steps to answer the missing sensor testing and selection procedure development. The steps (Figure 1) follow the direction from 1) criteria-based sensor selection to 2) (a) climate chamber and (b) in-situ testing.

Figure 1.

Study design flow chart indicating the two main stages for AIO and LC sensors selection and testing steps.

Sensor selection

Available AIO and LC sensors were identified through scientific literature and manufacturer webpages available since 2022. For AIO, we identified a total of 21 types of sensors that have been reviewed in detail (Figure S1 in the Online Supplementary Data). For the LC sensors, we identified five that meet the defined criteria of sensor accuracy, power independence, commercial availability, and easy installation. We omitted Do-It-Yourself (DIY) sensors from the study for comparison of user-friendly and ready-to-use applications.

The key defined criteria were: 1) high accuracy of air temperature sensor (±1°C); 2) compactness in design; 3) low weight (max. 2 kg) for stable mounting on street light poles, low price range (for AIO max. 3000 Euros and for LC sensors max. 600 Euros); 4) lower energy consumption; and 5) reliable solar panels for power independence and remote data accessibility (via cellular and cloud data). We selected these criteria based on existing literature and our own study approach that focused on user friendly installation in the field. The main arguments were based on accuracy of air temperature sensors for qualitative observations, low weight for stability on street light poles, flexible mounting set up for quick fixing and replacement, ease of transportation in bulk and requiring less people in field for installation. Low power consumption is important for sensors to be independent of a direct power supply. Small built-in solar panels provide flexibility in installation-location selection and for long-term data collection with no need for battery replacements.

As a result, we included three AIO (METSENS 600 (Campbell Scientific inc., 2022c), ClimaVUE50 (Campbell Scientific inc., 2022a), WXT536 (Vaisala, 2022)) and five LC sensors (Netatmo (Netatmo, 2022), SenseBox Home (SenseBox, 2022), Thermochron iButton (i-Button, 2022), HOBO-U23 (Onset Computer Corporation, 2022), nMETOS180 (Pessl Instruments, 2022)) in this study for analysis (see Table 1 for the sensors that were not selected, and the reasons for rejection).

Table 1.

An overview of the rejected AIO and LC sensors along with their reasons for rejection based on the initial selection criteria used in this study.

Name of sensor	Reason for rejection
ATMOS41 Weather station	Identical to ClimaVUE50 sensor
WS50PV compact All-In-One weather station	Air temperature sensor accuracy > ±0.3
MetConnect One	High cost (> 3000 Euros), weight > 2 kg
Clima Sensor D	Air temperature sensor accuracy > ±0.3, High power consumption (24 VDC)
All-In-One weather station	Air temperature sensor accuracy > ±0.3
MWS-C400	Air temperature sensor accuracy > ±0.3
Minimet5 All IN ONE	High cost (> 3000 Euros), weight > 2 kg
ZWS 700	Air temperature sensor accuracy > ±0.3
150 WXRS Weather Station	High power consumption (40 VDC)
Response One	Air temperature sensor accuracy > ±0.3
u[sonic]WS7	Air temperature sensor accuracy > ±0.3
Standard DNB 301-1	High cost (> 3000 Euros)
WS700-UMB	High cost (> 3000 Euros)
IMETOS 3.3 IMT280-USW	Weight > 2 kg
µMETOS ET	Weight > 2 kg
MaxiMet GMX600	High cost (> 3000 Euros)

All air temperature sensors need radiation shields to minimize the impact of direct radiation that could lead to a temperature difference of up to 4°C between shielded and unshielded sensors (Pohorsky et al., 2024). AIO sensors (METSENS 600, WXT536) had air temperature sensors installed inside the compact body except for ClimaVUE50 which uses default calibration for its correct readings (Campbell Scientific inc., 2022a). LC sensors (HOBO-U23, SenseBox HOME and i-Button) were provided with LC radiation shields, while nMETOS180 (Pessl Instruments, 2022) and Netatmo (outdoor module) (Netatmo, 2022) have inbuilt radiation shields from their respective manufacturers.

Testing phase

In total eight (three AIO and five LC) sensors were selected (Table 2) for the testing phase with the inclusion of HygroVUE10 sensor for reference with passive and active ventilation during in-situ testing. In CC testing, CC air temperature (±0.1 K to ±0.5 K) were taken as the reference value. SenseBox could not be operated due to technical failure in power supply, data observation and communication errors resulting in exclusion from the testing.

Table 2.

Overview of selected AIO and LC sensors provided by the manufacturers used in this study. The specifications are accurate as of December 2022. All sensor photos are from the In-situ test-bed site.

Climate chamber testing

A climate chamber (CC) also known as an environment chamber is a designed platform for simulation and testing with controlled environmental parameters ranging from small single sensor size to a walk-in size chamber (Dong et al., 2018; García-Contreras et al., 2018; Haka, 2020). In this study, the climate chamber WK3 Klima3000 (WEISS Technik North America Inc., 2022), having a temperature accuracy (±0.1 K to ±0.5 K), temperature range (-45°C to +180°C), change rate of 4K/min for heating and 3K/min for cooling, was used for sensor testing. The CC testing was held from 4 November 2024 to 8 November 2024. The sensors were not calibrated independently before the testing and sensor calibration report from manufactures were used in this study.

The sensors were all placed together on one tray (Figure 2) having an area of 9000 cm² (0.9m²) in the CC to check the response of sensors with three temperature amplitudes (Table 3). The air temperature sensors could not be removed from the radiation shields particularly for the AIO sensors. For this reason, we decided to keep the radiation shields for all sensors, even though the air exchange is strongly reduced in the CC setting.

Figure 2.

Climate chamber testing set-up of sensors. The observations were collected through a CR350 data logger provided with direct electric power cable (black) on the right bottom and communication USB cable (white) using sealed pathways on the left bottom of the climate chamber respectively. LC sensors are labelled in yellow and AIO in blue.

Table 3.

Defined temperature amplitude for the testing circle in the climate chamber with observation time interval for AIO, LC sensors and reference sensor.

Steps	Defined Testing	Allocated Time (minutes)	Temperature range (°C)	Observation time interval of referenced sensor (minutes)
1	Daily cycle	480	12 to 28	1
2	Extreme ranges	130	-20 to + 40	1
3	Stable temperature	60	15	1

In-situ testing

In the in-situ testing process, the sensors (METSENS 600, ClimaVUE50, WXT536, nMETOS180, HOBO-U23, Netatmo, SenseBox HOME and i-Button) were installed on a test bed (Figure 3) designed for an easy and portable setting up for open environment observations as recommended by Chapman et al. (2015). The test period was held from 17 April 2024 to 17 May 2024. The test bed was installed on a roof top of a building at Holländischer Platz campus of Kassel University, coordinate location: 51.325372°N and 9.504536°E. It had a dimension of 3m in length and 1.3m in height. The sensors were installed on the test bed at a height of approximately 2m from the surface of the rooftop which itself was 15m above ground level. The AIO sensors were aligned to the North with the help of a magnetic compass. All the sensors were positioned in a way that they had similar height position from the ground (WMO, 2014) and were covered with radiation shields for reducing the direct impact of solar radiation (Burt and Podesta, 2020; Huwald et al., 2009). The METSENS 600, ClimaVUE50, WXT536 and nMETOS180 sensors’ power supply came from solar panels with back up batteries. HOBO-U23, Netatmo (outdoor module), SenseBox and i-Button had their own internal batteries for energy supply except for Netatmo (indoor module) that required a constant direct power supply. The air temperature variable was taken from HygroVUE10 sensor while wind speed data was taken from MetSense600 for analysis. For in-situ testing, we selected HygroVUE10 (SHT35 air temperature and humidity sensor) with a manufactured defined accuracy (±0.2°C) (Campbell Scientific inc., 2022b; Fugazza et al., 2023; Heiser et al., 2024) reference sensor. The reference sensor was first used with active ventilation through a TS-200 Apogee’s 24volt fan-aspirated radiation shield (Apogee Instruments, Inc., 2022) and then with a natural ventilated radiation shield for the comparison.

Figure 3.

In-situ set up for the outdoor testing with AIO (blue) and LC (yellow) sensors, mounted on a testing bed with data logger boxes supported by solar panels for power supply except for fan-aspirated radiation shield that was given direct power supply.

Data analysis

The sensors were compared for their outputs for air temperature observations in both the CC and in-situ testing scenarios. The data collection was programmed for each sensor’s highest available temporal resolution. Data was collected in 1-minute to 15-minute intervals to represent the sensor response to fluctuating temperature change in the surrounding environment (Burt and Podesta, 2020; Chan et al., 2021; WMO, 2014). AIO sensors were programmed with 1-minute intervals. The LC sensors (Netatmo and HOBO-U23) were programmed with 5-minute intervals while nMETOS180 was 15 minutes. i-Button was programmed with 1-minute intervals for CC testing. Due to minimum memory storage capacity (512 bytes) and only 2048 readings with 1-minute intervals (i-Button, 2022), it was designed with 20-minute intervals with data collection after every 34 hours to cover the daily changes during in-situ testing data collection period.

As LC sensors had different temporal resolution and the observation data were found to have data gaps ranging from 4 to 19 minutes throughout the testing stages (Table 4). These data gaps were then filled with python program function (fill.na()) (Puertas, 2024) where the forward filling method was used that takes the last value in the data series and fills it in the forward gaps till the next available value. This could create some biases for data sets but it did not affect the temporal intervals (10 minutes in CC testing and hourly data in-situ testing) in this study. But such datasets need to be investigated in future research works if handling this method with big data gaps and short temporal intervals. The resulting datasets for each sensor were then statistically compared with the reference sensor based on Absolute Mean Deviation (MAD), Root mean square deviation (RMSD), regression analysis and coefficient of determination (R²) following other studies (DelSole, 2022; Mudelsee, 2019).

Table 4.

An overview of the LC sensors data collection intervals during the testing stages and corresponding data gaps that were filled using forward gap-filling method during statistical analysis in this study.

LC Sensors	Data collection intervals (minutes)		Data gaps (minutes)
	Climate Chamber testing	In-situ testing
HOBO-U23	5	5	4
Thermochron iButton	1	20	19
Netatmo	5	5	4
nMETOS180	10	15	14

Results and discussion

Climate chamber testing

Daily cycle test

The daily cycle testing (Figure 4a) analyzed with hourly air temperature values indicated that sensors followed the daily temperature fluctuation resembling a normal summer day with variability based on their accuracy, response time and insulation effects. The increasing temperature change from 12°C was observed by sensors in similar pattern except for i-Button and Netatmo, which showed a deviation of 4 K and 5 K respectively. Netatmo showed the maximum deviation of 3-4 K during both the decreasing and increasing temperature values than the rest of the sensors (Figure 4b), which is related to the slow thermal response of the Netatmo sensor and effect of radiation shield (Harrison, 2010; Meier et al., 2017).

Figure 4.

Hourly air temperature observations during daily cycle (12°C to 28°C) testing in the climate chamber (a) and mean air temperature deviations of sensors from reference sensor (b), observed during the 8-hour time period.

Extreme range test

During the extreme range testing with 10-minute air temperature observations (Figure 5), it was observed that all the sensors were responding with 5 to 15 K deviation from the referenced CC values. The sensors (iButton and Netatmo) showed higher deviation from the reference sensors while nMETOS180 showed a minimal deviation from the referenced values. AIO sensors (ClimaVUE50 and WXT536) had a consistency in their observations indicating their slow response to sudden temperature changes. Similar results were found by Dombrowski et al. (2021) indicating that AIO sensors’ response to extreme temperature fluctuations during the nighttime is lower compared to the reference weather station. Chodorek et al. (2022) identified that LC sensors tend to respond quickly to sudden temperature changes due to their response time characteristics based on manufactures’ accuracy and material quality. Our results also indicated that sensors tend to respond quickly to decreasing temperature due to internal condensation as identified by Aranda and Merlone (2024) during their on-site calibration of air temperature sensors. Increasing temperature is observed slowly because of slow thermal response of sensor materials which was concluded by Aranda and Merlone (2024) as well. The temperature range of -20°C to +40°C has larger impact on the uncertainties of conventional sensors such as thermocouples due to increased sensitivity levels with extreme temperature ranges (Foken, 2021; Oukaira et al., 2024). Wang et al. (2010) related the temperature variability observed differently by each sensor as due to the thermal inertia property of its components and defined thermal inertia as the observed delay in a body’s temperature response during heat transfers caused by factors like thermal conductivity, density and specific heat of the material body. Verbeke and Audenaert (2018) applied the principle of thermal inertia on buildings and its process of heat exchange relative to its environment.

Figure 5.

Extreme range test in climate chamber using 10-minute mean air temperature (−20°C to +40°C) testing results for all sensors including the reference climate chamber observed during the 130-minute time period.

Stable temperature test

The 15°C stable testing with 10-minute mean air temperature observation data (Figure 6) indicated that all sensors except for Netatmo took approximately 15-20 minutes to reach the referenced temperature value (15°C) of the CC. Chodorek et al. (2022) showed that each sensor tended to take a particular time to reach to specific level of temperature based on their sensor thermal response and accuracy scales.

Figure 6.

Stable temperature testing in climate chamber using 10-minute mean air temperature for all sensors including the reference climate chamber values observed during the 60-minute time period. It also demonstrates the response time of sensors to reach a defined temperature value from starting point.

Netatmo showed the longest duration before reaching the 15°C stable level with more than 50 minutes during the 60 minutes testing time period. This deviation is related to the compact design of Netatmo sensor cases without any ventilation and insulation to the surrounding environment. Meier et al. (2017) also found that the manufactured design of Netatmo sensor causes error in the air temperature observations for outdoor modules that are exposed to direct radiation, which needs to be corrected through quality control (QC) methods. He suggested that two hours are required for the sensor to reach a constant level. Büchau (2018) tested the outdoor module of Netatmo air temperature sensor (SHT20) with and without the cylindrical aluminum shell and found that the time constant for Netatmo sensor increase by 5 minutes when observed with the cylindrical aluminum shell. He suggested a 1-hour time period to reach a constant level for Netatmo sensors. We have included the effect of radiation shields on the observation values in this study.

Summary of the CC test

Comparing the results from statistical analysis (Table 5) for all the datasets with 10-minute mean air temperature values, HOBO-U23 had the lowest MAD in daily cycle test and nMETOS180 in extreme range test and in stable temperature test among the group of sensors respectively. The extreme range test results indicated that Netatmo had the highest MAD, MSD and RMSD and lowest coefficient of determination (R²) values among the group. On the other hand, nMETOS180 had the lowest MAD, RMSD, MSD and higher R² for air temperature observations. Even though it is evident that HOBO-U23 had the lowest MAD, MSD and RMSD in daily cycle testing, and second lowest MAD, MSD and RMSD during extreme range and stable temperature testing, it could not be considered for longer time period of air temperature observations collections due to manual data collection method (Onset Computer Corporation, 2022).

Table 5.

An overview of 10-minute mean air temperature observations comparative analysis for climate chamber testing indicating the statistical values of mean absolute deviation (MAD), mean square deviation (MSD), root mean square deviation (RMSD) and coefficient of determination (R²) for daily cycle, extreme range and stable testing periods for the sensors. All the sensors were analyzed with CC temperature values as reference. The highest values (dark blue) and lowest values (light blue) are indicated with R² values above 0.5 (green).

From the results of the extreme range test, it is recommended that nMETOS180 sensor could be applied for air temperature observations in extreme environments. The stable temperature test showed that AIO sensors (ClimaVUE50 and WXT536) with high R² values, tend to have low MSD values while LC sensors (nMETOS180, HOBO-U23 and iButton) have low MSD and low R² values indicating their quick response to the air temperature fluctuations in short periods of observations. During the daily cycle testing, HOBO-U23 and WXT536 sensors showed lowest MAD, MSD, RMSD and highest R² values among the LC and AIO sensors respectively. Kowal et al. (2020) applied different environmental conditions on two LC temperature sensors (HMP 155 and Vaisala’s HMP 45AC) and concluded that time, high humidity, high and low temperatures and rapid temperature changes have negative effect on the repeated testing results. Dombrowski et al. (2021) compared to AIO compact sensors ATMOS41 (similar to ClimaVUE50) and Vaisala WXT520 with reference weather station data. The authors identified that during nighttime and rainfall, the ATMOS41 showed lower daily maximum values due to its exposed air temperature sensor being affected by cooling as compared to shielded air temperature sensors. During the stable temperature testing, nMETOS180 had the lowest MAD (0.27) while Netatmo had the highest MAD (1.53) indicating the contrasting performance of LC sensors in the given experimental settings without ventilation to reach the stable temperature range. Burt and Podesta (2020) found through their sensor selection method that the response time of LC sensors should be critically observed along with differential experiment from a referenced weather station. Chan et al. (2021) concluded that cost effective LC electronic sensors, such as Arduino sensors and Internet-Of-Things (IoT), with mean bias errors below 20% could be used for environmental research instead of research grade sensors.

One important aspect for the CC observation analysis is that there was no ventilation provided to the sensors in CC testing and the sensor response variabilities were based on thermal insulation properties of each sensor based on type of radiation shield used as identified by Huwald et al. (2009) as well. Yang et al. (2021) concluded through the radiative error experiment that radiation shields heat up by direct or indirect radiation on the external surface. Nakamura and Mahrt (2005) identified air temperature measurement errors to be caused by insufficient wind flow through the sensors and internal radiative heat. This supports our argument that CC testing air temperature results cannot be directly compared to outdoor air temperature observations and need to go through detailed quality control parameters as recommended by Fenner et al. (2021).

In-situ testing

During the in-situ testing in an open environment, the sensors were exposed to ventilation and solar radiation as compared to the controlled CC testing. The time series representation of air temperature observations (Figure 7) of all sensors during the experimental time period showed daily fluctuations and deviations to the reference sensor values. We observed that LC sensors (HOBO-U23 and iButton) had extreme deviations during day and night air temperature observations compared to the other sensors.

Figure 7.

In-situ testing using hourly air temperature observation time series representation of all sensors’ analysis with the reference sensor (Fan aspirated HygroVUE10) during the 30-day time period. The highlighted red box represents the hottest day (1 May 2024) selected for day and night deviation analysis.

Day and night analysis

In order to understand the variabilities in air temperature deviations of sensors, we focused on the deviations occurring during day and nighttime observations. We classified the observed hourly data into day and night on the basis of observation time with data observed during 6:00 to 18:00 as daytime and from 19:00 to 5:00 as nighttime data, respectively. Based on this differentiation between day and nighttime data sets, we performed air temperature deviation analysis of all the sensors with respect to the referenced sensor. The air temperature deviation results for the hottest day (1 May 2024) during the 30-day time period was selected for day and nighttime deviation analysis. The results (Figure 8) showed that LC sensors (iButton and HOBO-U23) had the highest deviations from the reference sensor during daytime. One of the causes is that both the sensors (iButton and HOBO-U23) were inside a single radiation shield (Figure 3) that was exposed to solar radiation that could have an effect of the internal heating of the sensors but as indicated previously, we did not investigate the types of radiation shields and their qualities in our study.

Figure 8.

One day hourly air temperature deviations time series of sensors (shown with different colors) from the reference sensor (Fan-aspirated HygroVUE10) indicated as a black dotted line, observed during the selected hottest day (1 May 2024) in the in-situ testing. The daytime was classified as 06:00-18:00 and nighttime as 19:00-05:00 respectively.

During the nighttime observations, AIO sensor (WXT536) showed high deviation at the start of the night but LC sensor (iButton) showed highest deviations from the reference sensor during the later hours of nighttime observations. The AIO sensors (ClimaVUE50 and WXT536) indicated lower deviations from the reference sensor but their high cost and data management system makes them difficult to select for dense observation network projects, as identified by Chapman et al. (2023) and Ramírez-Moreno et al. (2021).

Dombrowski et al. (2021) observed that the exposed air temperature sensors of AIO compact sensors (such as ATMOS41) are affected by cooling during night and rainy wet days as compared to the radiation shielded reference station. All the sensors with different types of radiation shield observed the radiation and ventilation effect differently depending on the quality and type of radiation shield used. We have not analyzed the quality and material assessment of radiation shields for sensors in this study and these should be considered for comparative analysis in future. Huwald et al. (2009) identified that surface-reflected shortwave radiation (albedo effect) is a principal source of error in sensors even in the presence of ventilation and radiation shields. Harrison (2010) argued that ventilation errors at night with low wind speeds are often ignored for daily minimum temperature calibrations and affect the daily average air temperature results.

The statistical results of day and night air temperature observation analysis (Table 6) showed that AIO sensors have low MAD, MSD and RMSD values during daytime while LC sensors have high MAD, MSD and RMSD values. During nighttime observation, all sensors had low MAD, MSD, RMSD and high R² except for Netatmo which showed high MAD (5.76 K) and low R² (0.90). We related this exceptional behavior of Netatmo sensor to the slow thermal response of its outdoor module especially during nighttime observation. The variabilities in air temperature deviations, especially during daytime observations, were also related to the type of radiation shield used by that particular sensor but no further analysis was conducted on this aspect in this study. Netatmo performance in the in-situ testing made it a lower accuracy sensor for temperature observations but it has been used in many citizen science research studies because of its cost effectiveness, cloud based database and user-friendly interface (Brousse et al., 2022; Fenner et al., 2017, 2021; Meier et al., 2017; Ramírez-Moreno et al., 2021). The data from such networks requires quality control before applying for analysis as recommended by Fenner et al. (2021) and Ioannou et al. (2021). Büchau (2018) investigated the air temperature radiation deviations in Netatmo outdoor modules observed in different environmental conditions and concluded that air temperature deviations were caused by a combined effect of sensor outer casing (aluminum shell including the plastic inlet) response time and radiation observations from the surrounding environment. They suggested including the radiation effect of surrounding buildings as well for air temperature observations. nMETOS180 being from the LC sensor group performed close to the AIO sensors values during day and nighttime observations.

Table 6.

Statistical analysis of hourly air temperature deviations (K) during day and nighttime for in-situ testing indicated by lowest (light blue) to highest values (dark blue). R² values are represented as above 0.5 (green).

Ventilation analysis

In order to understand the impact of passive and active ventilation on the air temperature observations, we tested all sensors with HygroVUE10 as a reference sensor, first with natural ventilated radiation shield and later with fan-aspirated ventilation shield.

The statistical analysis (Table 7) values showed that AIO and LC sensors have low MAD, MSE, RMSD and high R², except for Netamo which showed high MAD and R² during natural and fan-aspirated ventilation observation. Among the AIO sensors, ClimaVUE50 showed the lowest MAD, MSD, RMSD and highest R² values indicating its high air temperature sensor accuracy (Table 2) performance in both the ventilation environments. That is why, it is applied in projects like TAHMO (Trans-African-Hydro-Meteorological-Observatory) (van de Giesen et al., 2014). Among the LC sensors, nMETOS180 performed with lowest MAD, MSD, RMSD and highest R² values in both ventilation scenarios. It was observed that during warm days there were high deviations by LC sensors as compared to low temperature days. It showed that AIO sensor tend to deviate less while LC sensors are influnced by the air temperature fluctuations due to their low accuracy.

Table 7.

Statistical analysis of hourly air temperature observations of all sensors as compared to natural ventilated and fan-aspirated ventilated sensor as reference station during in-situ testing indicated by lowest (light blue) to highest values (dark blue). R² values are represented as above 0.5 (green).

Correlation analysis

We examined the influence of wind speed and humidity on air temperature deviations observations by creating correlations between air temperature deviations, wind speed and humidity variables observed in the 30-day time period data. The correlation coefficient (r) between wind speed and natural ventilated HygroVUE10 sensor air temperature data showed a correlation (r=0.69) and with fan-aspirated ventilated HygroVUE10 air temperature values showed a correlation (r=0.70) value. This strong positive correlation indicated that the increase in wind speed had a similar impact on natural and fan-aspirated ventilation air temperature observations for the observed time period in this study.

The correlation between wind speed and relative humidity during day (r= - 0.01) with significance (p=0.58) and nighttime (r= - 0.02) with significance (p=0.66) respectively indicated their weak correlation relationship. LC sensors’ air temperature deviations have positive correlation values with relative humidity and wind speed during day and night except for Netatmo which shows negative correlation during daytime while AIO sensors have high negative correlations during daytime and low negative correlations during nighttime observations (Figure 9).

Figure 9.

Correlation coefficient (r) values for sensors’ air temperature deviations with relative humidity and wind speed indicated by day (blue) and night (orange).

Summary of in-situ testing

The in-situ testing showed that AIO and LC sensors are influenced by surrounding environment factors such as solar radiation, albedo effect and ventilation in particular. The influence of day and nighttime observation variabilities are related to sensor accuracy and ventilation effect especially for low temperature observation. The time series and statistical results for AIO sensors indicated that during day and nighttime analysis and ventilation testing, ClimaVUE50 had the lowest MAD, MSD, RMSD and highest R² values. Among the LC sensors, nMETOS180 showed lowest MAD, MSD, RMSD and highest R² values while Netatmo had the highest MAD, MSD, RMSD and lowest R² values during natural and fan-aspiration ventilation analysis. The correlation analysis indicated that wind and relative humidity have positive correlation for LC sensors while they have negative correlation for AIO sensors. We observed the impact of direct solar and internal radiation of sensors on the air temperature observations but these were not investigated in this study.

Limitations and way forward

The climate chamber testing observations excluded other environmental factors like humidity, ventilation, winds, shadow effect etc. Urban specific factors like air pollution, building typology, albedo effect from urban surfaces and anthropogenic heat should be investigated in detail for a better comparative analysis. These factors need to be investigated in future studies assessing sensor selection process based on multiple meteorological variables applied in climate chamber testing. Environmental factors affecting thermal analysis needs to be defined during a climate chamber experiments (Dong et al., 2018). The selection based on one variable (e.g. air temperature sensor) results could lead to misinformation of that observation station, especially in AIO compact sensors as suggested by Dombrowski et al. (2021) and van de Giesen et al. (2014). Chodorek et al. (2022) focused on LC sensors equipped with more than one sensor (e.g. humidity, radiation, precipitation, winds speed and directions etc.) to conclude that each sensor parameter must be defined for the project’s goals. The application of ventilation type for radiation shields should be observed with caution and the influence of surrounding environment must be taken into consideration during observation data analysis.

Our study design could be applied in other regions of the world for urban climate studies. The key environmental variables like wind speed, relative humidity and ventilation type correlation analysis indicated that variability in these variables have very low impact on the overall sensor analysis results. Other factors like solar radiation, type of radiation shields and its materials need to be considered while applying the sensor selection process. Urban factors should be investigated based on the scope and scale of the urban climate observation design.

Conclusion

Air temperature sensor selection was processed through two stages: firstly, criteria based selection from existing literature and commercial availability and secondly, through testing methods including climate chamber and in-situ testing. From the first selection method, we identified 11 AIO sensors based on the criteria for assessing if sensors are suitable for urban observation networks at city lights and other available installation mediums independent of external power supply. The key defined criteria were: high accuracy of air temperature sensor (±1°C), compactness in design, low weight, lower energy consumption, reliable solar panels for power independence and remote data accessibility (via cellular and cloud data). For the LC sensors, we considered five sensors that meet the defined criteria of sensor accuracy, power independence, commercial availability, and easy installation. In the second stage, all the sensors were tested in climate chamber with defined methods for daily cycle, extreme range and stable temperature testing. During in-situ testing, all sensors were tested in an open environment analyzing the performances by classifying them in day and nighttime observations and with the application of natural and fan-aspirated ventilation for radiation shields. The air temperature observations were analyzed through statistical methods like statistical analysis using mean absolute deviation (MAD), root mean square deviation (RMSD) and coefficient of determination (R²) performed on the observations. The overall analysis of AIO compact sensors (ClimaVUE50 and WXT536) highlighted their consistent performance with air temperature sensor high accuracy during long period of air temperature observations but also showed their lack of quick response to sudden change in temperature fluctuations. AIO sensors have the drawback of compactness in their design with less ventilation and they adapt slowly to the temperature changes due to thermal inertia. LC sensor (nMETOS180, iButton, Netatmo and HOBO-U23) testing performances showed variability in their results in this study. Netatmo had the highest deviations during CC testing and performed with highest MAD, MSD, RMSD and lowest R² values during natural and fan-aspiration ventilation analysis during in-situ testing. All sensors showed high deviations during daytime and lower during nighttime due to radiation errors with various types of radiation shields. LC sensors have in general higher deviations than AIO sensors but some LC sensors in this study showed promising results. For example, HOBO-U23 and nMETOS180 performance during CC testing (daily cycle test, extreme range test and stable temperature test) and in-situ testing (day and night analysis, ventilation and correlation analysis) showed their good performances and they could be reliable LC sensors for urban air temperature observations but they have their limitations as well. Our study analysis showed that LC sensor HOBO-U23 performed well but, due to its short memory feature, it would only be suitable for short-term observation campaigns but not for long term data collection. We have suggested nMETOS180 based on its features like cloud-based data management system, low weight, easy installation process, continuous solar based power supply and additional meteorological variables (relative humidity and precipitation) as compared to HOBO-U23 sensor. Based on these grounds, we suggested nMETOS180 to be a potential sensor for urban climate observation network.

With the advancements and innovations in technology, more and more cities will be installing such sensors, and more non-professionals will be interacting with such projects. New and advanced sensor types, data loggers and data management systems will be developing with time and could be applicable for more sophisticated urban climate observation data collection and analysis. It is important to focus on the development of more comparative studies for low-cost meteorological weather stations so that reliable conclusions are drawn for sensor selection and applications. Overall, our study provides an approach for air temperature sensor selection method using controlled and open environment testing methods that could be applied in other regions of the world. The testing parameters and variables could be changed according to the geographical region, meteorological conditions and project objectives.

Supplemental Material

sj-docx-1-tee-10.1177_2754124X251363130 – Supplemental material for Comparison and assessment of air temperature observations using All-In-One and Low-Cost sensors in an urban environment

Supplemental material, sj-docx-1-tee-10.1177_2754124X251363130 for Comparison and assessment of air temperature observations using All-In-One and Low-Cost sensors in an urban environment by Shakir Ahmed and Britta Jänicke in Transactions in Earth, Environment, and Sustainability

Footnotes

Acknowledgements

We acknowledge the contribution of student assistants Till Wiesehoff, Cornelius Hottenrott and Annika Schönewald for their support in field and installation during the testing stages. We thank Rolf Gross and our colleagues from the Chair of Building Physics, University of Kassel for providing the climate chamber and helping during its testing process. Moreover, we thank Nisha Patel and Dr. Stenka Vulova for proofreading and providing valuable suggestions throughout the research work.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Shakir Ahmed

Britta Jänicke

Data availability

The meteorological data used in this study containing sensor observations will be available upon request from the corresponding author.

Supplemental material

Supplemental material for this article is available online.

Author biography

Shakir Ahmed is a research associate and Ph.D. candidate at the Chair of Environmental Meteorology at the University of Kassel. His research focuses on urban climate observation, data collection and analysis and works with the development of the Kassel Urban Climate Network (KUCON). He holds a M.Sc. from Istanbul University, Turkiye and a second master’s degree in Climate and Environmental Science from Friedrich-Alexander-University Erlangen, Germany.

References

Anet

Burger

Spirig

, et al. (2024) Improving urban temperature measurements and two applications. City and Environment Interactions 23: 100148.

Apogee Instruments, Inc. (2022) Aspirated Radiation Shield. Available at: https://www.apogeeinstruments.com/ts-200-ss-24-v-aspirated-radiation-shield-shield-only/ (accessed 20 November 2022).

Aranda

Merlone

(2024) On-site calibration procedure and uncertainty contributions on air temperature sensors. International Journal of Thermophysics 45: 12.

Bai

Wang

, et al. (2024) Quantifying threshold and scale response of urban air and surface temperature to surrounding landscapes under extreme heat. Building and Environment 247: 111029.

Bechtel

Hüser

Weickhmann

, et al. (2024) Data2Resilience: Data-driven Urban Climate Adaption - A Biometeorological Sensor Network for Dortmund, EGU 2, Germany.

Brousse

Simpson

Walker

, et al. (2022) Evidence of horizontal urban heat advection in London using six years of data from a citizen weather station network. Environmental Research Letters 17(4): 44041.

Büchau

(2018) Modelling shielded temperature sensors: An assessment of the Netatmo citizen weather stations. Master Thesis, University of Hamburg, Germany.

Burt

Podesta

(2020) Response times of meteorological air temperature sensors. Quarterly Journal of the Royal Meteorological Society 146(731): 2789–2800.

Caluwaerts

Hamdi

Top

, et al. (2020) The urban climate of Ghent, Belgium: A case study combining a high-accuracy monitoring network with numerical simulations. Urban Climate 31: 100565.

10.

Campbell Scientific inc. (2022a) ClimaVue 50: Compact digital weather sensor. Available at: https://www.campbellsci.com/climavue-50 (accessed 29 September 2022).

11.

Campbell Scientific inc. (2022b) HygroVue 10: Digital temperature and relative humidity sensor with M12 connector. Available at: https://www.campbellsci.com/hygrovue10 (accessed 12 September 2022).

12.

Campbell Scientific inc. (2022c) MetSens600: Compact weather sensor for temperature, RH, barometric pressure, wind, and rainfall indication with compass. Available at: https://www.campbellsci.com/metsens600 (accessed 9 October 2022).

13.

Chan

Schillereff

Baas

ACW

, et al. (2021) Low-cost electronic sensors for environmental research: Pitfalls and opportunities. Progress in Physical Geography: Earth and Environment 45(3): 305–338.

14.

Chapman

Bell

Randall

(2023) Can crowdsourcing increase the durability of an urban meteorological network? Urban Climate 49: 101542.

15.

Chapman

Muller

Young

, et al. (2015) The Birmingham urban climate laboratory: An open meteorological test bed and challenges of the smart city. Bulletin of the American Meteorological Society 96(9): 1545–1560.

16.

Chen

Huang

, et al. (2022) Cross-examining precipitation products by rain gauge, remote sensing, and WRF simulations over a South American region across the Pacific coast and Andes. Atmosphere 13(10): 1666.

17.

Chen

Y-C

Liao

Y-J

Yao

C-K

, et al. (2019) The application of a high-density street-level air temperature observation network (HiSAN): The relationship between air temperature, urban development, and geographic features. The Science of the Total Environment 685: 710–722.

18.

Chodorek

Sitek

(2022) Response time and intrinsic information quality as criteria for the selection of low-cost sensors for use in mobile weather stations. Electronics 11(15): 2448.

19.

DelSole (2022) Statistical Methods for Climate Scientists. Cambridge University Press.

20.

Dombrowski

Hendricks Franssen

H-J

Brogi

, et al. (2021) Performance of the ATMOS41 all-in-one weather station for weather monitoring. Sensors (Basel, Switzerland) 21(3): 741.

21.

Dong

Zhou

, et al. (2018) Climate chamber experiment-based thermal analysis and design improvement of traditional Huizhou masonry walls. Sustainability 10(3): 694.

22.

Dumas

Masson

Hidalgo

, et al. (2021) Co-construction of climate services based on a weather stations network: Application in Toulouse agglomeration local authority. Climate Services 24: 100274.

23.

Fenner

Bechtel

Demuzere

, et al. (2021) CrowdQC+—a quality-control for crowdsourced air-temperature observations enabling world-wide urban climate applications. Frontiers in Environmental Science 9: 720747.

24.

Fenner

Meier

Bechtel

, et al. (2017) Intra and inter ‘local climate zone’ variability of air temperature as observed by crowdsourced citizen weather stations in Berlin, Germany. Meteorologische Zeitschrift 26(5): 525–547.

25.

Foken

(2021) Springer Handbook of Atmospheric Measurements. Cham: Springer International Publishing.

26.

Freitag

Ning

Berk

, et al. (2016) ATLAS, T-Flex, BaiLong meteorological sensor comparison test report. USA: NOAA technical memorandum OAR PMEL, USA, 148.

27.

Fugazza

Valle

Caccianiga

, et al. (2023) Glaciological and meteorological investigations of an Alpine debris-covered glacier: The case study of Amola Glacier (Italy). Cold Regions Science and Technology 216: 104008.

28.

García-Contreras

Gómez

Fernández-Yáñez

, et al. (2018) Estimation of thermal loads in a climatic chamber for vehicle testing. Transportation Research Part D: Transport and Environment 65: 761–771.

29.

Gubler

Christen

Remund

, et al. (2021) Evaluation and application of a low-cost measurement network to study intra-urban temperature differences during summer 2018 in Bern, Switzerland. Urban Climate 37: 100817.

30.

Haka

(2020) Climate test chambers and exposure testing. In: Martin

Mody

CCM

(eds) Between Making and Knowing. World Scientific, 227–237.

31.

Harrison

(2010) Natural ventilation effects on temperatures within Stevenson screens. Quarterly Journal of the Royal Meteorological Society 136(646): 253–259.

32.

Hart

Martinez

(2006) Environmental Sensor Networks: A revolution in the earth system science? Earth-Science Reviews 78(3–4): 177–191.

33.

Heiser

McMahon

Smith

(2024) Meteorological services annual data report for 2023.

34.

Huwald

Higgins

Boldi

M-O

, et al. (2009) Albedo effect on radiative errors in air temperature measurements. Water Resources Research 45(8): W08431.

35.

iButton (2022) iButton DS1922L Thermochron. Available at: https://iButton.co.uk/ (accessed 22 October 2022).

36.

Ioannou

Karampatzakis

Amanatidis

, et al. (2021) Low-cost automatic weather stations in the Internet of things. Information 12(4): 146.

37.

Irga

Fleck

Wooster

, et al. (2021) Green Roof & Solar Array -Comparative Research Project.

38.

Karagulian

Barbiere

Kotsev

, et al. (2019) Review of the performance of low-cost sensors for air quality monitoring. Atmosphere 10(9): 506.

39.

Kowal

Merlone

Sawiński

(2020) Long-term stability of meteorological temperature sensors. Meteorological Applications 27(1): e1795.

40.

Lelovics

Unger

Gál

, et al. (2014) Design of an urban monitoring network based on Local Climate Zone mapping and temperature pattern modelling. Climate Research 60(1): 51–62.

41.

Lubbe

Maritz

Bosserez

, et al. (2022) A multi-perspective analysis of microclimate dynamics for air-based solar hydrogen production. Heliyon 8(7): e09883.

42.

Lussana

Baietti

Båserud

, et al. (2023) Exploratory analysis of citizen observations of hourly precipitation over Scandinavia. Advances in Science and Research 20: 35–48.

43.

Mao

Khamis

Krause

, et al. (2019) Low-cost environmental sensor networks: Recent advances and future directions. Frontiers in Earth Science 7: 221.

44.

Meier

Fenner

Grassmann

, et al. (2017) Crowdsourcing air temperature from citizen weather stations for urban climate research. Urban Climate 19: 170–191.

45.

Mudelsee

(2019) Trend analysis of climate time series: A review of methods. Earth-Science Reviews 190: 310–322.

46.

Muller

Chapman

Grimmond

CSB

, et al. (2013) Sensors and the city: A review of urban meteorological networks. International Journal of Climatology 33(7): 1585–1600.

47.

Nakamura

Mahrt

(2005) Air Temperature Measurement Errors in Naturally Ventilated Radiation Shields. Journal of Atmospheric and Oceanic Technology 22: 1046–1058.

48.

Nalakurthi

NVSR

Abimbola

Ahmed

, et al. (2024) Challenges and opportunities in calibrating low-cost environmental sensors. Sensors (Basel, Switzerland) 24(11): 3650.

49.

Netatmo (2022) Netatmo. Available at: https://www.netatmo.com/de-de/smart-weather-station (accessed 13 October 2022).

50.

Onset Computer Corporation (2022) HOBO U23 Pro v2: External temperature/relative humidity data logger. Available at: https://www.onsetcomp.com/products/data-loggers/u23-002a (accessed 16 October 2022).

51.

Oukaira

Ettahri

Lakhssassi

(2024) Advancing quantum temperature sensors for Ultra-Precise Measurements (UPMs): A comparative study. Electronics 13(18): 3715.

52.

Peerlings

Vranic

Ommer

, et al. (2024) Indoor heat in Amsterdam: Comparing observed indoor air temperatures from a professional network and from a citizen science approach. City and Environment Interactions 24: 100173.

53.

Pessl Instruments (2022) μMETOS. Available at: https://metos.global/de/micrometos-nbiot/ (accessed 14 October 2022).

54.

Pohorsky

Baccarini

Tolu

, et al. (2024) Modular Multiplatform Compatible Air Measurement System (MoMuCAMS): A new modular platform for boundary layer aerosol and trace gas vertical measurements in extreme environments. Atmospheric Measurement Techniques 17(2): 731–754.

55.

Puertas

(2024) Data science tools. In: Igual

Seguí

(eds) Introduction to Data Science: Cham: Springer International Publishing, 5–28.

56.

Quanz

Ulrich

Fenner

, et al. (2018) Micro-scale variability of air temperature within a local climate zone in Berlin, Germany, during Summer. Climate 6(1): 5.

57.

Ramírez-Moreno

Keshtkar

Padilla-Reyes

, et al. (2021) Sensors for sustainable smart cities: A review. Applied Sciences 11(17): 8198.

58.

Rashid

Rehmani

(2016) Applications of wireless sensor networks for urban areas: A survey. Journal of Network and Computer Applications 60: 192–219.

59.

Romero Rodríguez

Guerrero Delgado

Castro Medina

, et al. (2024) Forecasting urban temperatures through crowdsourced data from Citizen Weather Stations. Urban Climate 56: 102021.

60.

SenseBox (2022) senseBox:home. Available at: https://sensebox.de/en/products-home (accessed 8 October 2022).

61.

Ulpiani

Hart

Di Virgilio

, et al. (2022) A citizen centred urban network for weather and air quality in Australian schools. Scientific data 9(1): 129.

62.

Vaisala (2022) Weather Transmitter WXT530 Series WXT530: A compact all-in-one weather instrument. Available at: https://store.vaisala.com/en/products/WXT530 (accessed 25 September 2022).

63.

van de Giesen

Hut

Selker

(2014) The Trans-African Hydro-Meteorological Observatory (TAHMO). WIREs Water 1(4): 341–348.

64.

Vasic

Milosevic

Savic

, et al. (2022) Micrometeorological measurements and biometeorological survey in different urban settings of Novi Sad (Serbia). Glasnik Srpskog geografskog drustva 102(2): 45–66.

65.

Verbeke

Audenaert

(2018) Thermal inertia in buildings: A review of impacts across climate and building use. Renewable and Sustainable Energy Reviews 82: 2300–2318.

66.

Nichersu

Wendel

(2019) Modeling, monitoring, and validating green roof and green facade solutions with semantic city models using low cost sensors and open software infrastructures. Urban Science 3(2): 39.

67.

Wang

Bras

Sivandran

, et al. (2010) A simple method for the estimation of thermal inertia. Geophysical Research Letters 37(5).

68.

Warne

(2017) Guidelines on Economical Alternative AWS.

69.

Weiss Technik North America Inc. (2022) WT3/WK3: Specialty environmental test chamber solutions. Available at: https://weiss-na.com/product/wt3-wk3/ (accessed 20 December 2022).

70.

Whiteman

Hubbe

Shaw

(2000) Evaluation of an inexpensive temperature datalogger for meteorological applications. Journal of Atmospheric and Oceanic Technology 17(1): 77–81.

71.

WMO (2014) Guide to Meteorological Instruments and Methods of Observation. Geneva: World Meteorological Organization.

72.

WMO (2006) Instruments and Observing Methods Report No. 81: Initial Guidance to Obtain Representative Meteorological Observations at Urban Sites.

73.

WMO (2024) Guidelines on the WMO Recognition of Long-term Observing Stations. Geneva: World Meteorological Organization.

74.

Writzl

Wollmann

Costa

, et al. (2022) Outdoor human thermal comfort along bike paths in Balneário Camboriú/SC, Brazil. Atmosphere 13(12): 2092.

75.

Yang

Liu

Chen

, et al. (2021) Optimized design and experimental validation of a temperature sensor for surface air temperature observation. Sensors and Actuators A: Physical 323: 112646.

76.

Young

Chapman

Muller

, et al. (2014) A low-cost wireless temperature sensor: Evaluation for use in environmental monitoring applications. Journal of Atmospheric and Oceanic Technology 31(4): 938–944.

77.

Zhou

S-Q

Z-W

W-Y

, et al. (2025) Vertical canopy structure dominates cooling and thermal comfort of urban pocket parks during hot summer days. Landscape and Urban Planning 254: 105242.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB