sensors

Article
Local Weather Station Design and Development for
Cost-Effective Environmental Monitoring and Real-Time
Data Sharing
Antonio Rivera 1 , Pedro Ponce 1, * , Omar Mata 1 , Arturo Molina 1 and Alan Meier 2

1 Institute of Advanced Materials for Sustainable Manufacturing, Tecnologico de Monterrey,

Monterrey 14380, Mexico; [email protected] (A.R.); [email protected] (O.M.); [email protected] (A.M.)
2 Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA; [email protected]
* Correspondence: [email protected]

Abstract: Current weather monitoring systems often remain out of reach for small-scale users and
local communities due to their high costs and complexity. This paper addresses this significant
issue by introducing a cost-effective, easy-to-use local weather station. Utilizing low-cost sensors,
this weather station is a pivotal tool in making environmental monitoring more accessible and
user-friendly, particularly for those with limited resources. It offers efficient in-site measurements of
various environmental parameters, such as temperature, relative humidity, atmospheric pressure, car-
bon dioxide concentration, and particulate matter, including PM 1, PM 2.5, and PM 10. The findings
demonstrate the station’s capability to monitor these variables remotely and provide forecasts with a
high degree of accuracy, displaying an error margin of just 0.67%. Furthermore, the station’s use of
the Autoregressive Integrated Moving Average (ARIMA) model enables short-term, reliable forecasts
crucial for applications in agriculture, transportation, and air quality monitoring. Furthermore, the
weather station’s open-source nature significantly enhances environmental monitoring accessibility
for smaller users and encourages broader public data sharing. With this approach, crucial in address-
ing climate change challenges, the station empowers communities to make informed decisions based
on real-time data. In designing and developing this low-cost, efficient monitoring system, this work
Citation: Rivera, A.; Ponce, P.; Mata,
provides a valuable blueprint for future advancements in environmental technologies, emphasizing
O.; Molina, A.; Meier, A. Local
sustainability. The proposed automatic weather station not only offers an economical solution for
Weather Station Design and
environmental monitoring but also features a user-friendly interface for seamless data communication
Development for Cost-Effective
between the sensor platform and end users. This system ensures the transmission of data through
Environmental Monitoring and
Real-Time Data Sharing. Sensors 2023,
various web-based platforms, catering to users with diverse technical backgrounds. Furthermore, by
23, 9060. https://doi.org/10.3390/ leveraging historical data through the ARIMA model, the station enhances its utility in providing
s23229060 short-term forecasts and supporting critical decision-making processes across different sectors.

Academic Editors: Lorena Parra and

Keywords: weather analytics; sustainable monitoring; atmospheric data; predictive modeling; data
Maria-Teresa Sebastia-Frasquet
collaboration; sensor networks
Received: 13 September 2023
Revised: 26 October 2023
Accepted: 3 November 2023
Published: 9 November 2023 1. Introduction
Monitoring local environments is a crucial undertaking that enables a better under-
standing of climate change’s challenges and how to respond to them. Environmental
Copyright: © 2023 by the authors.
changes usually refer to alterations in the natural surroundings of the Earth that can occur
Licensee MDPI, Basel, Switzerland. due to various factors such as human activities, climate change, natural disasters, and other
This article is an open access article natural processes. These changes can significantly impact the environment and affect many
distributed under the terms and areas, including ecosystems, biodiversity, water resources, and human health [1]. One of
conditions of the Creative Commons the main areas affected by environmental changes is ecosystems. Ecosystems are complex
Attribution (CC BY) license (https:// systems of living organisms, their physical environment, and their interactions [2].
creativecommons.org/licenses/by/ Environmental changes can disrupt the delicate balance of these systems and cause
4.0/). significant harm to plant and animal species and entire ecosystems. For example, defor-

Sensors 2023, 23, 9060. https://doi.org/10.3390/s23229060 https://www.mdpi.com/journal/sensors

Sensors 2023, 23, 9060 2 of 26

estation can lead to habitat loss and threaten the survival of many species of plants and
animals. At the same time, ocean acidification can harm marine ecosystems and disrupt
the food chain [3]. Environmental changes also affect biodiversity; biodiversity refers to
the variety of living organisms on Earth and the ecosystems they inhabit. Environmental
changes such as habitat destruction, climate change, and pollution can lead to biodiversity
loss, as species cannot adapt to the changing conditions [4]. This biodiversity loss can have
far-reaching consequences, including the loss of ecosystem services such as pollination and
water purification.
Water resources are also affected by environmental changes. Changes in temperature,
rainfall patterns, and land use can affect the availability and quality of water resources. For
example, climate change can lead to changes in rainfall patterns, which can affect water
availability and lead to droughts and floods. Pollution from human activities can also
contaminate water sources and affect the health of aquatic ecosystems. In addition, envi-
ronmental changes can also have significant impacts on human health. Air pollution, for
example, can cause respiratory problems, while exposure to chemicals and other pollutants
can lead to long-term health problems such as cancer and reproductive issues [5]. Tempera-
ture and precipitation patterns can also affect the spread of diseases such as malaria and
dengue fever [6,7]. As a result, environmental changes can significantly impact many areas,
including ecosystems, biodiversity, water resources, and human health.
Addressing these changes and working towards sustainable solutions is essential to
ensure a healthy and livable planet. Unfortunately, existing weather monitoring solutions
are constrained by the high costs associated with sensors, digital systems, communication
devices, power sources, trained personnel for maintenance and operation, limited coverage,
and insufficient data sharing with the public [8]. These limitations hinder small users
such as farmers, transportation companies, and local government agencies from accessing
reliable and affordable weather data. Weather monitoring technology continually evolves
to overcome these limitations and provide better user solutions.
The urgency for cost-effective weather monitoring is evident in today’s climate land-
scape. Such monitoring is pivotal, especially when considering the challenges of rapid
environmental changes and the need for timely responses. The National Oceanic and
Atmospheric Administration (NOAA) has been at the forefront of this, developing sur-
face moorings equipped with cost-effective sensors to monitor air-sea interactions, carbon
dioxide uptake, and upper ocean parameters, emphasizing the significance of real-time
data sharing [9]. One significant trend in weather monitoring technology is wireless sensor
networks (WSNs), a solution that is both affordable and efficient [10]. WSNs consist of
small, low-power sensors distributed over a large area and communicate with a central base
station to transmit data [11]. This technology collects high-resolution data in real time and
reduces traditional weather monitoring solutions’ cost and maintenance requirements [12].
The evolution of data sharing, facilitated by advancements in both public and pri-
vate online platforms, ensures that users are equipped with real-time insights, fostering
informed decision-making across various sectors [13]. As a result, weather data can be used
to develop applications that provide valuable insights and decision-making support for
various industries. State-of-the-art weather monitoring technology continually improves
with WSNs, IoT, AI, and data-sharing advancements [14]. These technologies are making it
more accessible and affordable for small users to access reliable and accurate weather data,
which is crucial for responding to the challenges of climate change.
Fixed stations are ground-based stations designed to receive and transmit signals
to and from satellites. These stations typically have satellite communication systems,
such as satellite television and internet services. Fixed stations are stationary and do not
move, making them ideal for establishing a stable connection with satellites and long-term
studies [15].
On the other hand, satellite images are pictures of the Earth taken from space using
satellites. These images can be used for various purposes, such as weather forecasting,
Sensors 2023, 23, 9060 3 of 26

urban planning, and natural resource management. Satellite images are captured using
satellite cameras orbiting the Earth [8].
To comprehensively address the complexities of environmental monitoring, it is essen-
tial to understand the diverse range of technological systems and methodologies employed
in this field:
• Fixed Stations: Ground-based stations that receive and transmit signals to and from
satellites.
• Satellite Images: Pictures of the Earth taken from space using satellite cameras.
• Mobile Stations: Stations that can move and are used in mobile satellite communica-
tion systems, such as satellite phones and GPS devices.
• Remote Sensing: The process of collecting data about the Earth from a distance, typi-
cally using satellites or aircraft. Remote sensing can be used for various applications,
such as environmental monitoring and land use mapping.
• Earth Observation: Using satellites and other platforms to monitor and study the
Earth’s surface, atmosphere, and oceans. Earth observation can be used for various
purposes, such as weather forecasting, disaster response, and agriculture management.
Local weather stations provide accurate, real-time data for various applications, from
agriculture to urban planning. However, with the rapid changes in environmental con-
ditions and the increasing complexity of weather patterns, there’s a growing need for
enhanced measurements and innovative monitoring techniques. Recent advancements,
such as integrating machine learning with city buses, have shown promising results in
real-time weather monitoring and prediction, emphasizing the importance of localized
data collection [16]. Moreover, incorporating wireless sensor networks in agricultural
management processes highlights the significance of tailored, on-field measurements for
optimizing production and ensuring eco-sustainability [17].
This paper presents an innovative solution to address the abovementioned limitations
by introducing a local open-source weather station (OSWS), a localized and affordable
sensing platform. The OSWS is designed to monitor a range of environmental variables,
including temperature, relative humidity, atmospheric pressure, CO2 concentration, and
particulate matter (PM 1, PM 2.5, and PM 10), offering rapid and cost-effective on-site
measurements. Moreover, the station possesses these variables and can forecast their values
by leveraging historical data obtained through the employment of the Autoregressive
Integrated Moving Average (ARIMA) model. The real-time monitoring capabilities of
the OSWS play a crucial role in providing accurate data that aids decision-making pro-
cesses in diverse domains such as agriculture, transportation, and air quality monitoring
applications [18].
The primary objectives of this research revolve around developing and implementing
an OSWS that offers a cost-effective solution for environmental monitoring, especially
catering to small users. This paper aims to harness the open-source nature of the OSWS
to enhance public data sharing and ensure seamless communication between the sensing
platform and end users, ensuring that valuable data is readily available for making informed
decisions and addressing environmental challenges stemming from climate change.
Furthermore, the design, development, and evaluation of this weather station serve as
a model for future low-cost and efficient monitoring systems. Its affordability, accessibility,
and real-time monitoring capabilities can significantly contribute to advancing sustainabil-
ity efforts and mitigating the adverse effects of climate change. The WS’s interface with
end users is crucial in facilitating effective data communication and enabling informed
decision-making.
Overall, the affordability, accessibility, and real-time monitoring capabilities of the
OSWS present a cost-effective solution for environmental monitoring and data sharing. The
comprehensive approach in designing, developing, and evaluating this device establishes
valuable information for developing similar low-cost and efficient monitoring systems,
ultimately promoting sustainability and mitigating climate change impacts.
 with end users is crucial in facilitating effective data communication and enabling in-
formed decision-making.
Overall, the affordability, accessibility, and real-time monitoring capabilities of the
OSWS present a cost-effective solution for environmental monitoring and data sharing.
Sensors 2023, 23, 9060 The comprehensive approach in designing, developing, and evaluating this device estab- 4 of 26
lishes valuable information for developing similar low-cost and efficient monitoring sys-
tems, ultimately promoting sustainability and mitigating climate change impacts.
To offer a clear outline of the paper, the structure is as follows: Beginning with the
To offer a clear outline of the paper, the structure is as follows: Beginning with the
Introduction in Section 1, the background and significance of the study are explored. Sec-
Introduction in Section 1, the background and significance of the study are explored.
tion 2 presents a thorough literature review, highlighting essential research and relevant
Section 2 presents a thorough literature review, highlighting essential research and relevant
developments.
developments. In Section 3, materials
In Section and and
3, materials methods are detailed
methods and subdivided
are detailed into Sec-
and subdivided into
tion 3.1, focusing on the design and development of the weather station, and
Section 3.1, focusing on the design and development of the weather station, and Section Section 3.2,
3.2,
discussing
discussing data validation
data validationmethods.
methods.Section 4 presents
Section the results,
4 presents further
the results, segmented
further into
segmented
Section 4.1, showcasing
into Section the design
4.1, showcasing theresults;
designSection
results;4.2, assessing
Section 4.2, the ergonomics;
assessing and Sec-
the ergonomics;
tion 4.3, comparing the data. Section 5 delves into the discussion, analyzing
and Section 4.3, comparing the data. Section 5 delves into the discussion, analyzing and interpret-
ing ourinterpreting
and findings. The
ourpaper culminates
findings. withculminates
The paper the conclusions in Section
with the 6, summarizing
conclusions in Section 6,
our research outcomes and their implications.
summarizing our research outcomes and their implications.

2. 2.
Literature Review
Literature Review
A Asystematic
systematicreview
reviewwaswasconducted
conductedtotoensure
ensurea acomprehensive
comprehensiveunderstanding
understandingofof
low-cost local weather prediction. The search
low-cost local weather prediction. The search strategy targeted reputable scientificjournals,
targeted reputable scientific jour-
nals, querying
querying various
various databases
databases focusing
focusing on articles
on articles emphasizing
emphasizing innovative
innovative weatherweather
monitor-
monitoring techniques
ing techniques or devices.
or devices. The selection
The selection criteriacriteria were refined
were refined to prioritize
to prioritize peer-
peer-reviewed
journal articles
reviewed journal that discussed
articles the challenges
that discussed and potential
the challenges of cost-effective
and potential and localized
of cost-effective and
weatherweather
localized prediction methods.
prediction After identifying
methods. the relevant
After identifying journal
the relevant papers,
journal a meticulous
papers, a me-
data analysis
ticulous method
data analysis was employed.
method was employed.This This
involved extracting
involved andand
extracting synthesizing
synthesizing data
related to the efficacy, cost implications, and materials used (Table 1). The
data related to the efficacy, cost implications, and materials used (Table 1). The synthe-synthesized data
provided
sized data insights
providedinto the advancements
insights and challenges
into the advancements andinchallenges
low-cost local weather predic-
in low-cost local
tion. Figure
weather 1 depicts
prediction. a cost
Figure comparison
1 depicts a costbetween commercial
comparison between weather stations,
commercial prototypes,
weather sta-
and prototypes,
tions, the OSWS presented in thispresented
and the OSWS paper. in this paper.

Figure 1. Cost comparison between various weather stations [19–27].

Figure 1. Cost comparison between various weather stations [19–27].
Sensors 2023, 23, 9060 5 of 26

Table 1. State-of-the-art based on related projects.

Ref. Variables IN/OUT DY ST WT MIT PS Aut. B Op-S IW Design OS GPS Forecast VR MCU C

Yes 300 mA Normal

Temperature and Both:
Solar panel Nearest- mode
RH: Ip65 Box RPI SQlite Configurable
with battery neighbors 150 mA without
[22] SHT75 Solar radiation No and send to Xbee RF between 24 h No No Compact Raspbian OS No No RPi
10 Ah and RBF ethernet, USB,
Solar radiation: shield ThingSpeak 30 s to 5 min
AC ANN etc.
SP110 2 m mast
4 h horizon 60 mA Xbee

NO3 , O3 :
NO2A43F/OXA431
Yes
PM 10, PM 2.5: Internal
Against
SDS011 Both: rechargeable
[28] No SD Card Bluetooth 25 s fixed 8h No No Compact No Yes No professional Arduino Not specified
Temperature, Ip 67 battery
WS ARPA
humidity, AC
Lazio
atmospheric
pressure

PM 2.5,
Alkaline 9 V
[19] temperature, No No SD Card No Not specified Not specified No No Compact No Yes No No Arduino Not specified
battery
humidity

NO2 , SO2 , CO2 , No local

Yes
CO, PM 2.5, storage. Only
[29] Both No Lora 10 s fixed Solar panel Not specified Yes No Compact No No No Against Arduino Not specified
temperature, Server
Aeroqual
and humidity connection

Air temperature,
rainfall, wind
speed, wind
direction, 55 W Solar No
[20] Both No 2 GB SD Card SIM900 GPRS 1 min fixed Not specified No No Compact No No No Arduino Not specified
relative panel C code
humidity,
atmospheric
pressure

Soil humidity
Soil temperature
Air temperature
1.5 months
Air humidity Solar panel
16 GB SD Only 1 mea- No
[21] Atmospheric Both No ESP32 WiFi 1 h fixed 55 W No No Compact No No No RPi Pico Not specified
Card surement per C code
pressure Li-ion battery
hour
Wind vane
Wind direction
Rain gauge

Air temperature,
relative Configurable: Solar panel Yes
32 GB SD Yes
humidity, Both 30 s, 1 min, 100 W Exportable Yes Yes
Yes Card ESP8266 48 h in full Compact No Against STM32 230 mA average
This work atmospheric IP65 box 10 min, 30 AC data in CSV Aluminum No ARIMA
2.4” TFT ThingSpeak WiFi mode All-in-one C code Schneider ARM-M3 in full mode
pressure, CO2 , 1.6 mast min, 1 h, 12 h, Internal API foil model
server thermostat
PM 1, PM 2.5, 24 h. Li-ion battery integration
PM 10

IN/OUT: Indoors and Outdoors. DY: Display. ST: Storage. WT: Wireless Technology. MIT: Measurement Interval Technology. PS: Power Sources. Aut. B: Autonomy with Battery. Op-S:
Open-source. IW: Isolated Wires. OS: Operative System. VR: Validation with Reference. PC: Power Consumption.
 ments to those of specialized devices. For indoor comparisons, a Schneider thermostat
was chosen as the reference tool, while for outdoor comparisons, a Purple Air sensor was
employed. It is crucial to emphasize that the reference sensors, although not entirely im-
mune to measurement errors, were selected based on their use in industry and market
Sensors 2023, 23, 9060 availability and their certifications guaranteeing minimal errors. Consequently, the com- 6 of 26
parison focused on evaluating the monotonic relationship between the OSWS and these
devices and determining errors (RMSE and MAPE) between them.
Moreover,
3. Materials andtoMethods
facilitate data storage, local storage (SD Card) and remote storage
(ThingSpeak
The designserver (https://thingspeak.com/,
steps deployed in this study accessed on 3 October
are depicted 2023))
in Figure were imple-
2, providing an
mented to capture and retain all collected sensor data. An ARIMA forecast model was also
overview of the research process. The initial stage, denoted by the green block, encompasses
applied to predict
developing values within
and fabricating a short-term
the sensing timeframe
platform using historical
for measuring data. Notably,
environmental the
variables
model employed was univariate, considering only one variable per model. Thus,
such as air temperature, relative humidity, atmospheric pressure, CO2 concentration, and the time
series containing
particulate mattervalues of a single
concentration (PMvariable (e.g.,
1, PM 2.5, PMtemperature) exclusively
10). During this predicted
phase, the val-
information
ues of thatprovided
the WMO same variable. Finally,
was used. Thisthe forecast model
international underwent
agency evaluation and
ensures high-quality datacompari-
delivery
son against
to end users.known data, resulting in calculating parameters to assess its performance.

Figure 2. The general methodology.

Figure 2. The general methodology.
Subsequently, once data for all variables were acquired, a comparison was conducted
3.1. Weather Station Design and Development
against commercial products to assess the proximity of the proposed station’s measure-
mentsTheto documents from the World
those of specialized Meteorological
devices. Organizationahave
For indoor comparisons, been followed
Schneider thermostatto
design a robust and accurate weather station as much as possible. There
was chosen as the reference tool, while for outdoor comparisons, a Purple Air sensor is a category
called Automated
was employed. Weather
It is crucial Stations (AWSs),
to emphasize thatand
the the currentsensors,
reference proposal fits intonot
although theentirely
multi-
purpose AWS category. The main feature of these devices is that they measure,
immune to measurement errors, were selected based on their use in industry and market store, and
transmit theand
availability datatheir
collected automatically
certifications [14]. Additionally,
guaranteeing they must
minimal errors. have at least
Consequently, thethree
com-
parts:
parison focused on evaluating the monotonic relationship between the OSWS and these
devices and determining errors (RMSE and MAPE) between them.
Moreover, to facilitate data storage, local storage (SD Card) and remote storage (ThingS-
peak server (https://thingspeak.com/, accessed on 3 October 2023)) were implemented to
capture and retain all collected sensor data. An ARIMA forecast model was also applied
to predict values within a short-term timeframe using historical data. Notably, the model
employed was univariate, considering only one variable per model. Thus, the time series
containing values of a single variable (e.g., temperature) exclusively predicted values of
that same variable. Finally, the forecast model underwent evaluation and comparison
against known data, resulting in calculating parameters to assess its performance.

3.1. Weather Station Design and Development

The documents from the World Meteorological Organization have been followed to
design a robust and accurate weather station as much as possible. There is a category called
Automated Weather Stations (AWSs), and the current proposal fits into the multi-purpose
AWS category. The main feature of these devices is that they measure, store, and transmit
the data collected automatically [14]. Additionally, they must have at least three parts:
Sensors 2023,
2023, 23,
23, 9060
x FOR PEER REVIEW 87of 27
of 26

1. Sensing
Sensinginstruments.
instruments.
2. Local
Localmodem
modemororinterface
interfacefor
foraanetwork
networkconnection.
connection.
3. Central processing unit.
Central processing unit.
Nevertheless, some other parts were added to the project to enhance their usability
and performance. The features included were:
1. Power
PowerSources.
Sources.
2. Human-Machine
Human-MachineInterface.
Interface.
summary,the
In summary, thefive
fivemodules
modulesare areshown
showninin Figure
Figure 3. 3. They
They areare handled
handled by by a micro-
a microcon-
controller
troller unitunit (MCU);
(MCU); in this
in this case,case, the STM32F103C8T6
the STM32F103C8T6 waswas implemented.
implemented.

Figure 3. Modules for the open-source weather station (OSWS).

For the sensing

For the sensing instrument,
instrument, three
three sensors
sensors were
were chosen
chosen and
and are
are shown
shown in
in Table
Table 22 (these
(these
sensors can measure the variables within the conventional ranges):
sensors can measure the variables within the conventional ranges):

Table 2. Sensors incorporated in the weather station.

Sensor Variables
Sensor Variables Measured
Measured Unit Unit Accuracy Accuracy Response
Response
Time Time
◦ ◦

BME280 [30]
Air Temperature
Air Temperature
Relative Humidity
°C % RH
C
±0.5 °C ±±3%
0.5 C
RH 1s
BME280
Relative Humidity
Atmospheric Pressure % RH hPa ±3% RH ±1 hPa 1s
[30]
SCD30 [31]Atmospheric Pressure
CO2 Concentration hPa ppm ±1 hPa ±30 ppm <20 s

SCD30 Particulate Matter 3

PMS5003 [32] µg/m ±10% <10 s
CO2 Concentration
Concentration ppm ±30 ppm <20 s
[31]
PMS5003 Particulate Matter
For versatility purposes, three differentµg/m3 power ±10% <10the
sources were chosen so s station
[32] Concentration
could be placed in scenarios with several conditions:
• 100
For W Solar Panel
versatility purposes, three different power sources were chosen so the station
•couldLi-ion Internal Battery
be placed in scenarios with several conditions:
• AC Plug connection
 100 W Solar Panel
Two storage
 Li-ion Internaloptions
Batteryare available for the storage unit: local and remote. A 32GB SD
card was added to the system for the first option. Since CSV is a flexible and compatible
 AC Plug connection
format, all the measurement values and timestamps were stored in this file.
Twothe
On storage
otheroptions
hand, theareremote
available for the
storage wasstorage
managedunit: through
local andtheremote.
modem A 32GB
ESP8266SD
card was added to the system for the first option. Since CSV is a flexible
with WiFi integrated; the values were sent to the ThingSpeak platform; this service stores and compatible
format,
each all the (i.e.,
variable measurement values
temperature, and timestamps
relative humidity, PM were2.5)
stored
in a in this file.
single channel, and the
valuesOn the other
collected canhand, the remote
be exported via anstorage
API, orwas managed
in CSV format.through the modem
In addition, ESP8266
since this project
with WiFi
aims to be integrated;
open source, the
allvalues were sent
the channels areto the ThingSpeak
open to the public.platform; this service
Table 3 shows stores
the order in
each variable (i.e., temperature, relative
which the columns with the values are stored. humidity, PM 2.5) in a single channel, and the
values collected can be exported via an API, or in CSV format. In addition, since this
Sensors 2023, 23, 9060 8 of 26

Table 3. Order of the columns stored in a CSV file.

Hour Second Date Month Year AT RH AP CO2 PM 1 PM 2.5 PM 10

[h] [s] [DD] [MM] [YY] [◦ C] [RH%] [hPa] [ppm] [µg/m3 ] [µg/m3 ] [µg/m3 ]
AT = Air Temperature. RH = Relative Humidity. AP = Atmospheric Pressure. CO2 = Carbon Dioxide.

Finally, a human–machine interface was implemented so users could operate the

OSWS efficiently. The elements included are:
• Display: to show the measurements on live.
• Display the on/off button.
• Interval selector button: to configure the measurement interval time.
• LED indicators: to indicate the device’s status.
• General on/off button.
Additionally, some other considerations were considered while designing the OSWS.
Table 4 describes the solutions applied to accomplish the WMO recommendations.

Table 4. Recommendations and proposed solutions.

Recommendation (Components) Implemented Solution (Components)

Data acquisition (sensing instruments) Specialized low-cost sensors
Plug-and-play connectors JST connector
CPU STM ARM Cortex-M3
ESP8266 module WiFi for one-way
Modem
communication
RTC clock Internal RTC clock
Internal battery
Power Supply PV module
AC plug
Configurable measurement interval
Sampling one minute at least
30 s, 1 min, 10 min, 30 min, 1 h, 12 h, 24 h
32 GB SD Card up to 10 years for
Available memory for hundreds of days
measurements each 30 s
Recommendation (Design) Implemented solution (Design)
Plug-and-Play sensors, components, display,
Easy component replacement
and MCU.
Stand-alone AWS All-in-one design
Visual indicators s
Extreme weather-proof case IP65 case to cover all the components
Sensors exposed to air but covered from 3D PLA case to protect the sensors exposed
direct sunlight to air
Supporting structure between 1.25 m and 2 m 1.6 m height mast

The forecast process is the final stage of the methodology shown in the diagram in
Figure 3. This part is described in detail in Figure 4.
Sensors 2023, 23, x FOR PEER REVIEW 10 of 27

Sensors 2023, 23, x FOR PEER REVIEW 10 of 27

Sensors 2023, 23, 9060 9 of 26

Figure 4. Methodology for forecast process.

Figure 4.
Figure 4. Methodology
Methodology for
for forecast
forecast process.
process.
As stated before, the forecast ARIMA model is univariate; hence, a single variable is
considered
As to predict
As stated
stated before,values
before, the from the
the forecast
forecast samemodel
ARIMA
ARIMA variable.
model is For instance,
is univariate;
univariate; if temperature
hence,
hence, aa single valuesis
single variable
variable is
are to be
considered expected,
to predictprevious
values temperature
from the same values are
variable. considered
For instance,
considered to predict values from the same variable. For instance, if temperature values inifthe model.
temperature Each set
values of
are
values
to (temperature,
arebetoexpected,
be expected, previousPM temperature
1, PMtemperature
previous 2.5, COvalues
2, etc.) are
is regarded
values considered as aintime
are considered series
the model.
in the since
Eachevery
model. set ofmeas-
Each values
set of
urement has
(temperature, a timestamp
PM 1, PM and
2.5, CO is ,evenly
etc.) is spaced.
regarded This
as a space
time between
series
values (temperature, PM 1, PM22.5, CO2, etc.) is regarded as a time series since every meas-since values
every is configured
measurement
with
has athe
urement button
timestamp that
andsets
has a timestamp theand
is evenlymeasurement
spaced.
is evenlyThis interval
space
spaced. described
between
This in theishuman–machine
space values
between configured with in-
values is configured the
terface.
button
with the that sets the
button thatmeasurement interval described
sets the measurement in the human–machine
interval described in the human–machineinterface. in-
The
Thetime
terface. timeseries
series(marked
(markedas asgreen
greenininFigure
Figure4)4)represents
representsall allthe
thevalues
valuescollected
collectedfrom
from
the
thedevice
device from a single variable. The following step was to split the
The time series (marked as green in Figure 4) represents all the values collected train
from a single variable. The following step was to split dataset
the into
dataset train
into and
from
test
the device from a single variable. The following step was to split the dataset into trainand
and data.
test The
data. first
The part
first trains
part the
trains model;
the model;the other
the other compares
compares thethepredicted
predicted values
values and
and
quantifies
quantifies the
the model’s
model’s performance
performance (Figure
(Figure 5).
5).
test data. The first part trains the model; the other compares the predicted values and
quantifies the model’s performance (Figure 5).

Timeseries
Figure5.5.Time
Figure seriesdivided
dividedinto
intotwo
twoparts.
parts.

Figure 5. Time seriessince

Furthermore, divided
theinto two parts.
ARIMA model requires the time series to be stationary, the
Furthermore, since the ARIMA model requires the time series to be stationary, the
meanand
mean andstandard
standarddeviation
deviationmust
must beconstant;
constant;therefore,
therefore,aastatistical
statisticaltest testcalled
calledAug-
Aug-
mentedFurthermore, sincewas
Dickey–Fuller ARIMAbe
the implemented
model torequires
assess thistherequisite.
time series If to
thebetime
stationary, the
series was
mented
mean Dickey–Fuller
and standard was implemented
deviation must be to assess this
constant; requisite.
therefore, a Ifstatistical
the time series
test was non-
called Aug-
non-stationary, a different method was applied to turn it into stationary. This method
stationary, a different method
mented Dickey–Fuller was appliedto
was implemented toassess
turn itthis
into stationary. This method
seriessubtracts
subtracts the previous value from the current value [33].requisite. If the
The following time
equation was non-
shows the
the previous value from
stationary,differencing the
a different method current value [33]. The following equation shows the first-
first-order process:was applied to turn it into stationary. This method subtracts
order differencing
the previous process:
value from the current value
zt = yt[33]. The
− yt− 1
following equation shows the first- (1)
order differencing process: zt = yt − yt−1 (1)
where:
•
where: zt is the new time series storing the zdifferences t = yt − yt−1 of the original time series. (1)
•where:
ztyist is
the thenewcurrent
time value
series of the original
storing time series.
the differences of the original time series.
• yztyis t−1theis the first
current previous
value of value
the of
original the actual
time time
series. series at time t.
t is the new time series storing the differences of the original time series.
  yyt−1
Onceis the
t is thethisfirst previous
method
current value
is applied,
value of the actual
the ADFtime
of the original time
test can be usedat
series
series. to time
the timet. series Zt to determine
 it Once
if is
yt−1stationary.
is this method
the first When thisvalue
is applied,
previous requisite
the isactual
ADF
of the accomplished,
test can
timebe the
used
series attoorder t.for the
the time
time model
series Zt tomust
deter-be
estimated.
mine ifOnce This model
it is stationary. is made up of three parts: Autoregressive AR(p), Integrated I(d), and
this methodWhen this requisite
is applied, the ADFistest accomplished,
can be usedthe order
to the forseries
time the model
Zt to must
deter-
Moving
be Average MA(q); therefore, three parameters must be estimated: p,Integrated
d, and q [34].
estimated. This
mine if it is stationary. When this requisite is accomplished, the order for the model I(d),
model is made up of three parts: Autoregressive AR(p), must
and The “d”Average
Moving parameterMA(q); represents
therefore, the number of timesmust the be differencing method was
be estimated. This model is made up of three
three parameters
parts: Autoregressive estimated: p, d, and
AR(p), Integrated q
I(d),
applied to the time series. In the case of the “p” parameter, it is estimated with the partial
[34].
and Moving Average MA(q); therefore, three parameters must be estimated: p, d, and q
autocorrelation function (PACF) plot; on the other hand, the “q” parameter is calculated
[34].The “d” parameter represents the number of times the differencing method was ap-
with
plied to the autocorrelation
the time function
the case(ACF).
series. Inrepresents of In both cases, theit lags considered forthe
thepartial
model
The “d” parameter thethe “p”
number parameter,
of times the isdifferencing
estimated with
method was ap-
are those
autocorrelation outside boundaries.
function (PACF) For example,
plot;ofonthe in
the“p”the case of
other hand, theFigure 6, a PACF plot is illustrated;
plied
in that tograph,
the time theseries.
lags 6, In9,the
and case
19 should be parameter,
considered itin“q”
isthe parameter
estimated is calculated
with
model [33,35–37]. the partial
The
with the autocorrelation
autocorrelation function function
(PACF) (ACF).
plot; on In both
the cases,
other the
hand, lags
the considered
“q” parameter foristhe model
calculated
boundaries or horizontal lines represent the confidence intervals, which are statistically
with thetoautocorrelation
derived help determinefunction (ACF). Inofboth
the significance cases, the
the partial lags considered
autocorrelations for the model
at different lags.
 are those outside boundaries. For example, in the case of Figure 6, a PACF plot is illus-
are those outside boundaries. For example, in the case of Figure 6, a PACF plot is illus-
trated; in that graph, the lags 6, 9, and 19 should be considered in the model [33,35–37].
trated; in that graph, the lags 6, 9, and 19 should be considered in the model [33,35–37].
The boundaries or horizontal lines represent the confidence intervals, which are statisti-
The boundaries or horizontal lines represent the confidence intervals, which are statisti-
Sensors 2023, 23, 9060 cally derived to help determine the significance of the partial autocorrelations at different
10 of 26
cally derived to help determine the significance of the partial autocorrelations at different
lags.
lags.

Figure 6. PACF plot of a temperature times series.

Figure 6.
Figure 6. PACF
PACF plot
plot of
of aa temperature
temperaturetimes
timesseries.
series.
3.2.
3.2. Data Validation
3.2. Data
Data Validation
Validation
The
The system proposed was tested on indoor and outdoor scenarios and compared
The system proposed
system proposed was was tested
tested on on indoor
indoor and and outdoor
outdoor scenarios
scenarios and and compared
compared
against
against a reference instrument. The first scenario was inside an enclosed room, and the
against a reference instrument. The first scenario was inside an enclosed room,
a reference instrument. The first scenario was inside an enclosed room, andand the
the
reference
reference instrument was Schneider’s thermostat (ST), which measures only temperature
reference instrument
instrumentwas was Schneider’s
Schneider’s thermostat
thermostat (ST),(ST), which
which measures
measures only
only temperature
temperature
and
and humidity and shows the live measurements through aa display.
and humidity
humidity and and shows
showsthethelive
live measurements
measurementsthrough through adisplay.
display.
Schneider
Schneider personnel gave access to the site where the thermostat was placed for 11hh
Schneider personnel gave access to the site where the
personnel gave access to the site where the thermostat
thermostat was was placed
placed for
for 1 h
(from
(from 12 pm to 1 pm); since photos were not allowed, Figure 7 shows a layout of the area.
(from12 12 pm
pm to to 11 pm);
pm); since
since photos
photos were
were notnot allowed,
allowed, Figure
Figure 77 shows
shows aa layout
layout of of the area.
the area.
As
As illustrated,
illustrated, the
the proposed
proposed OSWS
OSWS (red
(red rectangle)
rectangle) was was placed
placed next next
to theto the thermostat
thermostat (green
As illustrated, the proposed OSWS (red rectangle) was placed next to the thermostat
(green rectangle)
rectangle) and was and was configured
configured to havetoto have
the themeasurement
same same measurement interval,
interval, in this in this30
case, case,
s. It
(green rectangle) and was configured have the same measurement interval, in this case,
30
is s. It is essential
essential to to highlight
highlight that that were
there thereno were no windows,
windows, and and
the the single
single door doorclosed;
was was closed;
hence,
30 s. It is essential to highlight that there were no windows, and the single door was closed;
hence,
there there
was nowas no airflow.
airflow. DuringDuring this process,
this process, the OSWSthe OSWS was powered with an internal
hence, there was no airflow. During this process, the was
OSWS powered with an
was powered internal
with Li-ion
an internal
Li-ion
battery. battery.
Li-ion battery.

Thelayout
Figure7.7.The
Figure layoutof
ofthe
themeasurement
measurementsite.
site.
Figure 7. The layout of the measurement site.
Foroutdoor
For outdoorcomparison,
comparison,the the OSWS
OSWS waswas placed
placed onon a house
a house roofroof exposed
exposed to air
to the thefor
air
Forday.
outdoor comparison, the OSWS was placed on a sensor
house roof exposed to the air for
one day. The reference instrument was a Purple Air sensor 2.118 km away, which is con-is
for one The reference instrument was a Purple Air 2.118 km away, which
one day. The
considered reference
a small instrument
distance, thewas a Purple Air sensor 2.118 km away, which is con-
sidered a small distance, and and weather
the weather conditions.
conditions.
sidered a small
Figure88shows distance,
showsthethemap and
mapand the
andthe weather conditions.
thedistance
distancethe
theGoogle
GoogleMaps
Mapsappappestimated.
estimated.TheThepurple
purple
Figure
point Figure
is where8 shows
the the map
weather and the
station wasdistance
placed, the
andGoogle
the redMaps
point app estimated.
represents the The purple
Purple Air
point is where the weather station was placed, and the red point represents the Purple Air
point
sensor. is where
The line the weather
represents station
the was
distance placed,
between and
thethe red
OSWS point
and represents
the Purple the
Air Purple
sensor. Air
The
sensor. The line represents the distance between the OSWS and the Purple Air sensor. The
sensor.
red dotsThe line
arethe represents
thespecific
specific the distance between the OSWS and the Purple Air sensor. The
locations.
red dots are locations.
red dots are the specific locations.
Sensors 2023, 23, x FOR PEER REVIEW 12 of 27

Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 11 of 26
12 27

Figure 8. Distance between OSWS and Purple Air sensor .

The measuring time was 24 h (from 13th May at 01:30 to 14th May at 1:30), with a
Figure8.8.Distance
Figure
measurement Distance betweenOSWS
between
time interval OSWS
of andPurple
and
10 min. Purple
The Airsensor.
Air
data sensor . from the OSWS were stored on
obtained
an SD card for further analysis, and the data from the reference tool were downloaded
from Purple The measuring
The measuring
Air´s website. time was
time was 24 24 hh (from
(from 13th
13th May
May at at 01:30
01:30 toto 14th
14th May
May atat1:30),
1:30), with
with aa
measurement
measurementtime timeinterval
intervalofof1010min.
min.The data
The dataobtained
obtained from
fromthethe
OSWSOSWSwere stored
were on an
stored on
SD card for further analysis, and the data from the reference tool
an SD card for further analysis, and the data from the reference tool were downloaded
4. Results were downloaded from
Purple Air´s website.
from Purple Air´s website.
4.1. Design Results
4.4.Following
Results
Results the WMO recommendations described earlier and inspired by the state-
4.1. Design
of-the-art Results
projects, the OSWS prototype was developed. The final result is shown in Figure
4.1. Design Results
9, illustrating the size
Following the measurements.
WMO recommendations All the electronic
describedcomponents
earlier andthat handle
inspired byand
the man-
state-of-
agethe-art Following
the system are the WMO
placed insiderecommendations
this case. described earlier and inspired by the state-
projects, the OSWS prototype was developed. The final result is shown in Figure 9,
of-the-art
Moreover, projects,
Figure the
10 OSWS prototype
illustrates the was developed. The
the final result is shown in Figure
illustrating the size measurements. Allproposed solution;
the electronic components photosthatshow
handletheand
weather
manage
9,
station illustrating
attached the size measurements. All the electronic components that handle and man-
the system aretoplaced
a 1.65 inside
m aluminum
this case.base powered by the internal Li-ion battery.
age the system are placed inside this case.
Moreover, Figure 10 illustrates the proposed solution; the photos show the weather
station attached to a 1.65 m aluminum base powered by the internal Li-ion battery.

Figure
Figure 9. OSWS
9. OSWS sizesize measurements.
measurements.

Moreover, Figure 10 illustrates the proposed solution; the photos show the weather
station attached
Figure 9. to ameasurements.
OSWS size 1.65 m aluminum base powered by the internal Li-ion battery.
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 12 of
13 of 26
27

Figure 10. OSWS mounted in a 1.65 m height base.

Figure 3 shows three sensors in the sensing system that measure air temperature,
relative humidity, atmospheric pressure, CO2 concentration, and PM 1, PM 2.5, and PM
10 concentration. According to the WMO, the sensors should be exposed to the air but
protected from rain or direct sunlight. Figure 11 shows how the sensors are distributed
along the case, which has IP65 protection, keeping the internal electronic components
away from
Figure dust
10.OSWS
OSWS and water,in
mounted and1.65
themaccessory at the top of the device is designed to protect
Figure 10. mounted in aa1.65 heightbase.
m height base.
the sensors from rain and sun.
Figure
Figure
Figure 1133presents
shows the following
shows three
three sensors components:
sensors in
in the sensing
the sensing system
system that
that measure
measure airair temperature,
temperature,
1.relative
Accessory
relative for rain
humidity,
humidity, and sun. pressure,
atmospheric
atmospheric pressure,CO CO22 concentration,
concentration, andand PMPM 1,1, PM
PM 2.5,
2.5,and
andPMPM
2.10
10BME280 measures
concentration.
concentration. temperature,
According
According to therelative
to the WMO,humidity,
WMO, the sensors
the sensorsand atmospheric
should
should be pressure.
be exposed
exposed to the
to the air
airbut
but
3.protected
SCD30 for
protected fromCOrain
from 2 concentration
rain ordirect
or measurement.
directsunlight.
sunlight. Figure 11
Figure 11shows
showshow howthe
thesensors
sensorsare
aredistributed
distributed
4.along
along the
thecase,
PMS5003 forwhich
case, PM hashas
(PM
which IP65
1, PM protection,
IP65 keeping
2.5,protection,
and PM 10) the internal
concentration.
keeping electronic
the internal components
electronic away
components
5.from dust
IP65
away and
case.
from dustwater, and the
and water, accessory
and at theat
the accessory toptheoftop
theofdevice is designed
the device to protect
is designed the
to protect
sensors from rain and sun.
the sensors from rain and sun.
Figure 11 presents the following components:
1. Accessory for rain and sun.
2. BME280 measures temperature, relative humidity, and atmospheric pressure.
3. SCD30 for CO2 concentration measurement.
4. PMS5003 for PM (PM 1, PM 2.5, and PM 10) concentration.
5. IP65 case.

Figure
Figure11.
11.Sensor
Sensorand
andprotection
protectionfor
forrain
rainand
andsun.
sun.

Figure 11 presents the following components:

1. Accessory for rain and sun.
2. BME280 measures temperature, relative humidity, and atmospheric pressure.
3. SCD30 for CO2 concentration measurement.
4. PMS5003 for PM (PM 1, PM 2.5, and PM 10) concentration.
5.Figure
IP65
11. case.
Sensor and protection for rain and sun.
Sensors 2023, 23, 9060 13 of 26

Sensors 2023, 23, x FOR PEER REVIEW 14 of 27

Sensors 2023, 23, x FOR PEER REVIEW 14 of 27
Moreover, the HMI, shown in the diagram in Figure 3, was implemented as illustrated
in Figure 12, where:
Moreover, the HMI, shown in the diagram in Figure 3, was implemented as illus-
1. Indicator LEDs. Green: if the system is measuring. Yellow: if there is some trouble
tratedMoreover,
in Figure 12,the where:
HMI, shown in the diagram in Figure 3, was implemented as illus-
with a sensor.
trated
1. in Figure
Indicator 12, Green:
LEDs. where:if the system is measuring. Yellow: if there is some trouble with
2. 2.4” Display (Figure 13) showing:
1. aIndicator
sensor. LEDs. Green: if the system is measuring. Yellow: if there is some trouble with
2. 2.4″a. Live(Figure
Display measurements.
13) showing:
a sensor.
b.
a. Display
2. 2.4″ Interval selected.
Live measurements.
(Figure 13) showing:
c. Interval
b.
a. SD measurements.
Live status (inserted or removed).
selected.
d.
c.
b. SD Sensing (inserted orstatus
platform
statusselected.
Interval (measuring or not).
removed).
3. d. Interval
c. Sensingselector:
SD status used
platform
(inserted to
or configure
status the or
(measuring
removed). interval
not). measuring time:
3. Interval
d.
a. selector:
Sensing
30 s used
platform to configure
status the interval
(measuring or measuring time:
not).
a.
3. Interval
b. 30 s
selector:
1 min used to configure the interval measuring time:
b.
a.
c. 130min
s
10 min
c.
d. 10
b. minmin
1 min
30
d. 30
c.
e. 10 1min
min
h
e.
d. 130h min
f. 12 h
f.e. 12 1 hh
g. 24 h
g.
f. 24 12 hh
4. On/off
4. On/off
g. h button
24button for for
the the display.
display.
4. On/off button for the display.

Figure 12.
Figure 12. HMI
HMIimplemented
implementedininthe
thedevice.
device.
Figure 12. HMI implemented in the device.

Figure 13. Display.

Figure 13.
Figure 13. Display.
Display.
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 14 of 26
15 27

Two
Twomicroprocessors
microprocessorsand
anda wireless board
a wireless were
board installed
were to handle
installed all the
to handle allsensors and
the sensors
the HMI. This solution prevents overheating a single MCU and provides an easier
and the HMI. This solution prevents overheating a single MCU and provides an easier way to
modularize the sub-systems. Following this design, two PCBs were designed: main
way to modularize the sub-systems. Following this design, two PCBs were designed: main PCB
(Figure 14) and
PCB (Figure 14)second PCB (Figure
and second 15). 15).
PCB (Figure
Figure 14. Main PCB.

The main PCB includes:

1. The main MCU that manages the sensors stores the data in the SD card, controls the
timestamp obtained from the RTC, and sends the data shown through the display.
2. SD module controlled with SPI protocol.
3. Button for removing or inserting the SD safely.
4. 3V Coin battery to power the RTC even when the primary power source is discon-
nected.
5. Voltage regulators.
6. JST connectors for a secure component connection.
On the other hand, the second PCB includes:
1. MCU that manages the 2.4″ display and controls the WiFi board.
2. JST connector for communication with the main PCB.
3. WiFi board (ESP8266) that sends the data via the internet to ThingSpeak.
Figure
Figure
4. 14. Main
14.
HeadersMain PCB.
toPCB.
attach the display.

The main PCB includes:

1. The main MCU that manages the sensors stores the data in the SD card, controls the
timestamp obtained from the RTC, and sends the data shown through the display.
2. SD module controlled with SPI protocol.
3. Button for removing or inserting the SD safely.
4. 3V Coin battery to power the RTC even when the primary power source is discon-
nected.
5. Voltage regulators.
6. JST connectors for a secure component connection.
On the other hand, the second PCB includes:
Figure
1. MCU
Figure 15. Second
15. Second PCB.
that manages
PCB. the 2.4″ display and controls the WiFi board.
2. JST connector for communication with the main PCB.
4.2.
3. Ergonomics
The main
WiFi Assessment
boardPCB includes:
(ESP8266) that sends the data via the internet to ThingSpeak.
1.4. Headers
The
Oncemain MCU
attachthat
thetoprototype manages
thewas
display. the sensors
functional, it wasstores the data
evaluated in the of
in terms SDergonomics.
card, controls
It the
is a
timestamp obtained from the RTC, and sends the data shown through
science focused on equipment design with a human-centered approach to reduce operatorthe display.
2. SDor
fatigue module controlled
discomfort with SPI
considering protocol.
physical, cognitive, social, organizational, and environ-
3. Button
mental for [38,39].
factors removing or inserting
Since the focuses
this science SD safely.
on improving design performance, the
4. 3V Coin battery to power the RTC even when the primary power source is discon-
nected.
5. Voltage regulators.
6. JST connectors for a secure component connection.
On the other hand, the second PCB includes:
1. MCU that manages the 2.4” display and controls the WiFi board.
2. JST connector for communication with the main PCB.
3. WiFi board (ESP8266) that sends the data via the internet to ThingSpeak.
Figure
4. 15. Second
Headers PCB. the display.
to attach

4.2. Ergonomics
4.2. Ergonomics Assessment
Assessment
Once the prototype was
Once was functional,
functional,ititwas
wasevaluated
evaluatedininterms of of
terms ergonomics.
ergonomics.It isIta
science
is focused
a science on equipment
focused on equipmentdesigndesign
with a with
human-centered approach
a human-centered to reducetooperator
approach reduce
operator
fatigue orfatigue or discomfort
discomfort consideringconsidering
physical,physical, cognitive,
cognitive, social, organizational,
social, organizational, and
and environ-
environmental
mental factorsfactors
[38,39].[38,39]. Sincescience
Since this this science focuses
focuses on improving
on improving design
design performance,
performance, the
Sensors 2023, 23, 9060 15 of 26

the degree of usability must be quantified to accurately assess the ease of use of the
designed prototype; therefore, several helpful methods exist for this purpose. For instance,
keystroke-level model, link analysis, questionaries, checklist, hierarchical task analysis,
layout analysis, and interview, among others [39]; the selection of the methods for the
assessment depends on the project´s design stage in this case. the project has a functional
and physical prototype. Therefore, the methods selected were observation for the usability
link analysis and layout analysis for the ergonomics test [39].
The observation method was performed by giving the users the device and asking
them to perform planned activities; in this case, six users were selected, and six different
activities were established:
1. Move the station to a specific point.
2. Turn on the device.
3. Set the measurement interval time to 24 h.
4. Turn off the display.
5. Open the case to get access to the internal components.
6. Remove the SD card.
Only six users were selected since, when evaluating the usability of a product, it is
enough for only five users to use the product because they will find more than 75% of
the product issues. Using fewer users will lead to not covering most of these problems;
meanwhile, using more than six users, repeated issues will be found, making the results
meaningless [40].
Each user had ten points at the beginning of the test, and then each wrong action or
asking of a question was penalized by subtracting one point each time; in addition, the time
that the user took to complete all the activities was measured. Table 5 shows the average
score of each activity performed by the six users and the average time. In general, the
overall score was 8.7 regarding usability.

Table 5. Observation method results for the usability test.

Activity Score Average Average Time

Move the station to a
9.5
specific point
Turn on the device 9.5
Set the measurement interval
9
time to 24 h
4:14 min
Turn off the display 8.8
Open the case to get access to
8.6
the internal components
Remove the SD card 6.7
Average 8.7

Regarding ergonomics, the layout and link analyses were performed simultaneously
on the human–machine interface presented in Figure 12. The first method states that
the components from the same category must be as close as possible. There were only
two categories found: device control and power. On the other hand, the second method
consists of establishing a sequence of actions that would be one of the most common
sequences while using the device in a normal operational mode and tracing lines between
the components following that order [39]. In this case, the sequence was:
1. Turn on the device.
2. Check the indicator LEDs.
3. Turn on the display.
4. Watch the display and the measurements.
 Sensors 2023, 23, x FOR PEER REVIEW 17 of 27

Sensors
Sensors 23,23,
2023,
2023, 9060 PEER REVIEW
x FOR 17 16
of of
2726
Figure 16 depicts the two groups (red for device control and blue for power) and the
lines (green) that follow the order established. The device control group includes the com-
5. Figure
ponentsChange
labeled theB,measuring
16 depicts C, the
D, and interval
twoE; timefor
the power
groups (red to adevice
group different
includesvalue.
controlonly
andcomponent A.
blue for power) and the
Once
Figure
lines (green) the16lines
that are the
depicts
follow traced,
twothe
theorder method
groups (red
established. establishes
devicethat
for device
The no line
control andshould
groupblue cross
for
includes with
power)
the an-
and
com-
other
the to
linesenhance
(green) operability
that follow and
the ergonomics.
order In
established.this case,
The line
device
ponents labeled B, C, D, and E; the power group includes only component A. A-E crosses
control groupline C-D, indi-
includes the
cating that
components one or more
labeled B, components
C, D, and E; should
the powerbe relocated.
group includes only component
Once the lines are traced, the method establishes that no line should cross with an- A.
other to enhance operability and ergonomics. In this case, line A-E crosses line C-D, indi-
cating that one or more components should be relocated.

Figure16.
Figure Layoutanalysis
16.Layout analysisand
andlink
linkanalysis.
analysis.

Once the
17 lines are traced, the method establishes that no line should cross with another
FigureFigure
16. Layoutshows
analysisthe
andrearrangement
link analysis. of the components that satisfy the link analysis
to enhance operability
requirements. and ergonomics.
Furthermore, In thisthat
it can be observed case, line
the A-E
four crosses (B,
elements lineC,C-D, indicating
D, and E) are
that one
nowFigure or
groupedmore components
and satisfy should be relocated.
the layout analysis requirements.
17 shows the rearrangement of the components that satisfy the link analysis
Figure 17 shows the rearrangement of the components that satisfy the link analysis
requirements. Furthermore, it can be observed that the four elements (B, C, D, and E) are
requirements. Furthermore, it can be observed that the four elements (B, C, D, and E) are
now grouped and satisfy the layout analysis requirements.
now grouped and satisfy the layout analysis requirements.

Figure 17. Components relocated.

Figure
Figure 17.17. Components
Components relocated.
relocated.
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 17 of 26
18 27

4.3.
4.3. Data
Data Comparison
Comparison
The
The first comparison
first comparison was was between
between the
the designed
designed weather
weather station
station and
and the
the Schneider
Schneider
Electric thermostat. As stated, the two devices were placed together,
Electric thermostat. As stated, the two devices were placed together, as shown as shown in in
Figure 7.
Figure
The data from the reference instrument were provided directly by Schneider´s
7. The data from the reference instrument were provided directly by Schneider´s person- personnel.
Figure 18 shows
nel. Figure the two
18 shows signals;
the two the blue
signals; lineline
the blue represents the the
represents measurements
measurements performed
performedby
the current
by the proposal,
current andand
proposal, thethe
orange
orangeone is the
one reference
is the referencevalue.
value.At
Atfirst
firstglance,
glance,the
thetwo
two
signals
signalsseem
seemto tohave
havethe
thesame
samedirection,
direction,and
andaamonotonic
monotonicrelationship
relationshipcancanbebeinferred.
inferred.

Figure18.
Figure 18. OSWS
OSWS and
and reference
referencedevice
deviceplot
plotin
inan
anindoor
indoorscenario
scenariofor
fortemperature.
temperature.

Statistical parameters
Statistical parameters were were obtained
obtained toto compare
compare the the two
two signals
signals andand determine
determine theirtheir
relationship. The
relationship. The first
first non-parametric
non-parametric value value isis the
the Spearman
Spearman correlation,
correlation, which
which indicates
indicates
the strength of the
the the association,
association,a anumber
numberbetween
between −1−and 1, and
1 and the direction
1, and of theofasso-
the direction the
association,
ciation, which which
can can be positive
be positive or negative
or negative [40]. [40]. Additionally,
Additionally, the mean
the root root mean
square square
error
error
(RMSE)(RMSE) and mean
and mean absolute
absolute percentage
percentage errorerror
MAPE MAPE
werewere calculated
calculated to determine
to determine the
the per-
performance
formance of of thethe measurementprocess;
measurement process;these
theseparameters
parametersare aremainly
mainly focused
focused on rating
rating aa
forecast
forecastmodel’s
model’saccuracy,
accuracy,butbutthey
theycan bebe
can applied
applied to quantify
to quantify thethe
total error
total of two
error signals.
of two sig-
nals.Moreover, a Bland–Altman plot was generated to assess the agreement between these
two signals.
Moreover, Onathe y-axis, the differences
Bland–Altman between to
plot was generated theassess
measurements
the agreement are displayed;
between these on
twox-axis,
the signals.theOn
averages of the
the y-axis, measurements
the are plotted.
differences between Each plot on this
the measurements aregraph is a pair
displayed; on
of measurements taken by the two instruments. Finally, this Bland–Altman
the x-axis, the averages of the measurements are plotted. Each plot on this graph is a pair method has
two limits of agreement
of measurements takenatby ±1.96 times
the two the standardFinally,
instruments. deviation.
thisIdeally, all the points
Bland–Altman method should
has
lie between these two limits, representing an agreement between them
two limits of agreement at ±1.96 times the standard deviation. Ideally, all the points should [41]. Figure 19
shows the Bland–Altman plot generated by comparing the time series.
lie between these two limits, representing an agreement between them [41]. Figure 19
showsThisthefigure indicates aplot
Bland–Altman good agreement
generated between these
by comparing signals
the time since most points are
series.
between the interval limits [41]. The differences between the paired values increase when
the temperature is between 23.6 ◦ C and 23.8 ◦ C. Table 6 shows the values obtained for
the other statistical parameters. The correlation parameter is a high value indicating a
strong relationship between the two measurements [42]. In addition, RMSE and MAPE are
low values, demonstrating that the OSWS has similar measurement values to Schneider´s
thermostat.
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 18 of
19 of 26
27

Figure 19. Bland–Altman plot for temperatures measured indoors.

This figure indicates a good agreement between these signals since most points are
between the interval limits [41]. The differences between the paired values increase when
the temperature is between 23.6 °C and 23.8 °C. Table 6 shows the values obtained for the
other statistical parameters. The correlation parameter is a high value indicating a strong
relationship between the two measurements [42]. In addition, RMSE and MAPE are low
Figure 19.Bland–Altman
Bland–Altmanplot plotfor
for temperaturesmeasured
measuredindoors.
indoors.
Figure
values,19.demonstrating that thetemperatures
OSWS has similar measurement values to Schneider´s ther-
mostat.
This figure indicates a good agreement between these signals since most points are
Table 6. Statistical
between parameters
the interval limits for temperature
[41]. measurements.
The differences between the paired values increase when
Table 6. Statistical parameters for temperature measurements.
the temperature is
Spearman Correlation
between 23.6 °C and 23.8
RMSE
°C. Table 6 shows the values MAPE
obtained for the
Spearmanparameters.
other statistical CorrelationThe correlation RMSE
parameter is a high valueMAPE
indicating a strong
0.96 0.19 ◦ C 0.67%
relationship between0.96 the two measurements 0.19[42].
°C In addition, RMSE and 0.67% MAPE are low
values, demonstrating that the OSWS has similar measurement values to Schneider´s ther-
In
In assessing humidity
mostat. humiditymeasurements,
measurements,Figure Figure2020illustrates
illustrates a comparative
a comparative analysis
analysis be-
between
tween the theproposed
proposedOSWS OSWSand andthetheSchneider’s
Schneider’sthermostat
thermostatover overaa specified
specified time frame.
Table 6. Statistical
Concurrently,
Concurrently, parameters
Figure
Figure for temperature
2121presents
presents the
the measurements.
Bland–Altman
Bland–Altman plot, highlighting
plot, highlighting thatthat
most
mostvalues re-
values
side within
reside Spearmanthe stipulated
within the Correlation boundaries. This
stipulated boundaries. This pattern suggests a favorable
pattern suggests a favorable level of agreement
level of agree-
RMSE MAPE
between the two
ment between measurements
the two measurements (air temperature and relative
(air temperature humidity)
and relative since most
humidity) sincepoints
most
are between 0.96
the twothe red lines 0.19 °C 0.67%
points are between two redrepresenting the limits
lines representing of agreement
the limits in which
of agreement 95% 95%
in which of the
of
differences
the differences between the two
between measurements
the two measurements are expected to fall.
are expected to fall.
In assessing humidity measurements, Figure 20 illustrates a comparative analysis be-
tween the proposed OSWS and the Schneider’s thermostat over a specified time frame.
Concurrently, Figure 21 presents the Bland–Altman plot, highlighting that most values
reside within the stipulated boundaries. This pattern suggests a favorable level of agree-
ment between the two measurements (air temperature and relative humidity) since most
points are between the two red lines representing the limits of agreement in which 95% of
the differences between the two measurements are expected to fall.

Figure 20.
Figure 20. OSWS
OSWS and
and reference
referencedevice
devicein
inan
anindoor
indoorscenario
scenariofor
forrelative
relativehumidity.
humidity.

Figure 20. OSWS and reference device in an indoor scenario for relative humidity.
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 19 of 26
20 27

Figure21.
Figure 21. Bland–Altman
Bland–Altman plot
plotfor
forrelative
relativehumidity
humiditymeasured
measuredindoors.
indoors.

The
The OSWS
OSWS was was juxtaposed
juxtaposed withwith aa Purple
Purple Air
Air sensor
sensor located
located approximately
approximately 2.1182.118 km
km
away
awayininthetheoutdoor
outdoor environment.
environment. TheThemethodologies
methodologies employed
employedwere were
consistent with those
consistent with
used
thoseinused
the indoor experiments.
in the indoor Comprehensive
experiments. quantitative
Comprehensive findings
quantitative are enumerated
findings are enumer- in
Table 7. Table
ated in Moreover, Figure 22
7. Moreover, depicts
Figure the atmospheric
22 depicts pressure
the atmospheric graphs graphs
pressure where thewhereupper
the
blue
upperline represents
blue the signal
line represents from the
the signal fromproposed station.
the proposed In contrast,
station. the red
In contrast, theline
red at the
line at
bottom corresponds to the reference sensor. Notably, the difference in
the bottom corresponds to the reference sensor. Notably, the difference in readings be- readings between
the twothe
tween signals in Figure
two signals 22 could
in Figure 22 be attributed
could to the to
be attributed variance in altitude
the variance wherewhere
in altitude each
sensor was positioned.
each sensor was positioned.Figure 23 highlights
Figure the measurement
23 highlights the measurement discrepancy,
discrepancy,revealing an
revealing
average difference
an average of 1.07
difference of hPa.
1.07 Figure 24 shows
hPa. Figure a Bland–Altman
24 shows plot thatplot
a Bland–Altman affirms
that aaffirms
notablea
alignment between between
notable alignment the two signals,
the two with most
signals, of the
with mostpoints nestled
of the pointswithin
nestledthewithin
limitsthe
of
agreement. Additionally, it is paramount to consider the inherent biases
limits of agreement. Additionally, it is paramount to consider the inherent biases of the of the sensors,
which canwhich
sensors, introduce measurement
can introduce discrepancies.
measurement External noise,
discrepancies. including
External environmental
noise, including envi-
factors
ronmental factors or electronic interference, can also influence the readings,the
or electronic interference, can also influence the readings, affecting system’s
affecting the
accuracy.
system’s accuracy.

Table7.7.Statistical
Table Statisticalparameters
parametersfor
formeasurements
measurementsin
inan
anindoor
indoorscenario.
scenario.

Variable
Variable Spearman Correlation
Spearman Correlation RMSERMSE MAPE
MAPE
AirAir Temperature
Temperature 0.97 0.97 1.62 ◦ C1.62
°C 7.01%
7.01%
Relative
Relative Humidity
Humidity 0.81 0.81 4.0 %RH4.0 %RH 8.4%
8.4%
Atmospheric
Atmospheric Pressure
Pressure 0.99 0.99 1 hPa 1 hPa 0.14%
0.14%
PM 1 0.86 2.45 µg/m3 8.7%
PM 1 0.86 2.45 µg/m3 8.7%
PM 2.5 0.87 3.5 µg/m3 7.9%
PM 2.5 0.87 3.5 µg/m3 7.9%
PM 10 0.87 4.6 µg/m3 8.9%
PM 10 0.87 4.6 µg/m3 8.9%
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 20 of 26
21 27
Sensors 2023, 23, x FOR PEER REVIEW 21 of 27

Figure 22.
Figure 22. OSWS
OSWS and
and reference
referencedevice
deviceplot
plotin
inan
anoutdoor
outdoorscenario
scenariofor
foratmospheric
atmosphericpressure.
pressure.
Figure 22. OSWS and reference device plot in an outdoor scenario for atmospheric pressure.

Figure 23. Differences between the two pressure time series.

Figure23.
Figure 23.Differences
Differencesbetween
betweenthe
thetwo
twopressure
pressuretime
timeseries.
series.
Sensors 2023,23,
Sensors2023, 23,9060 21 of
of 26
Sensors 2023, 23, xx FOR
FOR PEER
PEER REVIEW
REVIEW 22
22 of 27
27

Figure24.
Figure
Figure 24.Bland–Altman
24. Bland–Altmanplot
Bland–Altman plotfor
plot forpressure
for pressuremeasured
pressure measuredoutdoors.
measured outdoors.
outdoors.

Furthermore,concerning
Furthermore,
Furthermore, concerningPM
concerning PM2.5
PM 2.5measurements,
2.5 measurements,Figure
measurements, Figure25
Figure 25captures
25 capturesthe
captures thesimultaneous
the simultaneous
simultaneous
readings
readings from the OSWS and the reference device in an outdoor scenario. FigureFigure
readings from the
the OSWS
OSWS and
and the
the reference
reference device
device in in
an an outdoor
outdoor scenario.
scenario. Figure 26
26 details
26 details
details
the the disparities
the disparities
disparities between
between between
the twothe
the two PMtwo
PM 2.5 PM
2.5 time2.5
time time shedding
series,
series, series, shedding
shedding onlight
light on
light on periods
periods
periods of
of conver-
of conver-
convergence
gence and and divergence.
divergence. Finally,
Finally, FigureFigure
27 27 employs
employs a a Bland–Altman
Bland–Altman
gence and divergence. Finally, Figure 27 employs a Bland–Altman plot to visualize the plotplot
to to visualize
visualize the
the agreement
agreement
agreement of of PM
of PM
PM 2.5 2.5
2.5 values
values
values measured
measured
measured outdoors
outdoors
outdoors between
between
between the the
the OSWS
OSWS OSWSandand
and the the
the reference
reference
reference de-
de-
device,
vice, reinforcing
reinforcing the
the consistency
consistency and
and reliability
reliability ofof the
the OSWS
OSWS
vice, reinforcing the consistency and reliability of the OSWS measurements. measurements.
measurements.

Figure 25.
Figure 25. OSWS
OSWS and
and reference
reference device
device plot
plot in
in an
an outdoor
outdoor scenario
scenario for
for PM
PM 2.5.
2.5.
Figure 25. OSWS and reference device plot in an outdoor scenario for PM 2.5.
Sensors 2023,23,
Sensors2023, 23,9060
x FOR PEER REVIEW 22 of
23 of 26
27
Sensors 2023, 23, x FOR PEER REVIEW 23 of 27

Figure 26. Differences

Differences betweenthe
the twoPM
PM 2.5time
time series.
Figure 26. Differencesbetween
Figure26. between thetwo
two PM2.5
2.5 timeseries.
series.

Figure 27. Bland–Altman plot for PM 2.5 measured outdoors.

Figure27.
Figure 27. Bland–Altman
Bland–Altman plot
plotfor
forPM
PM2.5
2.5measured
measuredoutdoors.
outdoors.
These numbers provide essential information and insights about the data gathered.
These numbers
These numbers provide
provide essential
essential information
information and insights about the data gathered.
Regarding the Spearman correlation, all the data collected from the sensors are considered
Regardingthe
Regarding theSpearman
Spearmancorrelation,
correlation,all
allthe
thedata
data collected
collectedfrom
fromthe
the sensors
sensors are
are considered
considered
to have a robust positive correlation since all the Spearman correlation values are above
to
to have
have aarobust
robustpositive
positive correlation
correlation since
since all
allthe
theSpearman
Spearmancorrelation
correlation values
values are
areabove
above
0.8 [40,42]. In the case of the RMSE parameter, all the measurements were below five units,
0.8
0.8[40,42].
[40,42]. In
In the case of the RMSE parameter,
parameter, all the measurements
measurements were were below
below five units,
and the MAPE was below 10%. It is essential to mention that the air temperature and
and
and the
the MAPE
MAPE was was below
below 10%.
10%. It is
is essential
essential toto mention
mention that
that the
the air
air temperature
temperature and
and
relative humidity may be affected by the buildings near the OSWS and the house where
relative
relative humidity
humidity may
may bebe affected
affected by
by the
the buildings
buildings near
near the
the OSWS
OSWS and and the
the house
house where
where
the device was placed [14].
the
the device
devicewas
was placed
placed[14].
[14].
5. Discussion
5.
5. Discussion
Discussion
5.1. Comparison
5.1. Comparison with
with the
the Existing
ExistingLiterature
Literature
5.1. Comparison with the Existing Literature
Theintroduction
The introductionof ofthe
theOSWS
OSWSin inthe
the realm
realmof of weather
weather monitoring
monitoringmarks
marksaa significant
significant
The introduction of the OSWS in the realm of weather monitoring marks a significant
stride. To
stride. Tocontextualize
contextualizeits itsplace
placeamidst
amidstthe
theextensive
extensiveliterature
literatureon
onthe
thesubject,
subject,ititisisvital
vitalto
to
stride. To contextualize its place amidst the extensive literature on the subject, it is vital to
delineate
delineatehowhow the
the OSWS
OSWS aligns
aligns with,
with, deviates
deviates from,
from, andand potentially
potentially surpasses contempo-
delineate how the OSWS aligns with, deviates from, and potentially surpasses contempo-
rary
rarysolutions.
solutions.The core
The strength
core of OSWS
strength lies inlies
of OSWS its commitment in democratizing
in its commitment weather
in democratizing
rary solutions. The core strength of OSWS lies in its commitment in democratizing
data access
weather using
data cost-effective
access sensors. Many
using cost-effective monitoring
sensors. systems insystems
Many monitoring the current
in theliterature
current
weather data access using cost-effective sensors. Many monitoring systems in the current
Sensors 2023, 23, 9060 23 of 26

either prioritize accuracy over affordability or vice versa. However, the OSWS strikes a
balance, ensuring that smaller communities and users are not left behind in the age of
data-driven decision-making. This holistic approach offers a twofold advantage: it ensures
affordability without compromising the essential quality of weather data.
Furthermore, real-time data sharing and monitoring capabilities distinguish the OSWS
from many conventional systems. In sectors where swift responses to changing weather
conditions are paramount, the immediacy of data becomes indispensable. The potential
of OSWS to create a connected network of weather stations exponentially amplifies this
benefit, enhancing data accessibility and utility across regions.
To encapsulate the distinctions and similarities between the OSWS and other research
endeavors, the comparative Table 8 offers a concise quantitative perspective, aiding in
discerning the unique propositions the OSWS offers. This table shows the root mean square
error that each station has against a professional measuring tool in an indoor scenario and
whose values are considered as the reference.

Table 8. Comparative table between this OSWS and other projects in RMSE.

Atmospheric
Temperature
Humidity Pressure PM 1 RMSE PM 2.5 RMSE PM 10 RMSE
Reference RMSE
RMSE [%RH] RMSE [µg/m3 ] [µg/m3 ] [µg/m3 ]
[◦ C]
[hPa]
This OSWS 1.62 4 1 2.45 3.5 4.6
[20] 0.99 4.16 0.53 * * *
[29] 1.5 1.3 * * 11 *
[21] 1.37 * * * * *
[22,28] * * * * * *
* Indicates that the variable was not compared with a reference.

5.2. Impact
The revolutionary aspect of the OSWS is its democratization of weather data through
low-cost sensors, facilitating accessible and affordable weather monitoring. By enabling
real-time data sharing, the OSWS fosters timely decisions, which is crucial for sectors
sensitive to weather fluctuations. Apart from real-time weather monitoring, the broader
vision behind the OSWS extends to its application across diverse sectors like agriculture
and urban development. Its educational utility, offering a tangible tool for learners, further
underscores its multifaceted applicability.
The OSWS, with its potential for regional weather analysis and integration from
multiple units, lays the groundwork for enhanced predictive capabilities, with notable
relevance in sectors relying heavily on accurate weather forecasts.

5.3. Limitations and Future Directions

While the OSWS heralds a transformative shift in weather monitoring, certain limi-
tations prevail. These encompass the system’s precision, which, while consistent, might
be inherently limited due to its cost-effective approach. Another pivotal concern is data
transmission and security, especially in regions with unreliable internet connectivity.
Future endeavors related to the OSWS ought to emphasize:
1. Secure and dependable data transmission, particularly in remote settings.
2. ExplorING alternative data transmission channels and offline data logging/retrieval
solutions.
3. Strengthening data security measures to preserve data integrity during transmission
and storage.
4. Investigating durable sensors and materials to withstand environmental challenges,
reinforcing the OSWS’s reliability.
Sensors 2023, 23, 9060 24 of 26

6. Conclusions
The OSWS offers a groundbreaking approach to weather monitoring, harnessing low-
cost sensors to overcome challenges posed by high costs and the need for expert personnel
inherent in existing technologies. Its capacity for real-time monitoring and forecasting of
pivotal environmental variables using the ARIMA model stands utmost for agriculture,
transportation, and air quality monitoring. Its open-source nature further amplifies data
accessibility, fostering informed decisions essential in our current climate-changing era.
The design and implementation of the OSWS laid the groundwork for the future
evolution of cost-effective and efficient monitoring systems. As we move ahead, focusing
on enhancing data transmission security, diversifying communication interfaces, and
enduring diverse environmental conditions will be pivotal in refining the OSWS, aiming to
bolster environmental sustainability and foster informed decision-making amidst escalating
ecological challenges.

Author Contributions: Conceptualization, A.R., P.P., A.M. (Arturo Molina) and A.M. (Alan Meier);
Methodology, A.R., P.P. and O.M.; Software, A.R., P.P. and A.M. (Arturo Molina); Validation, A.R.,
P.P. and O.M.; Formal analysis, A.R., P.P., O.M., A.M. (Arturo Molina) and A.M. (Alan Meier);
Investigation, A.R., P.P., O.M. and A.M. (Alan Meier); Resources, P.P.; Data curation, A.R., P.P. and
A.M. (Arturo Molina); Writing—original draft, P.P. and A.M. (Alan Meier); Writing—review & editing,
A.R. and A.M. (Arturo Molina); Visualization, A.R.; Supervision, P.P., A.M. (Arturo Molina) and A.M.
(Alan Meier); Project administration, P.P.; Funding acquisition, A.M. (Arturo Molina). All authors
have read and agreed to the published version of the manuscript.
Funding: Institute of Advanced Materials for Sustainable Manufacturing, Tecnológico de Monterrey.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data are contained within the article.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Effects|Facts—Climate Change: Vital Signs of the Planet. Available online: https://climate.nasa.gov/effects/ (accessed on 2
October 2023).
2. United Nations. What Is Climate Change? Available online: https://www.un.org/en/climatechange/what-is-climate-change
(accessed on 2 October 2023).
3. Lai, C. How Does Deforestation Affect the Carbon Cycle? Earth.Org. Available online: https://earth.org/how-does-deforestation-
affect-the-carbon-cycle/ (accessed on 2 October 2023).
4. US EPA. Causes of Climate Change. Available online: https://www.epa.gov/climatechange-science/causes-climate-change
(accessed on 2 October 2023).
5. US EPA. Introduction to Indoor Air Quality. Available online: https://www.epa.gov/indoor-air-quality-iaq/introduction-
indoor-air-quality (accessed on 2 October 2023).
6. How Climate Change Affects Vector-Borne Diseases|News|Wellcome. Available online: https://wellcome.org/news/how-
climate-change-affects-vector-borne-diseases (accessed on 2 October 2023).
7. Kulkarni, M.A.; Duguay, C.; Ost, K. Charting the Evidence for Climate Change Impacts on the Global Spread of Malaria and
Dengue and Adaptive Responses: A Scoping Review of Reviews. Glob. Health 2022, 18, 1. [CrossRef] [PubMed]
8. Weather Forecast with Meteodrones Weather Drones|Meteomatics. Available online: https://www.meteomatics.com/en/
meteodrones-weather-drones/ (accessed on 2 October 2023).
9. Surface Mooring Network in the Kuroshio Extension. Available online: https://ieeexplore.ieee.org/document/4579297 (accessed
on 2 October 2023).
10. Demetillo, A.T.; Japitana, M.V.; Taboada, E.B. A System for Monitoring Water Quality in a Large Aquatic Area Using Wireless
Sensor Network Technology. Sustain. Environ. Res. 2019, 29, 12. [CrossRef]
11. AE521/AE521: What Is a Wireless Sensor Network? Available online: https://edis.ifas.ufl.edu/publication/AE521 (accessed on
2 October 2023).
12. Pule, M.; Yahya, A.; Chuma, J. Wireless Sensor Networks: A Survey on Monitoring Water Quality. J. Appl. Res. Technol. 2017, 15,
562–570. [CrossRef]
13. Newlands, N.K. Model-Based Forecasting of Agricultural Crop Disease Risk at the Regional Scale, Integrating Airborne Inoculum,
Environmental, and Satellite-Based Monitoring Data. Front. Environ. Sci. 2018, 6, 63. [CrossRef]
Sensors 2023, 23, 9060 25 of 26

14. World Meteorological Organization. Guide to Instruments and Methods of Observation (WMO-No. 8). Available online:
https://community.wmo.int/en/activity-areas/imop/wmo-no_8 (accessed on 2 October 2023).
15. Fixed Site RAWS (Remote Automated Weather Station). Earth Sciences. Available online: https://www.essearth.com/product/
fixed-site-raws-remote-automated-weather-station/ (accessed on 2 October 2023).
16. Huang, Z.-Q.; Chen, Y.-C.; Wen, C.-Y. Real-Time Weather Monitoring and Prediction Using City Buses and Machine Learning.
Sensors 2020, 20, 5173. [CrossRef] [PubMed]
17. Bencini, L.; Palma, D.D.; Collodi, G.; Manes, A.; Manes, G.; Bencini, L.; Palma, D.D.; Collodi, G.; Manes, A.; Manes, G. Wireless
Sensor Networks for On-Field Agricultural Management Process. In Wireless Sensor Networks: Application—Centric Design;
IntechOpen: London, UK, 2010. [CrossRef]
18. US EPA. Climate Change Impacts on Agriculture and Food Supply. Available online: https://www.epa.gov/climateimpacts/
climate-change-impacts-agriculture-and-food-supply (accessed on 2 October 2023).
19. Zhuang, Y.; Lin, F.; Yoo, E.-H.; Xu, W. AirSense: A Portable Context-Sensing Device for Personal Air Quality Monitoring.
In Proceedings of the 2015 Workshop on Pervasive Wireless Healthcare, Hangzhou, China, 22 June 2015; MobileHealth ’15;
Association for Computing Machinery: New York, NY, USA, 2015; pp. 17–22. [CrossRef]
20. Bernardes, G.F.L.R.; Ishibashi, R.; Ivo, A.A.S.; Rosset, V.; Kimura, B.Y.L. Prototyping Low-Cost Automatic Weather Stations for
Natural Disaster Monitoring. Digit. Commun. Netw. 2023, 9, 941–956. [CrossRef]
21. Ioannou, K.; Karampatzakis, D.; Amanatidis, P.; Aggelopoulos, V.; Karmiris, I. Low-Cost Automatic Weather Stations in the
Internet of Things. Information 2021, 12, 146. [CrossRef]
22. Mestre, G.; Ruano, A.; Duarte, H.; Silva, S.; Khosravani, H.; Pesteh, S.; Ferreira, P.M.; Horta, R. An Intelligent Weather Station.
Sensors 2015, 15, 31005–31022. [CrossRef] [PubMed]
23. Amazon.com: Weather Sensor All-in-One Compact Outdoor Weather Station Weather Sensor for Weather Monitoring Systems
(HM-FLP): Patio, Lawn & Garden. Available online: https://www.amazon.com/Weather-Compact-Outdoor-Station-Monitoring/
dp/B0BWNDPCBJ?th=1 (accessed on 3 October 2023).
24. Davis Wireless Vantage Vue Weather Station with WeatherLink Console 6242UK. Available online: https://www.weathershop.co.
uk/davis-vantage-vue-weather-station-with-weatherlink-console-6242uk (accessed on 3 October 2023).
25. Weather Station Professional—Renke. Available online: https://www.renkeer.com/product/weather-station-professional/
(accessed on 3 October 2023).
26. ATMOS 41. METER. Available online: https://www.metergroup.com/en/meter-environment/products/atmos-41-weather-
station?utm_source=google&utm_medium=cpc&utm_campaign=Paid_search&utm_term=weather_atmos41&utm_content=
product&gclid=Cj0KCQjw1OmoBhDXARIsAAAYGSHQJG6A4A5ZJ8gacSA_F7XrON4QgEzpX4AqcampxDPxEeJ53i-pyO8
aAh0oEALw_wcB (accessed on 3 October 2023).
27. Industrial Weather Stations|Columbia Weather Systems. Available online: https://columbiaweather.com/products/weather-
stations/industrial/?gclid=Cj0KCQjw1OmoBhDXARIsAAAYGSEiCJORy-r13IuDQUwAqDLli5dMtstQ9hrD0zG0rfOxK4
hQPnR47Q4aAu1qEALw_wcB (accessed on 3 October 2023).
28. Shindler, L. Development of a Low-Cost Sensing Platform for Air Quality Monitoring: Application in the City of Rome. Environ.
Technol. 2021, 42, 618–631. [CrossRef] [PubMed]
29. Jabbar, W.A.; Subramaniam, T.; Ong, A.E.; Shu’Ib, M.I.; Wu, W.; de Oliveira, M.A. LoRaWAN-Based IoT System Implementation
for Long-Range Outdoor Air Quality Monitoring. Internet Things 2022, 19, 100540. [CrossRef]
30. Bosch Sensortec. Humidity Sensor BME280. Available online: https://www.bosch-sensortec.com/products/environmental-
sensors/humidity-sensors-bme280/ (accessed on 26 August 2023).
31. Smart Sensor Solutions. Available online: https://www.sensirion.com/products/catalog/SCD30/ (accessed on 26 August 2023).
32. PMS5003 Series Manual Datasheet by Adafruit Industries LLC|Digi-Key Electronics. Available online: https://www.digikey.jp/
htmldatasheets/production/2903006/0/0/1/pms5003-series-manual.html (accessed on 26 August 2023).
33. Bonakdari, H.; Zeynoddin, M. Chapter 2—Preparation & Stationarizing. In Stochastic Modeling; Bonakdari, H., Zeynoddin, M.,
Eds.; Elsevier: Amsterdam, The Netherlands, 2022; pp. 13–87. [CrossRef]
34. Routledge & CRC Press. Time-Series Forecasting. Available online: https://www.routledge.com/Time-Series-Forecasting/
Chatfield/p/book/9781584880639 (accessed on 2 October 2023).
35. Monigatti, L. Interpreting ACF and PACF Plots for Time Series Forecasting. Medium. Available online: https://
towardsdatascience.com/interpreting-acf-and-pacf-plots-for-time-series-forecasting-af0d6db4061c (accessed on 2 Octo-
ber 2023).
36. Ali, S. (28) Reading the ACF and PACF Plots—The Missing Manual/Cheatsheet|LinkedIn. Available online: https://www.
linkedin.com/pulse/reading-acf-pacf-plots-missing-manual-cheatsheet-saqib-ali/ (accessed on 2 October 2023).
37. Routledge & CRC Press. Guide to Methodology in Ergonomics: Designing for Human Use, Second Edition. Available on-
line: https://www.routledge.com/Guide-to-Methodology-in-Ergonomics-Designing-for-Human-Use-Second-Edition/Stanton-
Young-Harvey/p/book/9781466591851 (accessed on 2 October 2023).
38. Handbook of Human Factors and Ergonomics; Salvendy, G., Ed; 9780470528389: Amazon.com: Books. Available online:
https://www.amazon.com/Handbook-Factors-Ergonomics-Gavriel-Salvendy/dp/0470528389 (accessed on 2 October 2023).
39. Nielsen Norman Group. World Leaders in Research-Based User Experience. Why You Only Need to Test with 5 Users. Available
online: https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/ (accessed on 2 October 2023).
Sensors 2023, 23, 9060 26 of 26

40. What Is Correlation?|Definition from TechTarget. WhatIs.com. Available online: https://www.techtarget.com/whatis/

definition/correlation (accessed on 2 October 2023).
41. Giavarina, D. Understanding Bland Altman Analysis. Biochem. Med. 2015, 25, 141–151. [CrossRef]
42. Schober, P.; Boer, C.; Schwarte, L.A. Correlation Coefficients: Appropriate Use and Interpretation. Anesth. Analg. 2018, 126, 1763.
[CrossRef] [PubMed]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.