Trends and variation of PM2.5 in Antananarivo in the capital city of Madagascar

doi:10.21203/rs.3.rs-5289400/v1

Download PDF

Research Article

Trends and variation of PM2.5 in Antananarivo in the capital city of Madagascar

https://doi.org/10.21203/rs.3.rs-5289400/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

This study aims to evaluate the temporal variations of PM2.5 concentration , as well as its daily variation during the fire season in 2023 in tropical settings like Antananarivo Madagascar. PM2.5 concentrations were recorded in 8 localities using PurpleAir low-cost devices for one year. We then establish air quality patterns in conjunction with meteorological and environmental factors that may influence variations in PM2.5 concentrations. The annual PM2.5 concentrations ranged between 21.6 μg/m³-31.9 μg/m³, exceeding the World Health Organization annual threshold (5 μg/m³), where the daily PM2.5 concentrations were in the range 8.4 –130.42 μg/m³. During May-June the percentage of days where the air quality was unhealthy for sensitive groups is above 30%. It increased to 70% in October and November. A bimodal pattern with higher concentration during the morning (02:00 - 07:00 a.m) and evening (05:00-09:00 pm) were found. Dry weather, rainfall and atmospheric stability influenced the daily PM2.5 variations. Low wind speed and atmospheric stability favor its accumulation and increase the concentration of PM2.5.

PM2.5

air pollution

low-cost sensor

Madagascar

PM_2.5 are particles or droplets in the air that are 2.5 micrometers or smaller in size. It is generated from various sources (McDuffie et al., 2021) including road transport, industrial activities, wildfires, biomass burning, and natural sources such as dust storms (Reid et al., 2003), ocean, living vegetation. The source of PM_2.5 varies depending on the time and location, resulting in a diverse PM_2.5 composition. Depending on their origin, these particles can include organic and inorganic fraction and trace elements of heavy metal (Hao et al., 2019). PM_2.5 is widely considered one of the most extensively monitored air pollutants in the world due to the fact that it is the main contributor to the global burden of disease in the world (Lim et al., 2012). PM_2.5 is of particular concern due to its ability to penetrate deep into the lungs and even enter the bloodstream. Prolonged exposure to elevated levels of PM_2.5 has been linked to a variety of health problems, including respiratory and cardiovascular diseases (Manisalidis et al., 2020). WHO recommendations further restrict the maximum daily of 15 µg/m³ and annual thresholds of 5 µg/m³ for PM_2.5 (https://www.who.int/data/gho/data/themes/air-pollution).

Long term data monitoring and information on the variation of PM_2.5 are lacking in low income settings like Madagascar. This is due to limited financial and technological resources to invest in environmental monitoring, regulatory enforcement, and pollution control measures. Antananarivo, the capital city of Madagascar, represents a unique urban landscape where the intersection of rapid urbanization, industrial activities and means of transport, and diverse geographical features may contribute to a complex air quality dynamic. Due to limited financial resources and a lack of local vehicle production, Madagascar, relies on importing second-hand vehicles. However, these vehicles often lack modern emission control technologies, resulting in higher levels of harmful exhaust emissions (Ayetor et al., 2021a). Another significant contributor to emissions is the poor quality of the fuels used in the country (Ayetor et al., 2021b) combined with poor road infrastructure, vehicles traffic and congestion particularly at peak times. Air pollution is so severe that a layer of thick smog is often observed over the city. Additionally, open-air waste burning, brickmaking, charcoal-fueled garment factories and residential combustion have recently been identified as significant contributors to urban air pollution (Randrianarisoa et al., 2020). Traditional slash-and-burn farming for agriculture in the east and grassland clearing in the west are recurrent practices observed annually (Andriamanantena et al., 2021). These activities collectively impact the air quality in Antananarivo.

To address the gap in air quality knowledge and pollution sources, we aim to evaluate air quality patterns and the factors influencing pollution in the capital. We conducted a comprehensive analysis of PM_2.5 trends and spatial variations within Antananarivo. By examining these temporal patterns, we seek to provide a better understanding of the city's evolving environmental challenges, contributing valuable insights for informed policymaking and sustainable urban development.

2.1 Sampling sites description

Sampling sites are located within the urban center and the northern outskirts of the capital near or slightly further away from road traffic (Fig. 1). The locations were strategically chosen to represent PM_2.5 concentration behavior under various conditions, such as heavy traffic. They also help distinguish pollution sources, such as vehicle emissions, open burning, and other factors that may influence PM_2.5 concentrations.

2.2 Instrumentation

The study was conducted from January to December 2023. We used PurpleAir PA-II sensors, a low-cost sensor globally used for air quality monitoring. In 2019, a Met One Beta Attenuation Monitor 1020 (BAM-1020) was installed in the U.S. Embassy in Antananarivo. Prior dispatching the purple air data logger, they were calibrated by colocation near the BAM-1020.

The PurpleAir consists of two Plantower PMS5003 laser-scattering particle sensors, reporting real-time measurement of PM₁, PM_2.5, PM₁₀, and a photodiode particle counts in 6 bins from 0.3 to 10 µm. The two Plantower sensors, channels A and B sample at alternating intervals of 10s and provide 2min averaged data (Ardon-Dryer et al., 2020).

Data was downloaded using the PurpleAir Data Download Tool (https://community.purpleair.com/t/purpleair-data-download-tool/3787). The offline sensor data were manually downloaded from the SD cards connected to air sensors.

2.3 Correction factor and data curation

A comparison between the PurpleAir sensor and the BAM-1020 reference monitor (located at the US Embassy in Antananarivo) was conducted to calibrate and correct PurpleAir data. The analysis used hourly PM2.5 data alongside temperature and relative humidity as explanatory variables, collected from April to November 2023. Multiple Linear Regression, Polynomial Regression, XGBoost, and Random Forest models were tested to optimize the calibration.

The datasets were preprocessed to remove missing values, negative numbers, and extreme PM2.5 values above 1000 µg/m³. Measurements were also excluded if the relative humidity (RH) was higher than 100%. Model performance was evaluated using coefficient of determination (R²), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE) Eq. (1–3). The analysis used five-fold cross-validation, splitting the data with 80% for training and 20% for testing to ensure robust performance evaluation (Raheja et al., 2023; Westervelt et al., 2024).

For each valid and clean data point, the values recorded by channels A and B were averaged to ensure a more accurate representation of PM2.5 concentrations. The 2 min data were averaged up to hourly data, and hourly data were averaged to daily data using Eq. (4). Standard deviation was calculated to measure the variation of each PM_2.5 value from the average using Eq. (5). Due to measurement and transmission errors, some days' data were missing, so the dataset did not cover a full year. To avoid skewed results, thresholds were applied to the annual data, resulting in the retention of seven out of nine sensors (Fig. 1).

$$\:{R}^{2}\:=\:1\:-\:\frac{{\sum\:}_{i=1}^{n}{({y}_{i}-{\widehat{y}}_{i})}^{2}}{{\sum\:}_{i=1}^{n}{({y}_{i}-\underset{\_}{y})}^{2}}$$

$$\:MAE\:=\:\frac{1}{n}{\sum\:}_{i=1}^{n}{\left|{y}_{i}-{\widehat{y}}_{i}\right|}^{2}$$

$$\:RMSE\:=\:\sqrt{\frac{1}{n}{\sum\:}_{i=1}^{n}{({y}_{i}-{\widehat{y}}_{i})}^{2}}$$

Where:

n: number of data points

$\:{y}_{i}$ : observed values,

$\:{\widehat{y}}_{i}$ : predicted PM2.5 values,

$\:\underset{\_}{y}$ : mean of the observed PM2.5 concentration values

$$\:\underset{\_}{x}=\:\frac{1}{n}{\sum\:}_{i\:=1}^{n}{x}_{i}$$

$$\:\sigma\:\:=\:\sqrt{\frac{1}{n}{\sum\:}_{i\:=\:1}^{n}({x}_{i}-\:\underset{\_}{x}){}^{2}}$$

Where:

n : number of data points

$\:{x}_{i}$ : each individual PM_2.5 concentration value

$\:\underset{\_}{x}$ : mean of the PM_2.5 concentration values

2.4 Pattern and drivers of PM_2.5 concentration

Wind speed and direction and precipitation data were provided by the Direction Générale de la Météorologie Madagascar. Relative humidity and temperature were obtained from the purple air sensor.

To understand the relationship between daily PM2.5 concentration and meteorological factors, we calculated correlation coefficients and the Pearson correlation coefficient between daily PM2.5 and temperatures, relative humidity, wind speed, and wind direction. Precipitation was excluded from this analysis due to high frequency in the summer only. The Pearson correlation coefficient, calculated using Eq. (6), measures the strength and direction of the linear relationship between two variables ($\:{x}_{i\:}$ and $\:{y}_{i}$), where $\:\underset{\_}{x}$ and $\:\underset{\_}{y}$ are the average values of these variables, and n is the number of observations. Values of Pearson correlation coefficients range from − 1 to + 1, with higher absolute values indicating a stronger relationship. A positive value signifies a positive correlation, a negative value indicates a negative correlation, and 0 means no correlation.

$$\:{r}_{xy}=\:\frac{{\sum\:}_{i=1}^{n}\:({x}_{i}-\underset{\_}{x})({y}_{i}-\underset{\_}{y})}{\sqrt{{{\sum\:}_{i=1}^{n}({x}_{i}-\underset{\_}{x})}^{2}}\sqrt{{{\sum\:}_{i=1}^{n}({y}_{i}-\underset{\_}{y})}^{2}}}$$

Where n = number of data points

$\:{x}_{i}$ = each individual PM_2.5 concentration value

$\:{y}_{i}$ = each individual meteorological factor value

$\:\underset{\_}{x}$ = mean of PM_2.5 concentration values

$\:\underset{\_}{y}$ = mean of meteorological factor values

Active fire data were also used to assess the impact of fires on PM_2.5 concentrations. We chose the Visible Infrared Imaging Radiometer Suite (VIIRS) 375m active fire products because they provide global fire and thermal anomaly information twice a day with a spatial resolution of 375 meters. The VIIRS i4 band (3.74 µm) is particularly sensitive to thermal radiation from fires. By comparing the brightness temperature of the i4 band with other bands, the instrument can identify pixels likely containing active fires (Schroeder et al., 2014). We used active fire products VNP14IMGTDL (Suomi-NPP) and VJ114IMGTDL (NOAA-20), which provide data on the location, confidence, and intensity of fires worldwide and can be downloaded from the archive https://firms.modaps.eosdis.nasa.gov/download/ (last access, December 2023).

We then use the pollution rose, which graphically represents the frequency and intensity of pollutant concentrations based on wind direction and speed. By analyzing the pollution rose, we can determine the predominant wind directions dispersing pollutants in the study area, aiding in the identification of air pollution sources.

All data were analyzed using Python scripts for calculations and plots. WRPlot tool was used for the pollution rose.

3.1 PurpleAir sensor (PA) data correction factors

The PurpleAir sensor shows a moderate correlation with reference PM2.5 concentrations (r = 0.56) under varying humidity and temperature. At lower humidity levels, PurpleAir sensors tend to underestimate PM_2.5 and higher temperatures tend to cluster near the red dashed line, indicating better agreement between the sensor measurements in warm conditions. Humidity and temperature affect the spread of data points, though the overall correlation remains consistent. While these factors do not significantly affect the overall correlation, the variability in sensor performance suggests that humidity and temperature contribute to deviations from the reference PM2.5 concentrations. The MAE of 13.69 µg/m³ and a positive bias of 1.19 µg/m³ indicate a slight overestimation.

We compared the performance of four models in predicting PM2.5 concentrations using three key metrics: MAE, RMSE, and R². A lower MAE and RMSE along with a higher R² suggest a more accurate model in predicting PM2.5 concentrations. Selection of optimal model hyperparameters was conducted using parameter grid search and 5-fold cross-validation. The Polynomial Regression model has the best performance (MAE = 10.8 µg/m³, RMSE = 15.27 µg/m³, and R² = 0.35), explaining 35% of the variance in PM_2.5 level and underscoring its precision and reduced deviation from observed values. In contrast, the MLR correction model shows slightly lower performance (MAE = 11.31 µg/m³, RMSE = 15.65 µg/m³, and R² = 0.32), explaining 32% of the variance and showing that the model has higher error in its predictions. Random Forest shows low performance (MAE = 12.2 µg/m³, RMSE = 16.1 µg/m³, and R² = 0.30) while XGBoost performs the lowest (MAE = 13.5 µg/m³, RMSE = 17.2 µg/m³, and R² = 0.28). These results indicate that the Polynomial Regression model is more accurate and explains more variability in PM2.5. Overall, 65% − 75% of the variability in PM2.5 is not captured by the models, suggesting that important factors might be missing. The residual variability not captured by the models indicates that there are limitations in sensor’s accuracy and model design. In the city of Antananarivo, the influence of additional factors such as emission sources (vehicle, fire smoke), and other meteorological variables (wind pattern, pressure) may be very important but not included in this analysis. Incorporating those factors into the models could provide a more comprehensive understanding and yield higher predictive accuracy.

3.2 Temporal and spatial variations of PM_2.5 concentrations

The annual average of PM_2.5 concentrations for each sensor located at Ambatobe, Ampandrianomby, Andraharo, Antsakaviro, Soanierana, Ambohidahy, and Miandrarivo during the year 2023 are: 21.6µg/m³, 24.5µg/m³, 26.2µg/m³, 28.4µg/m³, 28.7µg/m³, 29.3µg/m³, and 31.9µg/m3 respectively. Furthermore, Ambatobe and Ampandrianomby are the least polluted outskirts while Soanierana, Ambohidahy, and notably Miandrarivo are the most polluted. These differences found for each measurement point could be due to the location of the sensors, which are either close to areas of dense circulation or in a slightly more remote and less polluted area. Each of these annual averages far exceeds the WHO’s 5 µg/m3 annual air quality standards. During the year 2023, 83% of the days exceed the WHO’s daily threshold standards (15 µg/m3). During the period from September to November, only 11 days are below this threshold. The maximum of the daily averages is observed on October 11, which has a value of 130.42 µg/m3, while the minimum is on January 15, which is 8.4 µg/m3 (Fig. 3).

In Fig. 3, each bar represents the standard deviation calculated from daily averages. A clear seasonal pattern in the standard deviation can be observed, with larger variances occurring between October and November, as well as May and July, and smaller variances during the remaining months. PM_2.5 levels exhibited higher standard deviations when daily averages were high, and lower standard deviations when daily averages were lower.

Figure 4 shows the monthly average concentrations of PM_2.5 for each sensor. Overall, the monthly averages exceed the daily WHO threshold (15 µg/m³ ) throughout the month.There are fluctuations in the average values for each month from January to December. Two peaks were observed, the first in May-June (winter) and the second in October-November.

Due to poor road infrastructure and limited public transportation, traffic congestion is common in Antananarivo's city center. Many older diesel vehicles remain in use, contributing to incomplete fuel combustion and higher PM_2.5 emissions. Additionally, the growing use of motorcycles, which also produce emissions, increases the pollution problem, especially as heavy traffic persists. Meteorological conditions such as weak wind speeds, a reduced boundary layer height, and increased temperature inversions are more frequent during the winter. These temperature inversions in the lower troposphere enhance atmospheric stability and inhibit the vertical dispersion of pollutants, creating favorable conditions for the accumulation of pollutants and leading to higher concentrations of PM_2.5 (Bai et al., 2023; Liu et al., 2019).

Figure 5 shows the monthly air quality classification level for 2023 according to WHO’s guideline, averaged across all sensors, with detailed breakdowns for each sensor shown in Fig. S1. Overall, Antananarivo experienced Moderate air quality throughout the year. January had the best air quality, with 12.9% of days classified as Good and 87.1% as Moderate. However, the air quality worsened during certain months, with Unhealthy for Sensitive Groups and Unhealthy categories accounting for over 30% of days in May and June, and reaching about 70% and 40%, respectively, in October and November (Fig. 4). Air pollution levels varied significantly between locations. Some areas maintained lower pollution levels, while others experienced very Unhealthy conditions during certain months. For example, a location near a tunnel exhibited high pollution early in the year, with 60% of days classified as Unhealthy for Sensitive Groups and 40% as Unhealthy (Fig. S1).

3.3 Diurnal variations of PM_2.5 concentrations

Figure 6 shows the overall diurnal variations of PM_2.5 concentration based, showcasing the consistent bimodal pattern. We analyzed PM_2.5 concentrations every one trimester: December-February, March-May, and June-August and September-November. In all seasons, particle mass concentrations exhibit a bimodal pattern, with notably higher levels in the mornings (around 2:00 am − 7:00 am) and evenings (around 5:00 pm − 9:00 pm). The amplitude and width of PM_2.5 are most pronounced during September-November. Air pollution in Antananarivo is at its worst during this period throughout the day, consistently surpassing the World Health Organization's 24-hour recommendation for air quality, which suggests levels below 15µg/m³.

3.4 Impact of fires, rainfall and wind direction on air quality during September and November

Figure 7A represents a zoomed-in temporal distribution of daily averages of PM_2.5 concentrations from September to November in 2023. Air quality reaches its worst levels between September and November, primarily due to the rise in forest fires during the dry season, which begins in July and August and peaks in September and October (Andriamanantena et al., 2021).

Figure 7B illustrates fire points detected by VIIRS SNPP and NOAA-20 and Fig. 7C presents daily precipitation data from the national weather station. In September, extensive fire activity was observed outside of Antananarivo, but PM_2.5 concentrations remained relatively low. A notable decrease in both fire activity and PM_2.5 levels occurred in early October, coinciding with the start of the rainy season in Antananarivo. However, in mid-November, a sharp drop in both fire incidents and PM_2.5 concentrations were observed, followed by a significant pollution episode.

On October 11, PM_2.5 concentrations ranged from 98 µg/m³ to 163 µg/m³, and on November 21, from 70 µg/m³ to 165 µg/m³. Fire points recorded in October and November were 3,239 and 2,470 for SNPP, and 4,128 and 4,727 for NOAA-20, respectively, with differences likely attributed to satellite overpass timings or cloud cover. The decrease in fire points may also be associated with precipitation events.

Robinson (2021) similarly reported a direct correlation between increased fire activity and elevated PM_2.5 during pollution episodes. Rainfall reduces atmospheric PM_2.5 concentration through collision-coalescence processes, where raindrops capture pollutants. However, on October 23, despite 38 mm of rainfall, PM_2.5 concentrations increased, possibly due to substantial fire activity the previous day, with smoke transported to Antananarivo after the rain had temporarily cleared the air. This highlights the complexity of fire-related pollution events, where localized meteorological factors and smoke transport can significantly influence air quality.

Pollution rose plots (Fig. 8) for September, October, and November highlight the critical role of wind direction in dispersing PM_2.5. The highest concentrations were linked to south-easterly winds, though October showed mixed wind directions, resulting in pollution peaks from multiple directions. Notably, PM_2.5 contributions were highest from the southeast, northwest, and northeast. This period coincides with widespread fires across Madagascar, particularly in the eastern regions due to slash-and-burn agriculture (“tavy”), which generate PM_2.5-laden smoke transported by trade winds (Andriamanantena et al., 2021).

The convergence of pollutants from distant fires and local sources, such as traffic emissions, contributed to elevated PM_2.5 levels. These findings underscore the significant influence of wind direction in transporting pollutants and confirm the relationship between fire activity and deteriorating air quality in Antananarivo (Fig. 7A and 7B).

In Antananarivo, the significant increase in air pollution during the hot period of October and November can be attributed to a combination of local and regional factors. Local emissions from traffic, industries, and urbanization activities, including the burning of waste and the seasonal production of bricks, contribute to elevated pollution levels. During this period, brick kilns are highly active, releasing substantial amounts of particulate matter and other pollutants. Long-range transport of smoke from widespread bush and forest fires, which are common during this season, significantly increases air quality issues. These fires, driven by agricultural practices such as slash-and-burn farming (also known as "tavy") and deforestation, result in the emission of large quantities of smoke and fine particulate matter (PM_2.5) that can travel across large distances, impacting urban areas. The combination of local pollution sources and regional smoke transport creates a situation of poor air quality, especially as meteorological conditions such as weak wind speeds and temperature inversions and rainfall delay further trap pollutants near the surface.

3.5 Analysis of the impact of meteorological parameters on PM_2.5

Figure 9 shows a weak positive correlation between PM_2.5 and temperature (+ 0.21). This suggests that temperatures may weakly lead to an increasing PM_2.5 concentration. This pattern can be likely attributed to smoke transport during the warm, dry season, and the impact of temperature inversions, which slows PM_2.5 dispersion. Temperature inversions limit vertical mixing and trap pollutants closer to the surface. Additionally, higher temperatures can enhance photochemical reactions, producing more precursors of PM_2.5 and other secondary pollutants, thus driving an increase in PM_2.5 concentrations (Chen et al., 2020; Zhang et al., 2015).

High concentrations of PM_2.5 may also influence temperature in the opposite way by scattering solar radiation, and therefore reduce surface temperatures (Zhong et al., 2018). This reduction in near-ground temperature weakens atmospheric mixing, allowing further accumulation of PM_2.5. During pollution episodes with temperature inversion layers, the cooling of the near-ground atmosphere and the stability of the inversion enhance PM_2.5 accumulation (Zhu et al., 2018).

Negative correlations were found for relative humidity (-0.46) and wind speed (-0.29). With an increasing relative humidity PM_2.5 concentrations decrease. When humidity surpasses 70%, suspended particles aggregate and fall out of the atmosphere through dry and wet deposition, leading to a significant reduction in PM2.5 levels (Wang and Ogawa, 2015; Li et al., 2015). Higher wind speeds promote the horizontal and vertical diffusion of PM2.5. Conversely, low wind speeds inhibit dispersion and promote the accumulation of PM2.5 near the surface. These findings are consistent with measurements taken in Antananarivo’s less polluted areas (INSTN, 2021), which also showed an inverse relationship between PM2.5 and both humidity and wind speed.

Here, we use PurpleAir low-cost air quality sensors to address gaps in air quality knowledge and pollution sources, using Antananarivo as a pilot study, and to develop collaborative monitoring tools. PurpleAir has been widely adopted due to its affordability and its capacity for real-time data monitoring, making this information accessible to a larger user community. This study provides a comprehensive analysis of the temporal variations of PM_2.5 concentrations in Antananarivo, Madagascar, throughout 2023, with a focus on the fire season's impact.

The PM_2.5 concentrations significantly exceeded the World Health Organization's guidelines, indicating severe air pollution throughout the year, particularly from September to November. During this period, PM_2.5 concentrations spiked during late afternoons and evenings, worsened by factors such as dry weather, low wind speeds, and atmospheric stability, all of which contributed to the accumulation of pollutants. In addition to local sources like vehicular emissions, industrial activities, and biomass burning, long-range transport of smoke from regional fires further deteriorated air quality.

In Antananarivo, temperature, humidity, and wind speed significantly affect PM2.5 concentrations. Higher temperatures tend to increase PM2.5 levels, while higher humidity and wind speeds reduce concentrations through deposition and diffusion processes. Understanding these dynamics is crucial for managing air quality in the city.

However, frequent power cuts may compromise data quality and significantly impact analyses of air quality in the city. Therefore, for future research we suggest focusing on refining model inputs, exploring advanced predictive algorithms, and enhancing sensor calibration techniques to improve the accuracy and reliability of air quality assessments in the capital. This also highlights the need for more robust air quality sensors that do not rely entirely on electricity.

The alarming levels of PM_2.5 observed in Antananarivo highlight the urgent need for targeted policies and interventions to mitigate air pollution in the city. By addressing local emissions, particularly from vehicular and industrial sources, and implementing strategies to minimize the impact of regional fires, substantial improvements in air quality can be achieved. This study's findings provide essential insights for policymakers and stakeholders, emphasizing the critical role of continuous monitoring and the development of sustainable urban planning to protect public health and the environment.

We recommend enhancing citizen science initiatives and raising awareness of the health impacts of air pollution to empower communities and foster collective action toward improved air quality.

Availability of data and materials

PM 2.5 from Purple air quality sensors are available at https://community.purpleair.com/t/purpleair-data-download-tool/3787 and https://github.com/heleneandria/AirQuality_Madagascar/tree/main/Andriamamonjy%20et%20al_Aerosol%20and%20Air%20Quality. Beta Attenuation Monitor (BAM) on PM2.5 data are available at the AirNow website (https://www.airnow.gov/international/us-embassies-and-consulates/#Madagascar$Antananarivo. Active fire data are available at https://firms.modaps.eosdis.nasa.gov/download/

Meteorological Data (precipitation and wind speed and direction) are “available on request”

Competing interests

The authors declare no competing interests

Funding

Regional Environment Officers (REO) small grants program from the U.S. Department of State

Authors' contributions (individual contributions)

MHA is the lead author of this paper. MHA is a PhD student from the University of Antananarivo.

MHA and LM was responsible for conceptualizing the study, performing data analysis, and drafting the initial manuscript.

LG and ZR provided essential contributions by verifying the facts presented in the paper. As experts in air pollution, they offered significant insights that helped refine the manuscript.

LM supervised the overall project and approved the final version.

ACKNOWLEDGMENTS

MHA is a PhD student at the University of Antananarivo. LM is a research associate at the Royal Botanic Garden Kew and the Association Vahatra in Madagascar. LM participation in this research is supported by the Jennifer Ward Oppenheimer Research Grant. This study is further supported by the Regional Environment Officers (REO) small grants program from the U.S. Department of State. The authors express their gratitude to Mrs. Manitra Dablin, Dr. Fidy Rasambainarivo, Sitraka Andriamiadana, and the citizen scientists who allowed us to install Purple air quality sensors at their homes. We also thank the NGO Association Vahatra and the meteorological service for facilitating manual data collection from their PA air quality sensors, as well as the Direction Générale de la Météorologie Madagascar for providing climatic data. Additionally, we appreciate the support from the U.S. Embassy and its environmental officer for enabling us to calibrate the air quality sensors at the embassy and for their active involvement in this research.

Andriamamonjy, M.H. (2022). Étude de la distribution des aérosols, liée aux feux de la biomasse à Madagascar par observations satellitaires (Master). Ecole Supérieure Polytechnique Antananarivo, Antananarivo. Madagascar.
Andriamanantena, H., Rakotondraompiana, S., Rakotoniaina, S., Razanaka, S. (2021). Répartitions spatiale et temporelle des feux à Madagascar. Rev. Fr. Photogrammétrie Télédétection 223, 38–58. https://doi.org/10.52638/rfpt.2021.520
Ardon-Dryer, K., Dryer, Y., Williams, J.N., Moghimi, N. (2020). Measurements of PM2.5; with PurpleAir under atmospheric conditions. Atmospheric Meas. Tech. 13, 5441–5458. https://doi.org/10.5194/amt–13-5441-2020
Ayetor, G.K., Mbonigaba, I., Ampofo, J., Sunnu, A. (2021a). Investigating the state of road vehicle emissions in Africa: A case study of Ghana and Rwanda. Transp. Res. Interdiscip. Perspect. 11, 100409. https://doi.org/10.1016/j.trip.2021.100409
Ayetor, G.K., Mbonigaba, I., Sackey, M.N., Andoh, P.Y. (2021b). Vehicle regulations in Africa: Impact on used vehicle import and new vehicle sales. Transp. Res. Interdiscip. Perspect. 10, 100384. https://doi.org/10.1016/j.trip.2021.100384
Bai, D., Liu, L., Dong, Z., Ma, K., Huo, Y. (2023). Variations and possible causes of the December PM2.5 in Eastern China during 2000–2020. Front. Environ. Sci. 11, 1134940. https://doi.org/10.3389/fenvs.2023.1134940
Chen, Z., Chen, D., Zhao, C., Kwan, M., Cai, J., Zhuang, Y., Zhao, B., Wang, X., Chen, B., Yang, J., Li, R., He, B., Gao, B., Wang, K., Xu, B. (2020). Influence of meteorological conditions on PM2.5 concentrations across China: A review of methodology and mechanism. Environ. Int. 139, 105558. https://doi.org/10.1016/j.envint.2020.105558
Hao, Y., Gao, C., Deng, S., Yuan, M., Song, W., Lu, Z., Qiu, Z. (2019). Chemical characterisation of PM2.5 emitted from motor vehicles powered by diesel, gasoline, natural gas and methanol fuel. Sci. Total Environ. 674, 128–139. https://doi.org/10.1016/j.scitotenv.2019.03.410
INSTN-Newsletter N°2. (2021). Nouvelle Rocade d’IARIVO - pollution de l’air. NEWSLETTER2.pdf (instn.mg)
Li, J., Chen, H., Li, Z., Wang, P., Cribb, M., Fan, X. (2015). Low-level temperature inversions and their effect on aerosol condensation nuclei concentrations under different large-scale synoptic circulations. Adv. Atmospheric Sci. 32, 898–908. https://doi.org/10.1007/s00376-014-4150-z
Lim, S.S., Vos, T., Flaxman, A.D., Danaei, G., Shibuya, K., Adair-Rohani, H., AlMazroa, M.A., Amann, M., Anderson, H.R., Andrews, K.G., Aryee, M., Atkinson, C., Bacchus, L.J., Bahalim, A.N., Balakrishnan, K., Balmes, J., Barker-Collo, S., Baxter, A., Bell, M.L., Blore, J.D., et al. (2012). A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet 380, 2224–2260. https://doi.org/10.1016/S0140–6736(12)61766–8
Liu, Z., Wang, H., Shen, X., Peng, Y., Shi, Y., Che, H., Wang, G. (2019). Contribution of Meteorological Conditions to the Variation in Winter PM2.5 Concentrations from 2013 to 2019 in Middle-Eastern China. Atmosphere 10, 563. https://doi.org/10.3390/atmos10100563
Manisalidis, I., Stavropoulou, E., Stavropoulos, A., Bezirtzoglou, E. (2020). Environmental and Health Impacts of Air Pollution: A Review. Front. Public Health 8, 14. https://doi.org/10.3389/fpubh.2020.00014
McDuffie, E.E., Martin, R.V., Spadaro, J.V., Burnett, R., Smith, S.J., O’Rourke, P., Hammer, M.S., Van Donkelaar, A., Bindle, L., Shah, V., Jaeglé, L., Luo, G., Yu, F., Adeniran, J.A., Lin, J., Brauer, M. (2021). Source sector and fuel contributions to ambient PM2.5 and attributable mortality across multiple spatial scales. Nat. Commun. 12, 3594. https://doi.org/10.1038/s41467-021-23853-y
Raheja, G., Nimo, J., Appoh, E.K.-E., Essien, B., Sunu, M., Nyante, J., Amegah, M., Quansah, R., Arku, R.E., Penn, S.L., Giordano, M.R., Zheng, Z., Jack, D., Chillrud, S., Amegah, K., Subramanian, R., Pinder, R., Appah-Sampong, E., Tetteh, E.N., Borketey, M.A., et al. (2023). Low-Cost Sensor Performance Intercomparison, Correction Factor Development, and 2 + Years of Ambient PM 2.5 Monitoring in Accra, Ghana. Environ. Sci. Technol. 57, 10708–10720. https://doi.org/10.1021/acs.est.2c09264
Randrianarisoa, J., Rajaonarintsoa, N.A., Perego, P., Silva, A.L., Siebert, F.W. (2020). Road safety in Antananarivo, capital of Madagascar: a naturalistic observation study of road users and motorcycle helmet use rates. https://doi.org/10.13140/RG.2.2.16465.35682
Reid, J.S., Jonsson, H.H., Maring, H.B., Smirnov, A., Savoie, D.L., Cliff, S.S., Reid, E.A., Livingston, J.M., Meier, M.M., Dubovik, O., Tsay, S. (2003). Comparison of size and morphological measurements of coarse mode dust particles from Africa. J. Geophys. Res. Atmospheres 108, 2002JD002485. https://doi.org/10.1029/2002JD002485
Robinson, J. (2021). Identification des conditions météorologiques et environnementales liées à l’apparition d’épisodes de pollution aux particules PM2.5: Cas de novembre 2020 (Master). Ecole Supérieure Polytechnique Antananarivo, Antananarivo, Madagascar.
Schroeder, W., Oliva, P., Giglio, L., Csiszar, I.A. (2014). The New VIIRS 375 m active fire detection data product: Algorithm description and initial assessment. Remote Sens. Environ. 143, 85–96. https://doi.org/10.1016/j.rse.2013.12.008
Wang, J., Ogawa, S. (2015). Effects of Meteorological Conditions on PM2.5 Concentrations in Nagasaki, Japan. Int. J. Environ. Res. Public. Health 12, 9089–9101. https://doi.org/10.3390/ijerph120809089
Zhang, H., Wang, Y., Hu, J., Ying, Q., Hu, X.-M. (2015). Relationships between meteorological parameters and criteria air pollutants in three megacities in China. Environ. Res. 140, 242–254. https://doi.org/10.1016/j.envres.2015.04.004
Zhong, J., Zhang, X., Dong, Y., Wang, Y., Liu, C., Wang, J., Zhang, Y., Che, H. (2018). Feedback effects of boundary-layer meteorological factors on cumulative explosive growth of PM2.5; during winter heavy pollution episodes in Beijing from 2013 to 2016. Atmospheric Chem. Phys. 18, 247–258. https://doi.org/10.5194/acp–18-247-2018
Zhu, W., Xu, X., Zheng, J., Yan, P., Wang, Y., Cai, W. (2018). The characteristics of abnormal wintertime pollution events in the Jing-Jin-Ji region and its relationships with meteorological factors. Sci. Total Environ. 626, 887–898. https://doi.org/10.1016/j.scitotenv.2018.01.083
Westervelt, D.M., Isevulambire, P.K., Yombo Phaka, R., Yang, L.H., Raheja, G., Milly, G., Selenge, J.-L.B., Mulumba, J.P.M., Bousiotis, D., Djibi, B.L., McNeill, V.F., Ng, N.L., Pope, F., Mbela, G.K., Konde, J.N. (2024). Low-Cost Investigation into Sources of PM 2.5 in Kinshasa, Democratic Republic of the Congo. ACS EST Air 1, 43–51. https://doi.org/10.1021/acsestair.3c00024

No competing interests reported.

AndriamamonjyetalAerosol2024SupMat.docx

Download PDF

Reviews received at journal
26 Nov, 2024
Reviewers agreed at journal
21 Nov, 2024
Reviewers agreed at journal
20 Nov, 2024
Reviewers agreed at journal
19 Nov, 2024
Reviewers invited by journal
19 Nov, 2024
Editor assigned by journal
22 Oct, 2024
Submission checks completed at journal
22 Oct, 2024
First submitted to journal
18 Oct, 2024

You are reading this latest preprint version

Trends and variation of PM2.5 in Antananarivo in the capital city of Madagascar

Status:

Version 1

Abstract

Figures

1 INTRODUCTION

2 METHODS

2.1 Sampling sites description

2.2 Instrumentation

2.3 Correction factor and data curation

2.4 Pattern and drivers of PM_2.5 concentration

3 RESULTS AND DISCUSSIONS

3.1 PurpleAir sensor (PA) data correction factors

3.2 Temporal and spatial variations of PM_2.5 concentrations

3.3 Diurnal variations of PM_2.5 concentrations

3.4 Impact of fires, rainfall and wind direction on air quality during September and November

3.5 Analysis of the impact of meteorological parameters on PM_2.5

4 CONCLUSION

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1

Trends and variation of PM2.5 in Antananarivo in the capital city of Madagascar

Status:

Version 1

Abstract

Figures

1 INTRODUCTION

2 METHODS

2.1 Sampling sites description

2.2 Instrumentation

2.3 Correction factor and data curation

2.4 Pattern and drivers of PM2.5 concentration

3 RESULTS AND DISCUSSIONS

3.1 PurpleAir sensor (PA) data correction factors

3.2 Temporal and spatial variations of PM2.5 concentrations

3.3 Diurnal variations of PM2.5 concentrations

3.4 Impact of fires, rainfall and wind direction on air quality during September and November

3.5 Analysis of the impact of meteorological parameters on PM2.5

4 CONCLUSION

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1

2.4 Pattern and drivers of PM_2.5 concentration

3.2 Temporal and spatial variations of PM_2.5 concentrations

3.3 Diurnal variations of PM_2.5 concentrations

3.5 Analysis of the impact of meteorological parameters on PM_2.5