Comparing Methods to Impute Missing Daily Ground-Level PM10 Concentrations between 2010–2017 in South Africa

Abstract

Good quality and completeness of ambient air quality monitoring data is central in supporting actions towards mitigating the impact of ambient air pollution. In South Africa, however, availability of continuous ground-level air pollution monitoring data is scarce and incomplete. To address this issue, we developed and compared different modeling approaches to impute missing daily average particulate matter (PM10) data between 2010 and 2017 using spatiotemporal predictor variables. The random forest (RF) machine learning method was used to explore the relationship between average daily PM10 concentrations and spatiotemporal predictors like meteorological, land use and source-related variables. National (8 models), provincial (32) and site-specific (44) RF models were developed to impute missing daily PM10 data. The annual national, provincial and site-specific RF cross-validation (CV) models explained on average 78%, 70% and 55% of ground-level PM10 concentrations, respectively. The spatial components of the national and provincial CV RF models explained on average 22% and 48%, while the temporal components of the national, provincial and site-specific CV RF models explained on average 78%, 68% and 57% of ground-level PM10 concentrations, respectively. This study demonstrates a feasible approach based on RF to impute missing measurement data in areas where data collection is sparse and incomplete.

Short-Term Joint Effects of PM10, NO2 and SO2 on Cardio-Respiratory Disease Hospital Admissions in Cape Town, South Africa

“…A different approach was applied in the prediction of missing PM 10 data in a different study using monitoring stations from Cape Town. It was reported that models for each site performed better in capturing the variability of PM 10 concentration [ 16 ].…”

Spatial and Temporal Variations in PM10 Concentrations between 2010–2017 in South Africa

“…SAAQIS can be reached via their website (https://saaqis.environment.gov.za/, accessed on 22 October 2018). For our study, we selected, for each year between 2010 and 2017, all sites with more than or equal to 70% of total daily measurement data available during a year [17]. Missing data were imputed using a random forest machine learning method, including spatiotemporal predictors, like meteorological, land use and source-related variables, as described in detail in our previous paper [17].…”

“…Our previous study described the quality of ground-level PM 10 measurements in four provinces of South Africa, Gauteng, Mpumalanga, Western Cape and KwaZulu-Natal, for the years 2010-2017 [17]. The earlier study explored methods to bridge the gap in daily PM 10 data by imputing missing daily PM 10 for some sites in these provinces for the study period.…”

Research trends in the field of ambient air quality monitoring and management in South Africa: A bibliometric review

“…The findings were that this approach is a feasible approach to compensate for missing data. ( Arowosegbe et al., 2021 ) Health Risk Assessment of PM 2.5 and PM 2.5 -Bound Trace Elements in Thohoyandou, South Africa The study investigated the health risks of fine particulate matter grade PM 2.5 in ambient air and its trace elemental components in a rural community of South Africa. The study established that health risks exposure was high particularly for children and infants.…”