EXPONENTIAL MOVING AVERAGE FOR AIR POLLUTION DATA: ASSESSING ITS ROLE IN PM10 MONITORING ACCURACY

Abstract

Accurate estimation of particulate matter (PM₁₀) concentrations is critical for assessing air quality and mitigating public health risks. Traditional monitoring data processing methods, such as simple moving averages (MA), often struggle to capture rapid fluctuations in pollutant levels due to their uniform weighting of historical data, potentially compromising real-time decision-making. This study evaluates the efficiency of the Exponential Moving Average (EMA) algorithm, which prioritizes recent observations through exponential weighting, to improve PM₁₀ concentration estimates. Using data from urban air quality monitoring stations, EMA was applied across varying time windows and compared against conventional MA approaches. Performance was assessed against ground-truth measurements. Results demonstrated that EMA significantly reduced estimation errors. The algorithm exhibited enhanced responsiveness to abrupt PM₁₀ spikes, attributed to its dynamic weighting mechanism. Sensitivity analysis revealed that optimal smoothing factors depended on the selected time window, balancing noise reduction and trend detection. These findings underscore EMA’s potential as a robust tool for air pollution monitoring data analyses, offering superior adaptability to temporal variability. Implementation of EMA in regulatory and public health frameworks could enhance early warning systems and pollution control strategies. Future research should explore integrating EMA with machine learning models and low-cost sensor networks to further optimize real-time air quality management.