Extension of AIRPACT5 forecasting to Day3 and evaluation of PM2.5 and ozone predictions across winter, spring and summer (2023–2024)
In this study, the forecast period of AIRPACT5 (Air Indicator Report for Public Awareness and Community Tracking) model was extended from two days to three days, and the accuracy of the third-day forecasts was subsequently evaluated. Two primary pollutants, PM 2.5 and O 3 , were considered for validation. The model's third-day forecasts were compared with observed data from AirNow monitoring stations, as well as with the model’s first and second-day forecasts. The evaluation covered a nine-month period from December 2023 to August 2024, encompassing the winter, spring, and summer seasons. Additionally, the model domain was categorized into urban, suburban and rural areas, and results were analyzed separately for each category. The comparison indicates that the performance trends of the third-day forecasts closely align with those of the first and second days. For PM 2.5 , model predictions were generally consistent with observational data, particularly in rural areas and across all seasons except during the wildfire season. In the case of O 3 , model performance was satisfactory in the summer but showed significant discrepancies in winter, especially in rural regions. • Day3 AIRPACT5 forecasts were generated and evaluated against observed PM 2 . 5 and O 3 data at categorized monitoring stations (urban, rural, and suburban). • AIRPACT5 performed best for PM 2 . 5 in rural areas during winter months. • Urban areas showed the lowest error values for O 3 across most months. • Mean Bias, Mean Error, and RMSE were used to assess model performance. • Seasonal and spatial variations significantly affect forecast accuracy.
Read more