Airborne particulate matter with an aerodynamic diameter of less than 2.5 & mu;m (PM2.5) is a major air pollutant worldwide. In Malaysia, transboundary 'haze' episodes with elevated PM2.5 concentrations linked to fires are common, causing health and economic harms. To reduce impacts, forecasting PM2.5 can enable effective PM2.5 management and decision-making. Until now, PM2.5 forecasts via a global mechanistic chemical transport model (CTM) have not been evaluated in the setting of Malaysia, where operational PM2.5 forecasting systems for preventive warnings are not yet deployed. Hence, this study aims to evaluate the performance of PM2.5 forecasts produced by a global CTM and to assess their suitability for use nation-wide in Malaysia. We used the surface PM2.5 forecasts from the Copernicus Atmosphere Monitoring Service's (CAMS) global atmospheric composition forecast dataset (CAMS-GACF) and evaluated them against hourly PM2.5 observations recorded throughout Malaysia from 2018 to 2020 via exceedance and accuracy analyses. We found that cycle 46r1 CAMS-GACF performance in Malaysia was generally weaker (critical success index (CSI) = 31%, R2 = 0.36) than reported in other studies (CSI = 20-54%, R2 = 0.32-0.79) focused on other countries, across multiple metrics in both analyses. We found CAMS-GACF did not accurately capture local-scale spatiotemporal variations in PM2.5 spatially and diurnally. However, we found CAMS-GACF captured better the increased regional PM2.5 pollution during the transboundary 'haze' episode of 2019. Based on our findings, we also propose recommendations on integrating CAMS-GACF in early-warning systems in Malaysia and on improving forecasts via bias-correction.
View source
