Understanding global changes in fine-mode aerosols during 2008-2017 using statistical methods and deep learning approach.

Despite their extremely small size, fine-mode aerosols have significant impacts on the environment, climate, and human health. However, current understandings of global changes in fine-mode aerosols are limited. In this study, we employed newly developed satellite retrieval data and an attentive interpretable deep learning model to explore the status, changes, and association factors of the global fine-mode aerosol optical depth (fAOD) and aerosol fine-mode fraction (FMF) from 2008 to 2017. At the global scale, the results show a significant increasing trend in land FMF (2.34 × 10-3/year); however, the FMF over the ocean and the fAOD over land and ocean did not reveal significant trends. Between 2008 and 2017, high levels of both fAOD (>0.30) and FMF (>0.75) were identified over China, southeastern Asia, India, and Africa. Seasonally, global land FMF showed high values in summer (>0.70) and low values in spring (<0.65), while land fAOD was high in summer (>0.15) but low in winter (<0.13). Importantly, Australia and Mexico experienced significant increasing trends in FMF during all four seasons. At the regional scale, a significant decline in fAOD was identified in China, which indicates that government emission controls and reductions have been effective in recent decades. The deep learning model was used to interpret the result and showed that O3 was significantly associated with changes in both the FMF and fAOD. This finding suggests the importance of synergizing the regulations for both O3 and fine particles. Our work comprehensively examined global spatial and seasonal fAOD and FMF changes and provides a holistic understanding of global anthropogenic impacts.

A Spatial-Temporal Interpretable Deep Learning Model for improving interpretability and predictive accuracy of satellite-based PM2.5.

Being able to monitor PM2.5 across a range of scales is incredibly important for our ability to understand and counteract air pollution. Remote monitoring PM2.5 using satellite-based data would be incredibly advantageous to this effort, but current machine learning methods lack necessary interpretability and predictive accuracy. This study details the development of a new Spatial-Temporal Interpretable Deep Learning Model (SIDLM) to improve the interpretability and predictive accuracy of satellite-based PM2.5 measurements. In contrast to traditional deep learning models, the SIDLM is both "wide" and "deep." We comprehensively evaluated the proposed model in China using different input data (top-of-atmosphere (TOA) measurements-based and aerosol optical depth (AOD)-based, with or without meteorological data) and different spatial resolutions (10 km, 3 km, and 250 m). TOA-based SIDLM PM2.5 achieved the best predictive accuracy in China, with root-mean-square errors (RMSE) of 15.30 and 15.96 µg/m3, and R2 values of 0.70 and 0.66 for PM2.5 predictions at 10 km and 3 km spatial resolutions, respectively. Additionally, we tested the SIDLM in PM2.5 retrievals at a 250 m spatial resolution over Beijing, China (RMSE = 16.01 µg/m3, R2 = 0.62). Furthermore, SIDLM demonstrated higher accuracy than five machine learning inversion methods, and also outperformed them regarding feature extraction and the interpretability of its inversion results. In particular, modeling results indicated the strong influence of the Tongzhou district on the principle PM2.5 in the Beijing urban area. SIDLM-extracted temporal characteristics revealed that summer months (June-August) might have contributed less to PM2.5 concentrations, indicating the limited accumulation of PM2.5 in these months. Our study shows that SIDLM could become an important tool for other earth observation data in deep learning-based predictions and spatiotemporal analysis.