CSLTNet: A CNN-LSTM Dual-Branch Network for Particulate Matter Concentration Retrieval

The concentrations of atmospheric particulate matter (PM10 and PM2.5) significantly impact global environment, human health, and climate change. This study developed a particulate matter concentration retrieval method based on multi-source data, proposing a dual-branch retrieval network architecture named CSLTNet that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. The CNN branch is designed to extract spatial features, while the LSTM branch captures temporal characteristics, with attention modules incorporated into both the CNN and LSTM branches to enhance feature extraction capabilities. Notably, the model demonstrates robust spatial generalization capability across different geographical regions.Comprehensive experimental evaluations demonstrate the outstanding performance of the CSLTNet model. For the Beijing–Tianjin–Hebei region in China: in PM10 retrieval, sample-based 10-fold cross-validation achieved R2 = 0.9427 (RMSE = 16.47μg/m3), while station-based validation yielded R2 = 0.9213 (RMSE = 19.50μg/m3); for PM2.5 retrieval, sample-based 10-fold cross-validation resulted in R2 = 0.9579 (RMSE = 6.49μg/m3), with station-based validation reaching R2 = 0.9296 (RMSE = 8.32μg/m3). For Northwest China: in PM10 retrieval, sample-based 10-fold cross-validation achieved R2 = 0.9236 (RMSE = 34.52μg/m3), while station-based validation yielded R2 = 0.9046 (RMSE = 37.24μg/m3); for PM2.5 retrieval, sample-based 10-fold cross-validation resulted in R2 = 0.9279 (RMSE = 10.56μg/m3), with station-based validation reaching R2 = 0.8787 (RMSE = 13.71μg/m3).