Data Integration for ML-CNPM2.5: A Public Sample Dataset Based on Machine Learning Models and Remote Sensing Technology Applied for Estimating Ground-Level PM2.5 in China

Ambient fine particulate matter (PM2.5) has significant adverse effects on human health, thereby urgent hunger for accurate monitoring of ground-level PM2.5, especially its space distribution. Since satellites can observe the Earth on a large spatial scale, remote sensing technology can be applied to estimate PM2.5 concentrations at the national level. Based on it and machine learning (ML) methods, numerous studies mapped high-accuracy, wholesale and continuous PM2.5. However, different models and data in these studies made their results incomparable, and more samples were needed to be provided. Here, a large-column and long-term sample dataset (ML-CNPM2.5) applied for ML-based models was constructed with 5076608 data records and 24 features from 2014 to 2023 in China. Multiple approaches were used to guarantee the quantity and quality of the sample dataset. Due to its comprehensiveness and objectivity, the ML-CNPM2.5 can be used to train and validate different models, thereby further improving the accuracy of PM2.5 estimating. Using the ML-CNPM2.5, eight basic ML-based models were also constructed as the baseline for judging other derivative models. These models can estimate daily full-coverage PM2.5 and most performed well, with ten-fold cross-validation (10 CV) RMSE of 16.94- 11.21 mu g/m(3) and R-2 of 0.71-0.89, which is consistent with previous studies and can effectively capture spatial trends of PM2.5 in a period suffered from high pollution. Overall, our ML-CNPM2.5 can be applied to effectively construct, validate, and compare various ML-based models for PM2.5 estimation, helping to develop new algorithms with higher accuracy and robustness.