Urban land-use information is important for urban land-resource planning and management. However, current methods using traditional surveys cannot meet the demand for the rapid development of urban land management. There is an urgent need to develop new methods to overcome the shortcomings of conventional methods. To address the issue, this study used the random forest (RF), support vector machine (SVM), and artificial neural network (ANN) models to build machine-leaning methods for urban land-use classification. Taking Hangzhou as an example, these machine-leaning methods could all successfully classify the essential urban land use into 6 Level I classes and 13 Level II classes based on the semantic features extracted from Sentinel-2A images, multi-source features of types of points of interest (POIs), land surface temperature, night lights, and building height. The validation accuracy of the RF model for the Level I and Level II land use was 79.88% and 71.89%, respectively, performing better compared to SVM (78.40% and 68.64%) and ANN models (71.30% and 63.02%). However, the variations of the user accuracy among the methods depended on the urban land-use level. For the Level I land-use classification, the user accuracy was high, except for the transportation land by all methods. In general, the RF and SVM models performed better than the ANN model. For the Level II land-use classification, the user accuracy of different models was quite distinct. With the RF model, the user accuracy of educational and medical land was above 80%. Moreover, with the SVM model, the user accuracy of the business office and educational land classification was above 75%. However, the user accuracy of the ANN model on the Level II land-use classification was poor. Our results showed that the RF model performs best, followed by SVM model, and ANN model was relatively poor in the essential urban land-use classification. The results proved that the use of machine-learning methods can quickly extract land-use types with high accuracy, and provided a better method choice for urban land-use information acquisition.

Land use change has an important influence on the spatial and temporal distribution of PM_2.5 concentration. Therefore, based on the particulate matter (PM_2.5) data from remote sensing instruments and land use change data in long time series, the Getis-Ord Gi* statistic and SP-SDM are employed to analyze the spatial distribution pattern of PM_2.5 and its response to land use change in China. It is found that the average PM_2.5 increased from 25.49 μg/m³ to 31.23 μg/m³ during 2000-2016, showing an annual average growth rate of 0.97%. It is still greater than 35 μg/m³ in nearly half of all cities. The spatial distribution pattern of PM_2.5 presents the characteristics of concentrated regional convergence. PM_2.5 is positively correlated with urban land and farmland, negatively correlated with forest land, grassland, and unused land. Furthermore, the average PM_2.5 concentrations show the highest values for urban land and decrease in the order of farmland > unused land > water body > forest > grassland. The impact of land use change on PM_2.5 is a non-linear process, and there are obvious differences and spillover effects for different land types. Thus, reasonably controlling the scale of urban land and farmland, optimizing the spatial distribution pattern and development intensity, and expanding forest land and grassland are conducive to curbing PM_2.5 pollution. The research conclusions provide a theoretical basis for the management of PM_2.5 pollution from the perspective of optimizing land use.

The lockdown of cities in the Yangtze River Delta (YRD) during COVID-19 has provided many natural and typical test sites for estimating the potential of air pollution control and reduction. To evaluate the reduction of PM_2.5 concentration in the YRD region by the epidemic lockdown policy, this study employs big data, including PM_2.5 observations and 29 independent variables regarding Aerosol Optical Depth (AOD), climate, terrain, population, road density, and Gaode map Point of interesting (POI) data, to build regression models and retrieve spatially continuous distributions of PM_2.5 during COVID-19. Simulation accuracy of multiple machine learning regression models, i.e., random forest (RF), support vector regression (SVR), and artificial neural network (ANN) were compared. The results showed that the RF model outperformed the SVR and ANN models in the inversion of PM_2.5 in the YRD region, with the model-fitting and cross-validation coefficients of determination R² reached 0.917 and 0.691, mean absolute error (MAE) values were 1.026 μg m⁻³ and 2.353 μg m⁻³, and root mean square error (RMSE) values were 1.413 μg m⁻³, and 3.144 μg m⁻³, respectively. PM_2.5 concentrations during COVID-19 in 2020 have decreased by 3.61 μg m⁻³ compared to that during the same period of 2019 in the YRD region. The results of this study provide a cost-effective method of air pollution exposure assessment and help provide insight into the atmospheric changes under strong government controlling strategies.