Air pollution has become a serious problem threatening human health. Effective prediction models can help reduce the adverse effects of air pollutants. Accurate predictions of air pollutant concentration can provide a scientific basis for air pollution prevention and control. However, the previous air pollution-related prediction models mainly processed air quality prediction, or the prediction of a single or two air pollutants. Meanwhile, the temporal and spatial characteristics and multiple factors of pollutants were not fully considered. Herein, we establish a deep learning model for an atmospheric pollutant memory network (LSTM) by both applying the one-dimensional multi-scale convolution kernel (ODMSCNN) and a long-short-term memory network (LSTM) on the basis of temporal and spatial characteristics. The temporal and spatial characteristics combine the respective advantages of CNN and LSTM networks. First, ODMSCNN is utilized to extract the temporal and spatial characteristics of air pollutant-related data to form a feature vector, and then the feature vector is input into the LSTM network to predict the concentration of air pollutants. The data set comes from the daily concentration data and hourly concentration data of six atmospheric pollutants (PM_2.5, PM₁₀, NO₂, CO, O₃, SO₂) and 17 types of meteorological data in Xi’an. Daily concentration data prediction, hourly concentration data prediction, group data prediction and multi-factor prediction were used to verify the effectiveness of the model. In general, the air pollutant concentration prediction model based on ODMSCNN-LSTM shows a better prediction effect compared with multi-layer perceptron (MLP), CNN, and LSTM models.

Urban agglomeration is a primary source of global energy consumption and CO₂ emissions. It is employed as a major means of modern economic and social activities. Analysis of the temporal and spatial characteristics of CO₂ emissions in urban agglomerations and prediction of the future trends of CO₂ emissions in urban agglomerations will help in the implementation of CO₂ reduction policies within region-wide areas. So, based on that, this study contains four aspects. Firstly, it calculates the energy CO₂ emissions of China’s Chengdu-Chongqing urban agglomeration. Secondly, it analyzes the time and space changes in the area by using ArcGIS. Then, the STIRPAT model is used to investigate the factors influencing CO₂ emissions, and the elasticity coefficient of the influencing factors is estimated using the ridge regression method, and the important influencing factors are screened on the basis of the estimated results, which are then used as input features for prediction. Finally, a combined prediction model based on the improved GM (1, N) and SVR models is constructed, and then the optimal solution is found through the particle swarm optimization algorithm. It sets up different CO₂ emission scenarios to predict the energy CO₂ emission of the region and its cities. The results show that, first, the CO₂ emissions of the Chengdu-Chongqing urban agglomeration have accumulated year by year, but by 2030, as predicted, it will not reach its peak. The spatial layout of CO₂ emissions in this region is not expected to undergo major changes by 2030. Second, population, GDP, gas and electricity consumption, and industrial structure have served as important factors affecting energy CO₂ emissions in the region. Third, on the basis of the prediction results for different scenarios, the CO₂ emissions in the baseline scenario are low in the short term, but the CO₂ emissions in the low-carbon scenario are low in the long run. This study also puts forward some policy recommendations on how to reduce CO₂ emissions.

Fine particulate matter (PM_2.5) has a continuing impact on the environment, climate change and human health. In order to improve the accuracy of PM_2.5 estimation and obtain a continuous spatial distribution of PM_2.5 concentration, this paper proposes a LUR-GBM model based on land-use regression (LUR), the Kriging method and LightGBM (light gradient boosting machine). Firstly, this study modelled the spatial distribution of PM_2.5 in the Chinese region by obtaining PM_2.5 concentration data from monitoring stations in the Chinese study region and established a PM_2.5 mass concentration estimation method based on the LUR-GBM model by combining data on land use type, meteorology, topography, vegetation index, population density, traffic and pollution sources. Secondly, the performance of the LUR-GBM model was evaluated by a ten-fold cross-validation method based on samples, stations and time. Finally, the results of the model proposed in this paper are compared with those of the back propagation neural network (BPNN), deep neural network (DNN), random forest (RF), XGBoost and LightGBM models. The results show that the prediction accuracy of the LUR-GBM model is better than other models, with the R² of the model reaching 0.964 (spring), 0.91 (summer), 0.967 (autumn), 0.98 (winter) and 0.976 (average for 2016–2021) for each season and annual average, respectively. It can be seen that the LUR-GBM model has good applicability in simulating the spatial distribution of PM_2.5 concentrations in China. The spatial distribution of PM_2.5 concentrations in the Chinese region shows a clear characteristic of high in the east and low in the west, and the spatial distribution is strongly influenced by topographical factors. The seasonal variation in mean concentration values is marked by low summer and high winter values. The results of this study can provide a scientific basis for the prevention and control of regional PM_2.5 pollution in China and can also provide new ideas for the acquisition of data on the spatial distribution of PM_2.5 concentrations within cities.

As VOCs pose a threat to human health, it is important to accurately capture changes in VOCs concentrations and sense VOCs concentrations in relevant areas. Therefore, it is necessary to improve the accuracy of VOCs concentration prediction and realise the VOCs aggregation situation sensing. Firstly, on the basis of regional grid division, the inverse distance spatial interpolation method is used for spatial interpolation to collect regional VOCs data information. Secondly, extreme gradient boosting (XGBoost) is used for spatio-temporal feature selection, combined with graph convolutional neural network (GCN) to construct regional spatial relationships of VOCs, and multiple linear regression (MLR) to process VOCs time series data and predict the VOCs concentration in the grid. Finally, the aggregation potential values of VOCs are calculated based on the prediction results, and the potential perception results are visualised. A VOCs aggregation perception method based on concentration prediction is proposed, using the XGBoost-GCN-MLR method with a scenario-aware approach for VOCs to perceive the VOCs aggregation in the relevant region. VOCs concentration prediction and VOCs aggregation trend perception were carried out in Xi’an, Baoji, Tongchuan, Weinan and Xianyang. The results show that compared with the GCN model, XGBoost model, MLR model and GCN-MLR model, the XGBoost-GCN-MLR model reduces the input variables, achieves the optimisation of the input parameters of the VOCs concentration prediction model, reduces the complexity of the prediction model and improves the prediction accuracy. Intelligent sensing of VOCs aggregation can visualise the regional VOCs. The intelligent sensing of VOCs aggregation can visualise the development trend and status of regional VOCs aggregation and convey more information, which has practical value.