Obtaining PM_2.5 data for the entirety of a research region underlies the study of the relationship between PM_2.5 and human spatiotemporal activity. A professional sampler with a filter membrane is used to measure accurate values of PM_2.5 at single points in space. However, there are numerous PM_2.5 sampling and monitoring facilities that rely on data from only representative points, and which cannot measure the data for the whole region of research interest. This provides the motivation for researching the methods of estimation of particulate matter in areas having fewer monitors at a special scale, an approach now attracting considerable academic interest. The aim of this study is to (1) reclassify and particularize the most frequently used approaches for estimating the PM_2.5 concentrations covering an entire research region; (2) list improvements to and integrations of traditional methods and their applications; and (3) compare existing approaches to PM_2.5 estimation on the basis of accuracy and applicability.

The frequency of marine oil spills has increased in recent years. The growing exploitation of marine oil and continuous increase in marine crude oil transportation has caused tremendous damage to the marine ecological environment. Using synthetic aperture radar (SAR) images to monitor marine oil spills can help control the spread of oil spill pollution over time and reduce the economic losses and environmental pollution caused by such spills. However, it is a significant challenge to distinguish between oil-spilled areas and oil-spill-like in SAR images. Semantic segmentation models based on deep learning have been used in this field to address this issue. In addition, this study is dedicated to improving the accuracy of the U-Shape Network (UNet) model in identifying oil spill areas and oil-spill-like areas and alleviating the overfitting problem of the model; a feature merge network (FMNet) is proposed for image segmentation. The global features of SAR image, which are high-frequency component in the frequency domain and represents the boundary between categories, are obtained by a threshold segmentation method. This can weaken the impact of spot noise in SAR image. Then high-dimensional features are extracted from the threshold segmentation results using convolution operation. These features are superimposed with to the down sampling and combined with the high-dimensional features of original image. The proposed model obtains more features, which allows the model to make more accurate decisions. The overall accuracy of the proposed method increased by 1.82% and reached 61.90% compared with the UNet. The recognition accuracy of oil spill areas and oil-spill-like areas increased by approximately 3% and reached 56.33%. The method proposed in this paper not only improves the recognition accuracy of the original model, but also alleviates the overfitting problem of the original model and provides a more effective monitoring method for marine oil spill monitoring. More importantly, the proposed method provides a design principle that opens up new development ideas for the optimization of other deep learning network models.

Air-borne particulate matter, PM_2.5 (PM having a diameter of less than 2.5 micrometers), has aroused widespread concern and is a core indicator of severe air pollution in many cities globally. In our study, we present a validated framework to predict the daily PM_2.5 distributions, exemplified by a use case of Shijiazhuang City, China, based on daily aerosol optical depth (AOD) datasets. The framework involves obtaining the high-resolution spatiotemporal AOD distributions, estimation of the spatial distributions of PM_2.5 and the prediction of these based on a convolutional long short-term memory (ConvLSTM) model. In the estimation part, the eXtreme gradient boosting (XGBoost) model has been determined as the estimation model with the lowest root mean square error (RMSE) of 32.86 µg/m³ and the highest coefficient of determination regression score function (R²) of 0.71, compared to other common models used as a baseline for comparison (linear, ridge, least absolute shrinkage and selection operator (LASSO) and cubist). For the prediction part, after validation and comparison with a seasonal autoregressive integrated moving average (SARIMA), which is a traditional time-series prediction model, in both time and space, the ConvLSTM gives a more accurate performance for the prediction, with a total average prediction RMSE of 14.94 µg/m³ compared to SARIMA’s 17.41 µg/m³. Furthermore, ConvLSTM is more stable and with less fluctuations for the prediction of PM_2.5 in time, and it can also eliminate better the spatial predicted errors compared to SARIMA.

The rapid, accurate extraction of water body information is critical for water resource management and disaster assessment. Its data foundation was mostly provided by remote sensing images through deep learning methods. However, the methods still require the improvement of recognition accuracy and reduction of model size. As a solution, this paper proposed a new high-precision convolutional neural network for water body extraction. This network’s structural design is based on the assumption that the extraction effect of a convolutional neural network is independent from its parameters number, thus the recognition effect could be effectively improved through reasonable adjustment of the network structure according to characteristics of water bodies on high-resolution remote sensing images. It brings two critical improvements. Firstly, the number of downsampling layers was reduced to adapt to the low resolution of remote sensing imagery. Secondly, the bottleneck structure has also been updated to fit the decoder–encoder framework. The improved bottleneck structures were nested to ensure the transmission of water characteristics information in the model. In comparison with the other five commonly used networks, the new network has achieved the best results (average overall accuracy: 98.31%, parameter benefit value: 0.2625), indicating the extremely high practical value of this approach.

The popular simulation process that uses traditional cellular automata with a fixed time step to simulate forest fire spread may be limited in its ability to reflect the characteristics of actual fire development. This study combines cellular automata with an existing forest fire model to construct an improved forest fire spread model, which calculates a speed change rate index based on the meteorological factors that affect the spread of forest fires and the actual environment of the current location of the spread. The proposed model can adaptively adjust the time step of cellular automata through the speed change rate index, simulating forest fire spread more in line with the actual fire development trends while ensuring accuracy. When used to analyze a forest fire that occurred in Mianning County, Liangshan Prefecture, Sichuan Province in 2020, our model exhibited simulation accuracy of 96.9%, and kappa coefficient of 0.6214. The simulated fire situation adapted well to the complex and dynamic fire environment, accurately depicting the detailed fire situation. The algorithm can be used to simulate and predict the spread of forest fires, ensuring the accuracy of spread simulation and helping decision makers formulate reasonable plans.

Estimating and mapping population distributions dynamically at a city-wide spatial scale, including those covering suburban areas, has profound, practical, applications such as urban and transportation planning, public safety warning, disaster impact assessment and epidemiological modelling, which benefits governments, merchants and citizens. More recently, call detail record (CDR) of mobile phone data has been used to estimate human population distributions. However, there is a key challenge that the accuracy of such a method is difficult to validate because there is no ground truth data for the dynamic population density distribution in time scales such as hourly. In this study, we present a simple and accurate method to generate more finely grained temporal-spatial population density distributions based upon CDR data. We designed an experiment to test our method based upon the use of a deep convolutional generative adversarial network (DCGAN). In this experiment, the highest spatial resolution of every grid cell is 125125 square metre, while the temporal resolution can vary from minutes to hours with varying accuracy. To demonstrate our method, we present an application of how to map the estimated population density distribution dynamically for CDR big data from Beijing, choosing a half hour as the temporal resolution. Finally, in order to cross-check previous studies that claim the population distribution at nighttime (from 8 p.m. to 8 a.m. on the next day) mapped by Beijing census data are similar to the ground truth data, we estimated the baseline distribution, first, based upon records in CDRs. Second, we estimate a baseline distribution based upon Global Navigation Satellite System (GNSS) data. The results also show the Root Mean Square Error (RMSE) is about 5000 while the two baseline distributions mentioned above have an RMSE of over 13,500. Our estimation method provides a fast and simple process to map people’s actual density distributions at a more finely grained, i.e., hourly, temporal resolution.