Springer Science and Business Media LLC

Machine Learning technologies have the potential to deliver new nonlinear mineral prospectivity mapping (MPM) models. In this study, Back Propagation (BP) neural network Support Vector Machine (SVM) methods were applied to MPM in the Hatu region of Xinjiang, northwestern China. First, a conceptual model of mineral prospectivity for Au deposits was constructed by analysis of geological background. Evidential layers were selected and transformed into a binary data format. Then, the processes of selecting samples and parameters were described. For the BP model, the parameters of the network were 9–10 − 1; for the SVM model, a radial basis function was selected as the kernel function with best C = 1 and γ = 0.25. MPM models using these parameters were constructed, and threshold values of prediction results were determined by the concentration-area (C-A) method. Finally, prediction results from the BP neural network and SVM model were compared with that of a conventional method that is the weight- of- evidence (W- of- E). The prospectivity efficacy was evaluated by traditional statistical analysis, prediction-area (P-A) plots, and the receiver operating characteristic (ROC) technique. Given the higher intersection position (74% of the known deposits were within 26% of the total area) and the larger AUC values (0.825), the result shows that the model built by the BP neural network algorithm has a relatively better prediction capability for MPM. The BP neural network algorithm applied in MPM can elucidate the next investigative steps in the study area.

A local anomaly detection algorithm based on sliding windows in spectral space has been proposed in this research. The traditional local anomaly detection algorithms are implemented in spatial windows because local data of an image scene is more suitable for a single statistical model than global data. However, from the aspect of geometric structure of a dataset, this assumption is not entirely proper. As multivariate data, the hyperspectral image dataset can be considered as a low-dimensional manifold, embedded in the high-dimensional spectral space. The nonlinear spectral mixture occurs more frequently, as well as a low dimensional manifold being nonlinear. The traditional spatial local anomaly detection algorithms based on linear projection would not be appropriate to deal with this kind of data. This paper studies the local linear ideas in manifold learning, and an anomaly detection algorithm has been implemented based on the linear projections in a local area of spectral space. The key concept is that a small neighborhood areas of nonlinear manifold can be considered as a local linear structure. The classic spatial local algorithms and proposed algorithm are compared by using real hyperspectral images from vehicle and aviation platforms. The results demonstrated the effectiveness of the proposed algorithm in improving detection of the weak anomalies that decreases the number of false alarms.

After an earthquake, the goal of emergency rescue work is to minimize the number of casualties and save lives. The most important initial task is the rapid assessment of the number of earthquake casualties. At present, the rapid assessment of casualties relies primarily on census data rather than real-time population data in an earthquake area. Features of the census data, such as their low accuracy and time lag, can easily cause large errors in the assessment result; thus, the assessment result cannot always reflect the actual situation in an earthquake area. In this paper, we use phone signal data to construct a population density model and estimate the real-time population in a seismic area. The results show that the estimated population is consistent with the actual data in the study area. Finally, we obtain a real-time population distribution map of the study area. The real-time distribution of the population is consistent with the actual situation based on the economic and social development of the study area. The population in an urban area is relatively dense, whereas that in a rural area is less dense. The results show that phone signal data can play a useful role in estimating the real-time population when an earthquake occurs and can be used to support earthquake emergency rescue work.

In present study, the variability in wintertime maximum (Tmax) and minimum (Tmin) temperature patterns over India using observed and deep learning techniques have been assessed. The analysis has been caried out for the period 1979–2018 during the months from November to February. The month of February depicted strongest variability in Tmax and Tmin over Northwest India (NWI) with significant + ve trend for upper half of the country. Wintertime temperature variability was seen to be dominant in the Indo-Gangetic plain area covering some parts of NWI and Northeast India (NEI) for Tmax and Tmin. Also, a gradual increase in the spatial coverage, engulfing majority of South Peninsular India (SPI) and Central India (CI) of the rising Diurnal Temperature Range (DTR) was found from November to January. Decreasing DTR was observed only for January extending along Indo-Gangetic plains. The model Random Forest (RF) performed quite well relative to Long Short-Term Memory model (LSTM) in predicting the winter temperatures (especially for Tmax) during all the considered months. The RF made a robust Tmax forecast during NDJF over all India (RMSE – 0.51, MAPE – 1.4). However, its performance is not up to the mark during the month of February over NEI (RMSE – 1.63, MAPE – 4.5). The maximum fluctuating patterns of temperature have been found during the month of February. The study emphasizes on algorithm-based approaches to study the temperature, so that better understanding could be developed for the meteorological sub-divisions over India.

The analysis of groundwater resources is extremely important to the cultivation of crops, our daily lives, and sustainable growth. Thus, a precise and credible estimation of groundwater levels is crucial and helps to prevent resource depletion. Research in the past has shown that machine-learning approaches such as Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) are effective at mimicking complicated nonlinear problems and also integrating multiple techniques to develop an enhanced tool that improves the accuracy of the prediction model. This study presented prediction models that constitute hybrid techniques integrating ANN, ANFIS, and an Improved Reptile Search Algorithm (IRSA). In this case, IRSA is applied to identify the parameters of the ANN and ANFIS in order to improve the overall effectiveness of the forecasting models. After this, the developed hybrid approaches for groundwater level modelling were evaluated using four seasons (January to March, April to June, July to September, and October to December) of groundwater level data for India collected from the India Water Resources Information System. Also, the comparative study has done between ANN-IRSA, ANFIS-IRSA, and traditional ANN as well as ANFIS, which were evaluated on the same datasets. The ANFIS-IRSA model achieved optimal RMSE (0.4074, 0.3927, 4.5591, 1.8408), MAE (0.2329, 0.2516, 1.3644, 0.8612), MAPE (0.0201, 0.022, 0.04, 0.0664), R2 (0.9907, 0.9861, 0.9747, 0.9809), WI (0.9975, 0.9963, 0.9926, 0.9948), and PBIAS (0.2862, 0.0833, 2.7417, 1.7682) for four distinct seasons, which is exceptional in comparison with other models. Based on the results of simulations and comparisons, ANFIS-IRSA outscored other models on the same datasets and proved to be a robust method to predict time-series data.

Air quality has a significant impact on human health, in order to alleviate the air pollution and improve the ability to predict the air quality. In this paper, a prediction model of air quality index composed of variational mode decomposition and temporal convolutional network was proposed. First, in order to reduce the non-stationarity and randomness of the time series, the original air quality index sequence was decomposed by variational mode decomposition, and the decomposition number was determined by the central frequency method to decompose into multiple relatively stable sub-sequences with different frequency scales. Then, the decomposed sub-stable sequence was predicted by the time convolutional network. Finally, the prediction data were integrated and reconstructed to obtain the final prediction results. Comparing the results of other forecasting models by performance evaluation metrics, the combined forecasting model proposed in this paper reduced RMSE by 20.9%, 19.2%, 5.1%, 29.9%, 23.7% on the Beijing dataset. MAPE reduced by 26.6%, 22.3%, 19.5%, 28.9%, 15.0%, respectively. MAE decreased by 19.1%, 20.6%, 9.6%, 29.5%, 23.5%. R2 increased by 4.6%, 4.0%, 0.8%, 14.9%, 5.5% respectively. This proves the accuracy and reliability of the proposed model.

The Wubaochang structure is located at the edge of the eastern Sichuan structural belt, at the intersection of the Dabashan and eastern Sichuan arcuate fold-and-thrust belts. The tectonic setting is complex and yields poor seismic imaging results. Within this structure, the Jurassic Shaximiao Formation is a low-efficiency tight sandstone gas reservoir with low single-well production. Currently, horizontal wells are installed in the main horizons to enhance productivity and efficiency. The drilled wells show that the reservoir consists of multiple horizons with rapid lateral facies changes and highly heterogeneous physical properties. Production well pressure patterns are segmented in striped zones and significantly controlled by the sedimentary microfacies. The microfacies, geometry, and distribution patterns of the sand beds are poorly understood. As a result, there have been few high-quality horizontal wells drilled in the reservoir, significantly restricting the exploration and development efficiency of the Shaximiao Formation in this area. In this study, we examined the regional sedimentary background and relative paleo topography of the study area and synthesized the results of core descriptions, rock composition analysis, and well log and seismic response feature evaluations. Finally, we determined that the sandstones in the study area are braided river delta deposits, mainly developed with the microfacies of braided channel, flood plain, mid-channel bar, mid-channel bar complex, underwater distributary channel, and interdistributary bay. We also examined the planar distribution pattern of the sand bodies with different microfacies, based on single-well microfacies division, paleo topographic feature evaluation, and seismic information entropy analysis. The results show that the mid-channel bar and mid-channel bar complex sands are the most favorable reservoir sands in this gas reservoir, followed by the braided channel sands and underwater distributary channel sands. The topography controlled the sand deposition distribution patterns and the development of the different microfacies. Seismic entropy reflected the amount of seismic signal information and degree of chaos. The examination of seismic entropy information provides a more stable and effective stratigraphic interpretation than the use of conventional seismic methods for identifying different microfacies of braided river deltas. Seismic entropy information can provide a reliable seismic and geological basis for optimizing microfacies targets and horizontal well placement.