Search

It is essential to predict carbon prices precisely in order to reduce CO2 emissions and mitigate global warming. As a solution to the limitations of a single machine learning model that has insufficient forecasting capability in the carbon price prediction problem, a carbon price prediction model (GWO–XGBOOST–CEEMDAN) based on the combination of grey wolf optimizer (GWO), extreme gradient boosting (XGBOOST), and complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is put forward in this paper. First, a random forest (RF) method is employed to screen the primary carbon price indicators and determine the main influencing factors. Second, the GWO–XGBOOST model is established, and the GWO algorithm is utilized to optimize the XGBOOST model parameters. Finally, the residual series of the GWO–XGBOOST model are decomposed and corrected using the CEEMDAN method to produce the GWO–XGBOOST–CEEMDAN model. Three carbon emission trading markets, Guangdong, Hubei, and Fujian, were experimentally predicted to verify the model's validity. Based on the experimental results, it has been demonstrated that the proposed hybrid model has enhanced prediction precision compared to the comparison model, providing an effective experimental method for the prediction of future carbon prices. [ABSTRACT FROM AUTHOR]

The rainfall can contribute significantly to landslide events, especially in hilly areas. The landslide susceptibility map (LSM) usually helps to mitigate disasters. However, how to accurately predict the susceptibility of landslides is still a difficult point in the field of disaster research. In this study, five advanced machine learning technologies (MLTs), including the Light Gradient Boosting Machine, extreme gradient boost, categorical boosting (CatBoost), support vector machine, and random forest, are utilized to landslide susceptibility modeling and their capabilities are compared through evaluation indicators. The northern part of Yanping, Fujian Province, China, is selected as the research object, because this area experienced mass landslide events due to extremely heavy rainfall in June 2010, resulting in many casualties and a large number of public facilities destroyed. The influencing factors for landslides, namely topographic, hydrological, geologic and human activities, are prepared from various data sources based on the availability. Through the analysis of the actual situation in the study area, 13 suitable landslide condition factors are considered and the availability of relevant factors is checked according to the multicollinearity test. The landslide inventory including 631 samples in this study area is obtained from historical information, satellite data in Google earth and performed field surveys. The landslide inventory is randomly divided into two datasets for model training and testing with a 7:3 ratio. The area under the curve of ROC, accuracy rate, Kappa index and F1 score are applied to compare the MLTs capabilities. In this paper, the results of factor importance analysis show that the first three important condition factors are the distance to faults, the distance to drainages and the slope. According to the LSMs, in the study area, the central and western regions are at high and very high landslide susceptibility levels, while almost all the eastern and northeastern regions are at medium and low landslide susceptibility levels. The CatBoost model is a very promising technology in landslide research according to the evaluation results, which means that for landslide susceptibility research, gradient boosting algorithms may get more accurate results and show better prospects in the future. Finally, the results of this paper will contribute to environmental protection to a certain extent. [ABSTRACT FROM AUTHOR]

In recent years, artificial intelligence technology has widely influenced the design field, introducing new ideas to efficiently and systematically solve urban renewal design problems. The purpose of this study is to create a stylized generation technology for building facade decoration in historic districts, which will aid in the design and control of district style and form. The goal is to use the technical advantages of the conditional generative adversarial network (CGAN) in image generation and style transfer to create a method for independently designing a specific facade decoration style by interpreting image data of historical district facades. The research in this paper is based on the historical district of Putian in Fujian Province and facilitates an experiment of image data acquisition, image processing and screening, model training, image generation, and style matching of the target area. The research found the following: (1) CGAN technology can better identify and generate the decorative style of historical districts. It can realize the overall or partial scheme design of the facade. (2) In terms of adaptability, this method can provide a better scheme reference for historical district reconstruction, facade renovation, and renovation design projects. Especially for districts with obvious decorative styles, the visualization effect is better. In addition, it also has certain reference significance for the determination and design of the facade decoration style of a specific historical building. (3) Lastly, this method can better learn the internal laws of the complex district style and form to generate a new design with a clear decoration style attribute. It can be extended to other fields of historical heritage protection to enhance practitioners' stylized control of the heritage environment and improve the efficiency and capability of professional design. [ABSTRACT FROM AUTHOR]

Landslide susceptibility mapping of mountain roads is frequently confronted by insufficient historical landslide sample data, multicollinearity of existing evaluation index factors, and inconsistency of evaluation factors due to regional environmental variations. Then, a single machine learning model can easily become overfitting, thus reducing the accuracy and robustness of the evaluation model. This paper proposes a combined machine-learning model to address the issues. The landslide susceptibility in mountain roads were mapped by using factor analysis to normalize and reduce the dimensionality of the initial condition factor and generating six new combination factors as evaluation indexes. The mountain roads in the Youxi County, Fujian Province, China were used for the landslide susceptibility mapping. Three most frequently used machine learning techniques, support vector machine (SVM), random forest (RF), and artificial neural network (ANN) models, were used to model the landslide susceptibility of the study area and validate the accuracy of this evaluation index system. The global minimum variance portfolio was utilized to construct a machine learning combined model. 5-fold cross-validation, statistical indexes, and AUC (Area Under Curve) values were implemented to evaluate the predictive accuracy of the landslide susceptibility model. The mean AUC values for the SVM, RF, and ANN models in the training stage were 89.2%, 88.5%, and 87.9%, respectively, and 78.0%, 73.7%, and 76.7%, respectively, in the validating stage. In the training and validation stages, the mean AUC values of the combined model were 92.4% and 87.1%, respectively. The combined model provides greater prediction accuracy and model robustness than one single model. [ABSTRACT FROM AUTHOR]

Copyright of Revista Música Hodie is the property of Revista Musica Hodie and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

China's landslide disasters are serious, and regional landslide disaster early-warning is one of the important means of disaster prevention and mitigation. The traditional regional landslide disaster early-warning model, however, is limited by the complex landslide induction mechanism, limited data accumulation, and insufficient big data analysis methods, and has problems such as limited early-warning accuracy and insufficient refinement. In this paper, a machine learning method was introduced into the field of regional landslide disaster warning. From the model construction process of training sample-set construction, sample learning and training, model parameter optimization, model preservation, warning output, and so on, a method for constructing a regional landslide early-warning model based on machine learning was systematically proposed. In the sample learning and training, 80% of the training sample-set was used as the training set, and 20% was used as the test set for five-fold cross validation. The Bayesian Optimization algorithm was used to optimize the model parameters, and the accuracy, ROC curve, and AUC value were used to verify the model accuracy and model generalization ability. With China's Fujian province as an example, based on nine years of geological and meteorological data (2010–2018), geological environment factors, factors of hazard-affected bodies and historical disaster situations, and rainfall-induced factors in four categories, a total of 26 indicators were used as input characteristic parameters. Six machine learning algorithms were adopted to improve model training; the results showed that the Random Forest algorithm performed the best, giving an accuracy of 92.3%, and was the model with the best generalization ability (AUC was 0.955). The second best was the Artificial Neural Network model, with an accuracy of 0.937 and an AUC of 0.935. Next were the Nearest Neighbor model, the Logistic Regression model, and the Support Vector Machine; the poorest results were from the Decision Tree model. Finally, the typical rainfall-type landslide disaster process in Fujian Province was selected as an example to verify the Random Forest algorithm model. The results showed that compared with the early-warning results of the original explicit statistical model, the hit rate of the new model was 6 times, or equal to that of the original model, and the landslide density in the early-warning area of the new model was 1.6–1.7 times that of the original model. Preliminary verification showed that the new model based on the Random Forest method has obvious advantages, a higher hit rate and a smaller warning area, and can achieve more accurate warnings. The follow-up will continue to track the new landslide disaster situation in the study area and carry out model verification and correction. [ABSTRACT FROM AUTHOR]

Due to the inherent uncertainty and the mismatch between renewable energy output and load demand, renewable energy-based system optimization is increasingly important in China. Therefore, electricity demand response (EDR) is critical to the stability and efficiency of an integrated renewable energy system (IRES). The customers' demand response under the time-of-use (TOU) mechanism is the key to the entire process. However, how to optimize integrated renewable energy system based on accurately identifying the electricity demand response participation of customers is one of the main issues at this stage. To solve the problem, this paper presents a novel approach for integrated renewable energy system optimization considering electricity demand response management and multistage energy storage systems from the perspective of Fujian Province, China. To enable large-scale renewable energy integration, the original output is processed using the Hampel identifier (HI)-Savitzky-Golay filter (SGF) technology. A machine learning (ML)-based identification model is developed for EDR to determine whether the customers are participating or not. Driven by the time-of-use mechanism, customers can make dynamic adjustments based on the Logistics function. After that, an integrated renewable energy system model optimization strategy is proposed based on satisfying customers' electricity demand response. Throughout the optimization process, the multistage energy storage system plays a vital role in the residual fluctuation absorption for renewable energy filtering, the dynamic adjustment process of the demand response, and the integrated renewable energy system optimization. Through extensive case studies and discussions, it is demonstrated that random forest (RF) and support vector machine (SVM) reach 80 % identification accuracy with a 20 % testing ratio. According to the optimization approach proposed in this paper, the minimum operation cost can be calculated to be 9.336*106 CNY. The methodology presented in this paper establishes a special relationship between electricity demand response identification and dynamic adjustments, and provides an optimization model for Fujian Province, China, which is a reliable reference for future work. [ABSTRACT FROM AUTHOR]

Extracting geochemical anomalies from geochemical exploration data is one of the most important activities in mineral exploration. Geochemical anomaly detection can be regarded as a binary classification problem. The similarity between geochemical samples can be measured by their distance. The key issue of this classification is to find the intrinsic relationship and distance between geochemical samples to separate geochemical anomalies from background. In this paper, a hybrid method that integrates random forest and metric learning (RFML) is used to identify geochemical anomalies related to Fe-polymetallic mineralization in Southwest Fujian Province of China. RFML does not require any specific statistical assumption on geochemical data, nor does it depend on sufficient known mineral occurrences as the prior knowledge. The geochemical anomaly map obtained by the RFML method showed that the known Fe deposits and the generated geochemical anomaly area have strong spatial association. Meanwhile, the receiver operating characteristic curves for the results of RFML and another method, namely maximum margin metric learning, indicated that the RFML method exhibited better performance, suggesting that RFML can be effectively applied to recognize geochemical anomalies. [ABSTRACT FROM AUTHOR]

Landslide disasters, due to their widespread distribution and clustered occurrences, pose a significant threat to human society. Rainfall is considered a primary triggering factor, and the frequent clustering of landslides underscores the importance of early warning systems for regional landslide disasters in preventing and mitigating rainfall-induced landslides. Research on early warning models is crucial for accurately predicting rainfall-induced landslides. However, traditional models face challenges such as the complexity of landslide causes, insufficient data, and limited analysis methods, resulting in low accuracy and inadequate precision. This study focuses on Fujian Province, China, proposing a four-step process for building a regional landslide early warning model based on machine learning. The process includes data integration and cleaning, sample set construction, model training and validation, and practical application. By integrating and cleaning the latest and most detailed data, a training sample set (15,589 samples) for the regional landslide disaster early warning model is established. Three machine learning algorithms--Random Forest, Multilayer Perceptron, and Convolutional Neural Network--are employed and compared, the evaluation results indicated that the RF-based warning model achieved an accuracy of 0.930-0.957 and an AUC value of 0.955. The CNNbased warning model demonstrated an accuracy of 0.945-0.948 with an AUC value of 0.940. The MLP-based warning model achieved an accuracy of 0.930-0.953 and an AUC value of 0.930. The results showed comparable accuracy metrics among the three models, with RF exhibiting a significant advantage in AUC values. Finally, the models are applied to the regional landslide disasters induced by heavy rainfall in Fujian Province on 5 August 2021. The results showed that in the binary classification warning strategy, the accuracy of the Random Forest and Convolutional Neural Network was 92.9%, while that of the Multilayer Perceptron was 85.8%, all performing well. In the multi-classification hierarchical warning strategy, the Random Forest excelled, while the performance of the Convolutional Neural Network and Multilayer Perceptron was relatively limited. The findings of this study Frontiers in Earth Science 01 frontiersin.org Liu et al. 10.3389/feart.2024.1419421 contribute to valuable attempts in landslide disaster warning model research, with anticipated further improvements through the gradual accumulation of samples and practical application verification. [ABSTRACT FROM AUTHOR]

The optimal selection of characteristic bands and retrieval models for the hyperspectral retrieval of soil heavy metal concentrations poses a significant challenge. Additionally, satellite-based hyperspectral retrieval encounters several issues, including atmospheric effects, limitations in temporal and radiometric resolution, and data acquisition, among others. Given this, the retrieval performance of the soil arsenic (As) concentration in Pingtan Island, the largest island in Fujian Province and the fifth largest in China, is currently unclear. This study aimed to elucidate this issue by identifying optimal characteristic bands from the full spectrum from both statistical and physical perspectives. We tested three linear models, namely Multiple Linear Regression (MLR), Partial Least Squares Regression (PLSR) and Geographically Weighted Regression (GWR), as well as three nonlinear machine learning models, including Back Propagation Neural Network (BP), Support Vector Machine Regression (SVR) and Random Forest Regression (RFR). We then retrieved soil arsenic content using ground-based soil full spectrum data on Pingtan Island. Our results indicate that the RFR model consistently outperformed all others when using both original and optimal characteristic bands. This superior performance suggests a complex, nonlinear relationship between soil arsenic concentration and spectral variables, influenced by diverse landscape factors. The GWR model, which considers spatial non-stationarity and heterogeneity, outperformed traditional models such as BP and SVR. This finding underscores the potential of incorporating spatial characteristics to enhance traditional machine learning models in geospatial studies. When evaluating retrieval model accuracy based on optimal characteristic bands, the RFR model maintained its top performance, and linear models (MLR, PLSR and GWR) showed notable improvement. Specifically, the GWR model achieved the highest r value for the validation data, indicating that selecting optimal characteristic bands based on high Pearson's correlation coefficients (e.g., abs(Pearson's correlation coefficient) ≥0.45) and high sensitivity to soil active materials successfully mitigates uncertainties linked to characteristic band selection solely based on Pearson's correlation coefficients. Consequently, two effective retrieval models were generated: the best-performing RFR model and the improved GWR model. Our study on Pingtan Island provides theoretical and technical support for monitoring and evaluating soil arsenic concentrations using satellite-based spectroscopy in densely populated, relatively independent island towns in China and worldwide. [ABSTRACT FROM AUTHOR]

Background: In recent years, identifying players with injury risk through physical fitness assessment has become a hot topic in sports science research. Although practitioners have conducted many studies on the relationship between physical fitness and the likelihood of injury, the relationship between the two remains indeterminate. Consequently, this study utilized machine learning to preliminary investigate the relationship between individual physical fitness tests and injury risk, aiming to identify whether patterns of physical fitness change have an impact on injury risk. Methods: This study conducted a retrospective analysis by extracting the records of 17 young female basketball players from the sport-specific physical fitness monitoring and injury registration database in Fujian Province. Sports-specific physical fitness tests included physical performance, physiological, biochemical, and subjective perceived responses. The data for each player was standardized individually using Z-scores. Synthetic minority over-sampling techniques and edited nearest neighbor algorithms were used to sample the training set to address the negative impact of class imbalance on model performance. Feature extraction was performed on the dataset using linear discriminant analysis, and the prediction model was constructed using the cost-sensitive neural network. Results: The 10 replicate 5-fold stratified cross-validation showed that the lower limb non-contact injury prediction model based on the cost-sensitive neural network had achieved good discrimination and calibration (average Precision: 0.6360; average Recall: 0.8700; average F2-Score: 0.7980; average AUC: 0.8590; average Brier-score: 0.1020), which could be well applied in training practice. According to the attribution analysis, agility and speed were important physical attributes that affect youth female basketball players’ non-contact lower limb injury risk. Specifically, there was enhance in the performance of the 1-min double under, accompanied by an increase in urinary ketone and urinary blood levels following the agility test. The 3/4 basketball court sprint performance improved, while urinary protein and RPE levels decreased after the speed test. Conclusion: The sport-specific physical fitness change pattern can impact the lower limb non-contact injury risk of young female basketball players in Fujian Province, specifically in terms of agility and speed. These findings will provide valuable insights for planning athletes’ physical training programs, managing fatigue, and preventing injuries. [ABSTRACT FROM AUTHOR]

Identifying the geographical origins of white tea is of significance because the quality and price of white tea from different production areas vary largely from different growing environment and climatic conditions. In this study, we used near-infrared spectroscopy (NIRS) with white tea (n = 579) to produce models to discriminate these origins under different conditions. Continuous wavelet transform (CWT), min-max normalization (Minmax), multiplicative scattering correction (MSC) and standard normal variables (SNV) were used to preprocess the original spectra (OS). The approaches of principal component analysis (PCA), linear discriminant analysis (LDA) and successive projection algorithm (SPA) were used for features extraction. Subsequently, identification models of white tea from different provinces of China (DPC), different districts of Fujian Province (DDFP) and authenticity of Fuding white tea (AFWT) were established by K-nearest neighbors (KNN), random forest (RF) and support vector machine (SVM) algorithms. Among the established models, DPC-CWT-LDA-KNN, DDFP-OS-LDA-KNN and AFWT-OS-LDA-KNN have the best performances, with recognition accuracies of 88.97%, 93.88% and 97.96%, respectively; the area under curve (AUC) values were 0.85, 0.93 and 0.98, respectively. The research revealed that NIRS with machine learning algorithms can be an effective tool for the geographical origin traceability of white tea. [ABSTRACT FROM AUTHOR]

Magnetite geochemistry is crucial for the discrimination of ore deposit genetic type. Traditional two-dimensional discrimination diagrams based on particular data for limited deposits cannot meet the requirements of high-precision classification. The continuous compilation of magnetite geochemical big data and high-precision machine learning algorithms provide a new research route. In this study, a laser ablation inductively coupled plasma mass spectrometry (LA–ICP–MS) magnetite geochemical dataset compiled from published literature for different geological environments worldwide was used to train different machine learning (ML) classifiers, including random forest, support vector machine, and multilayer perceptron neural network, to predict the genetic type of ore deposits. To verify the efficacy of the classifier, LA–ICP–MS analysis was performed on magnetite samples collected from the Makeng and Luoyang iron deposits in Fujian Province, southeastern China. The obtained data were used to predict the ore deposit genetic type with the aid of the ML classifiers. The classifiers established by all three algorithms yielded good predictive outcomes. The results were consistent with those obtained from detailed petrogeochemical and isotopic studies, showing that the deposits are skarn-type, validating the suitability of the classifier. This article also provides an executable program of the classifiers for anyone to use. The executable program meets the requirements for ore deposit-type classification and can be considered a powerful tool for magnetite exploration and prospecting. [ABSTRACT FROM AUTHOR]

Deep autoencoder (AE) networks show a powerful ability for geochemical anomaly identification. Because of little contribution to the AE network, small probability samples (again, please check this) having comparatively high reconstructed errors can be recognized by the trained model as anomalous samples. However, different autoencoder networks have different abilities for anomaly identification. To test these methods for geochemical anomaly identification, we based our study on stream sediment data of the Cu‐Zn‐Ag metallogenic area in southwest Fujian province as samples. Three unsupervised deep learning models: the autoencoder (AE), multi‐convolutional autoencoder (MCAE), and fusion convolutional autoencoder (FCAE), were used to extract the combined structural, spatial distribution, and mixed features of multiple‐elements. The results showed that the anomalous area delineated by the FCAE model had the best consistency with the known copper mineral occurrences, followed by the MCAE and AE models, with area under the curve values (AUC) of 0.80, 0.78, and 0.61, respectively. FCAE and AE were insensitive to changes in convolution window size, while MCAE extracted more spatial distribution or mixed features. Overall, FCAE focused more on structural distribution or mixed features, combining the advantages of both MCAE and AE. Therefore, FCAE performed best among the three deep learning methods. This study provides a practical basis for selecting and constructing geochemical anomaly recognition models based on deep learning algorithms. Plain Language Summary: The recognition of abnormal geochemical concentrations of elements is important for mineral exploration, and deep learning method is a novel way to accomplish this task. Deep learning methods can extract multifeature information from the geological environment, which effectively improves the efficiency of mineral prospecting. Here, we employed three unsupervised deep learning models to extract the integrated structure, spatial distribution, and mixed features of multiple elements. This study analyzes the characteristics of deep learning models with different network structures, thus it provides a practical basis for selecting a method to recognize abnormal geochemical concentrations of elements. Key Points: The combined structural features extracted by deep learning algorithms have a strong anti‐interference ability for noiseProvides a practical approach for constructing geochemical anomaly recognition models based on deep learning algorithmsComprehensively analyzes the characteristics of three deep learning models for geochemical anomaly identification [ABSTRACT FROM AUTHOR]

• A new sequence to sequence (seq2seq) architecture is developed for hourly runoff prediction. • The new model uses a linear layer in the decoder to process historical runoff instead of using RNN cells. • The new model predicts more accurate hourly runoff than other benchmark seq2seq models. • Prediction accuracy is further improved by combining the new model with the attention mechanism. Machine learning especially deep learning methods have been widely used for runoff prediction in recent years. Models based on the sequence-to-sequence (seq2seq) framework were recently introduced into the field of rainfall-runoff modeling. In this paper, we demonstrate a new design of a seq2seq model which shows higher prediction accuracy than previous seq2seq models. Our new design takes advantage of the high linear correlation between historical runoff and future runoff, and processes the runoff input through a linear layer. Our new seq2seq model, named seq2seq-lrp (linear runoff processing), is combined with the attention mechanism (seq2seq-lrp-a), and compared with three seq2seq benchmark models (seq2seq-1, seq2seq-2, and seq2seq-3). The five models (seq2seq-lrp, seq2seq-lrp-a, seq2seq-1, seq2seq-2, seq2seq-3) were used to simulate runoff in the Yutan sub-catchment, Fujian Province, Southeast China, using hourly runoff and rainfall measurements from 2000 to 2014. The results show that the novel seq2seq-lrp and seq2seq-lrp-a models are better suited in hourly flood prediction than the seq2seq benchmark models and the seq2seq-lrp-a model performs best among the five models. [ABSTRACT FROM AUTHOR]

Landslide susceptibility prediction has the disadvantages of being challenging to apply to expanding landslide samples and the low accuracy of a subjective random selection of non-landslide samples. Taking Fu'an City, Fujian Province, as an example, a model based on a semi-supervised framework using particle swarm optimization to optimize extreme learning machines (SS-PSO-ELM) is proposed. Based on the landslide samples, a semi-supervised learning framework is constructed through Density Peak Clustering (DPC), Frequency Ratio (FR), and Random Forest (RF) models to expand and divide the landslide sample data. The landslide susceptibility was predicted using high-trust sample data as the input variables of the data-driven model. The results show that the area under the curve (AUC) valued at the SS-PSO-ELM model for landslide susceptibility prediction is 0.893 and the root means square error (RMSE) is 0.370, which is better than ELM and PSO-ELM models without the semi-supervised framework. It shows that the SS-PSO-ELM model is more effective in landslide susceptibility. Thus, it provides a new research idea for predicting landslide susceptibility. [ABSTRACT FROM AUTHOR]

Improving prediction accuracy is a prominent modeling issue in relation to forest simulations, and ensemble learning is a new effective method for improving the precision of crown profile model simulations in order to overcome the disadvantages of statistical modeling. Background: Ensemble learning (a machine learning paradigm in which multiple learners are trained to achieve better performance) has strong nonlinear problem learning ability and flexibility in terms of analyzing longitudinal data, and it remains rarely explored so far in the field of crown profile modeling forest science. In this study, we explored the application of ensemble learning to the modeling and prediction of crown profiles. Methods: We evaluated the performance of ensemble learning procedures and marginal model in modeling crown profile using the crown profile database from China fir plantations in Fujian, in southern China. Results: The ensemble learning approach for the crown profile model appeared to have better performance and higher efficiency (R2 > 0.9). The crown equation model 18 showed an intermediate performance in its estimation, whereas GBDT (MAE = 0.3250, MSE = 0.2450) appeared to have the best performance and higher efficiency. Conclusions: The ensemble learning method can combine the advantages of multiple learners and has higher model accuracy, robustness and overall induction ability, and is thus an effective technique for crown profile modeling and prediction. [ABSTRACT FROM AUTHOR]

The coarseModerate Resolution Imaging Spectroradiometer (MODIS) aerosol optical depth (AOD) product (spatial resolution: 3 km) retrieved by the dark-target algorithm always generates the missing values when being adopted to estimate the ground-level PM2.5 concentrations. In this study, we developed a two-stage random forest using MODIS 3-km AOD to obtain the PM2.5 concentrations with full coverage in a contiguous coastal developed region, i.e., Yangtze River delta--Fujian--Pearl River delta (YRD--FJ--PRD) region of China. A first-stage random forest--integrated six meteorological fields was employed to predict the missing values of AOD product, and the combined AOD (i.e., random forest--derived AOD and MODIS 3-km AOD) incorporated with other ancillary variables were developed for predicting PM2.5 concentrations within a second-stage random forest model. The results showed that the first-stage random forest could explain 94% of the AOD variability over YRD--FJ--PRD region, and we achieved a site-based cross validation (CV) R² of 0.87 and a time-based CV R² of 0.85. The full-coverage PM2.5 concentrations illustrated a spatial pattern with annual-mean PM2.5 of 46, 40, and 35 µg m-3 in YRD, PRD, and FJ, respectively, sharing the same trend with previous studies. Our results indicated that the proposed two-stage random forest model could be effectively used for PM2.5 estimation in different areas. [ABSTRACT FROM AUTHOR]

The majority of machine learning algorithms that have been applied in data-driven predictive mapping of mineral prospectivity require a sufficient number of training samples (known mineral deposits) to obtain results with high performance and reliability. Semi-supervised learning can take advantage of the huge amount of unlabeled data to benefit the supervised learning tasks and hence provide a suitable scheme for mapping mineral prospectivity in cases where only few known mineral deposits are available. Semi-supervised random forest was used in this study to map mineral prospectivity in the southwestern Fujian metallogenic belt of China, where there is still excellent potential for mineral exploration due to the large proportion of areas covered by forest. The findings obtained from the current study include: (1) semi-supervised learning can make use of both the labeled and unlabeled samples to help improve the performance of mapping mineral prospectivity; (2) multi-dimensional scaling can be used to explore the clustering structure within the samples, which provides a tool to validate the usability of semi-supervised learning algorithms. In addition, the prospectivity map obtained in this study can be used to guide further mineral exploration in the southwestern Fujian of China. [ABSTRACT FROM AUTHOR]