Multi-source Data Research Articles

Integrative factor-adjusted sparse generalized linear models

With the increasing accessibility of data, sample sizes remain limited in most studies, integrating multiple-source datasets is thus desirable to improve model performance. The high dependence among covariates is a common characteristic of high-dimensional data. However, traditional approaches fail to achieve variable selection consistency when the covariates are highly correlated. To address the issues stemming from correlation, we propose the penalized factor-adjusted approach to reduce the correlations in the integrative analysis within a generalized linear model (GLM) framework for multi-source high-dimensional data. By utilizing the latent factors and idiosyncratic components as predictors, the proposed approach enables model estimation with weakly correlated covariates. We rigorously establish the consistency properties under certain conditions. Simulation demonstrates the superior and competitive performance of the proposed approach in estimation, prediction, and variable selection. Analysis of genetic data on prostate cancer confirms its practical usefulness.

Coseismic Deformation and Kinematic Rupture Process of the 2023 Türkiye MW7.8-7.5 Earthquake Doublet Deduced from High-rate GNSS Observations

Summary On February 6, 2023, a devastating earthquake doublet struck in Türkiye, which caused tremendous damage including about 54000 fatalities. We use high-rate GNSS data to study the coseismic static and kinematic displacement, and invert for fault slip distribution of the earthquake doublet. The maximum static coseismic displacements for the Mw7.8 (EQ1) and Mw7.5 earthquake (EQ2) are 0.38 m and 4.4 m respectively. The kinematic displacement recorded by GNSS stations reflect clearly the earthquake rupture process in more detail. We utilize 1-Hz GNSS waveforms, GNSS static displacements and surface rupture measurements to invert for the kinematic rupture model of the earthquake doublet. Our results show a rupture duration of ∼80 s, faulting length ∼350 km, and maximum slip ∼7.4 m for EQ1, and a rupture duration of ∼35 s, rupture length ∼190 km, and maximum slip ∼9.2 m for EQ2. The slip distributions inverted solely from GNSS coseismic deformation are basically consistent with the results from multi-source data. The rupture velocity of EQ1 towards east is about 4.5 km/s towards the east and 3.2 km/s towards the west. EQ2 nucleates on Çardak fault with bilateral supershear rupture velocity of 5.0∼5.5 km/s within the first 0-12 s, subsequently transitioning to subshear rupture at a speed of 1.2 km/s towards the east and 3.0 km/s towards the west. Our finding of the bilateral supershear ruptures of EQ2 contrasts with the results of unilateral westward supershear rupture of previous studies. We think that this discrepancy could be attributed to the utilization of more geodetic data of the relatively denser GNSS stations in the east flank of EQ2’s epicenter and calculating the mean velocities for individual rupture intervals in this study.

Measurement and optimization paths of the multidimensional development levels of counties in the Yellow River Basin: based on the sustainable livelihoods framework

The Yellow River Basin (YRB) faces intense man-land conflicts. However, existing studies rarely focus on the comprehensive and multidimensional development levels in the YRB, and there is a lack of refined county-level studies, making it difficult to fully support the implementation of ecological protection and high-quality development strategy in the YRB. Under the Sustainable Livelihoods Framework, the spatial distribution, differences, and correlation characteristics of the multidimensional development levels of counties in the YRB are disclosed. This is achieved by comprehensively utilizing geospatial, socio-economic, and other multi-source data in combination with methods such as the entropy weight method, the Theil index, and spatial analysis. Optimal development paths are proposed with a focus on the development types of counties. The findings are as follows: (1) Counties in the YRB have a low overall multidimensional development level, presenting a spatial distribution pattern of “high in the east and low in the west” overall. In terms of each dimension, the average value ranking of lower reach > middle reach > upper reach is observed in each dimension; the exception is the financial dimension (with the ranking of middle reach > lower reach > upper reach). (2) Differences between counties in the development indices vary across different dimensions. The differences in the multidimensional development indices and in the development indices of each dimension (except for the natural dimension) are mainly attributable to differences between counties within each reach. Differences between reaches are non-significant. (3) The multidimensional development levels of counties in the YRB and their development levels in each dimension show strong spatial correlation. And significant counties are mainly identified as LL-type (clusters of low value areas) and HH-type (clusters of high value areas). (4) Depending on the measurement results and advantageous factors of the county multidimensional development index, counties are classified into four development types: weak development type, single-dimension-led development type, potential coordinated development type, and multidimensional coordinated development type. Based on this classification, targeted optimal development paths are proposed with a focus on the specific characteristics of different development types. The research findings can provide effective scientific support for ecological protection and high-quality development in the YRB.

A global product of 150-m urban building height based on spaceborne lidar

Urban building height, as a fundamental 3D urban structural feature, has far-reaching applications. However, creating readily available datasets of recent urban building heights with fine spatial resolutions and global coverage remains a challenging task. Here, we provide a 150-m global urban building heights dataset around 2020 by combining the spaceborne lidar (Global Ecosystem Dynamics Investigation, GEDI), multi-sourced data (Landsat-8, Sentinel-2, and Sentinel-1), and topographic data. The validation results revealed that the GEDI-estimated building height samples were effective compared to the reference data (Pearson’s r = 0.81, RMSE = 3.58 m). The mapping product also demonstrated good performance, as indicated by its strong correlation with the reference data (Pearson’s r = 0.71, RMSE = 4.73 m). Compared with the currently existing datasets, it holds the ability to provide a spatial resolution (150 m) with a great level of inherent details about the spatial heterogeneity and flexibility of updating using the GEDI samples as inputs. This product will boost future urban studies across many fields, including environmental, ecological, and social sciences.

A prior-knowledge-guided dynamic attention mechanism to predict nocturnal hypoglycemic events in type 1 diabetes

Nocturnal hypoglycemia is a critical problem faced by diabetic patients. Failure to intervene in time can be dangerous for patients. The existing early warning methods struggle to extract crucial information comprehensively from complex multi-source heterogeneous data. In this paper, a deep learning framework with an innovative dynamic attention mechanism is proposed to predict nocturnal hypoglycemic events for type 1 diabetes patients. Features related to nocturnal hypoglycemia are extracted from multi-scale and multi-dimensional data, which enables comprehensive information extraction from diverse sources. Then, we propose a prior-knowledge-guided attention mechanism to enhance the network's learning capability and interpretability. The method was evaluated on a public available clinical dataset, which successfully warned 94.91% of nocturnal hypoglycemic events with an F1-score of 96.35%. By integrating our proposed framework into the nocturnal hypoglycemia early warning model, issues related to feature redundancy and incompleteness were mitigated. Comparative analysis demonstrates that our method outperforms existing approaches, offering superior accuracy and practicality in real-world scenarios.

Target Detection and Characterization of Multi-Platform Remote Sensing Data

Detecting targets in remote sensing imagery, particularly when identifying sparsely distributed materials, is crucial for applications such as defense, mineral exploration, agriculture, and environmental monitoring. The effectiveness of detection and the precision of the results are influenced by several factors, including sensor configurations, platform properties, interactions between targets and their background, and the spectral contrast of the targets. Environmental factors, such as atmospheric conditions, also play a significant role. Conventionally, target detection in remote sensing has relied on statistical methods that typically assume a linear process for image formation. However, to enhance detection performance, it is critical to account for the geometric and spectral variabilities across multiple imaging platforms. In this research, we conducted a comprehensive target detection experiment using a unique benchmark multi-platform hyperspectral dataset, where man-made targets were deployed on various surface backgrounds. Data were collected using a hand-held spectroradiometer, UAV-mounted hyperspectral sensors, and airborne platforms, all within a half-hour time window. Multi-spectral space-based sensors (i.e., Worldview and Landsat) also flew over the scene and collected data. The experiment took place on 23 July 2021, at the Rochester Institute of Technology’s Tait Preserve in Penfield, NY, USA. We validated the detection outcomes through receiver operating characteristic (ROC) curves and spectral similarity metrics across various detection algorithms and imaging platforms. This multi-platform analysis provides critical insights into the challenges of hyperspectral target detection in complex, real-world landscapes, demonstrating the influence of platform variability on detection performance and the necessity for robust algorithmic approaches in multi-source data integration.

Mapping land cover in the low Arctic using multiple sources of earth observation data

Land cover maps in high latitude regions are important for permafrost modeling and mapping and other climate change related studies. National and global land cover products cannot meet the requirements for regional permafrost mapping due to their broad land cover types. This study produced a land cover map in a low Arctic region using multi-source earth observation data, including different modes of RADARSAT-2 (RS2), Sentinel 2 (S2), and the high resolution ArcticDEM. We assessed the performances of two machine learning classification algorithms (Random Forest (RF) and Support Vector Machine (SVM)) with different combinations of input data, including evaluating the effects of incidence angle from different RS2 modes on classification results. This study concludes that the combination of two imaging modes of RS2, S2 composites and the high resolution ArcticDEM achieves a more accurate land cover map (the overall accuracy is 93.6%) than other combinations with the overall accuracies ranging from 38% to 87%; the performance of RF is better than that of SVM, the overall accuracy difference ranges from 21% to 1%; and the shallow SAR incidence angles outperform steep ones in classification results of different combinations for both classifiers.

Urban Greenway Planning and Designing Based on MGWR and the Entropy Weight Method

Travelers’ attention to high-quality human habitats is increasing, and the role of urban greenways in improving the quality of travelling spaces has also been appreciated. This research aims at making the weight calculation of suitability more scientific and reasonable, clustering the shared bicycle travelling OD points according to suitability, and analyzing the distribution of OD points. Taking Xiamen as an example, multiscale geographically weighted regression and entropy weight methods were used to calculate the weights of variables using multi-source big data. The clustering of origin-destination (OD) points for shared bicycle travel are identified using the DBSCAN clustering algorithm, which can provide accurate support for greenway planning and shared bicycle placement. The results show that the density of tourist attractions, POI entropy index, road density, and intermediate are four important factors affecting the suitability of greenways. The clustering results of the shared bicycle OD points show that the high-aggregation areas of origin and destination points are located in the northeast and southwest directions as well as west and east directions. This study provides a theoretical and modelling analysis reference for greenway planning and design.

Research on rolling bearing fault diagnosis method based on improved multi-source fusion convolutional neural network

Abstract As sensor technology advances, the variety and number of sensors increase, leading to the capture of more signals. Existing multi-source fusion methods often face issues such as increased model complexity or the failure to fully utilize the potential correlations among multi-sensor data, thereby affecting the accuracy and reliability of fault diagnosis. To address this issue, this paper proposes a multi-source fusion convolutional neural network (MFCNN) that diagnoses bearing faults by integrating features from multi-source signals. Firstly, multiple convolution blocks with gradually increasing one-dimensional kernel sizes are utilized to extract features from the integrated multi-source data. This approach enhances feature extraction efficiency and simplifies the network architecture. Secondly, a feature fusion based on the CBAM attention mechanism is proposed, which refines feature representation through channel and spatial attention modules. This makes the model more focused on important information, thereby improving recognition accuracy. The diagnostic capabilities of the proposed MFCNN are evaluated utilizing two datasets.

Spatiotemporal Variation and Driving Mechanisms of the Global Production-Living-Ecological Space Coupling Coordination Degree

The coupling coordination degree (CCD) of the production-living-ecological space (PLES) functional index is an indicator of regional sustainable development potential. However, previous studies have failed to reveal the driving mechanisms of the CCD of PLES functional index on a global scale. Therefore, this study employed the CCD model to evaluate the CCD of the PLES functional index and spatial regression models to measure the heterogeneous drivers using multi-source data in 2000, 2010, and 2020. The results demonstrated that ecological spaces dominated (85%) globally, while living spaces comprised the smallest share (3%). The ecological functional index was higher than the production and living functional indices. Further analyses revealed that topographic factors were the main restricting factors for PLES; the proportion of production and living spaces decreased with increasing altitude and slope, whereas ecological spaces showed the opposite trend. The global CCD values of the PLES functional index in 2000, 2010, and 2020 were 0.186, 0.189, and 0.198, respectively, showing an increasing trend. High CCD was generally observed in areas with dense population and industry where human activity systems interact with natural ecosystems. The formation of the CCD of the global PLES functional pattern results from the joint action of natural and socioeconomic factors, with pronounced spatial heterogeneity. Our findings can help optimize global territorial space utilization, improve territorial space utilization efficiency, and realize global sustainable development goals.

Spatial Pattern and Influencing Factors of Tourist Attractions in Coastal Cities: A Case Study of Qingdao

The spatial distribution of tourist attractions plays a critical role in the development of coastal cities. Qingdao, with its coastal geography, rich cultural heritage, and rapid urbanization, serves as a representative case. This study integrates POI and multi-source data, employing methods such as the average nearest neighbor index, kernel density estimation, standard deviational ellipse, and Geodetector to analyze the spatial characteristics and influencing factors of Qingdao’s tourist attractions. Additionally, path dependence theory is innovatively applied to elucidate the mechanisms of the city’s development trajectory. Both natural and social factors influence this distribution, where the resource environment forms the foundational basis, the economic development provides impetus, and the urban development orientation exerts a regulatory effect. The findings are broadly applicable to other coastal tourist cities and offer strategic insights for sustainable development in such contexts.

Data mining techniques for LULC analysis using sparse labels and multisource data integration for the hilly terrain of Nilgiris district, Tamil Nadu, India

Data mining techniques for LULC analysis using sparse labels and multisource data integration for the hilly terrain of Nilgiris district, Tamil Nadu, India

Probabilistic Seismic Hazard Assessment for the North China Plain Earthquake Belt: Sensitivity of Seismic Source Models and Ground Motion Prediction Equations

AbstractIn this study, a multi-source data fusion method was proposed for the development of a Hybrid seismic hazard model (HSHM) in China by using publicly available data of the 5th Seismic Ground Motion Parameter Zoning Map (NSGM) and historical seismic catalogues and integrating with modern ground motion prediction equations (GMPEs). This model incorporates the characteristics of smoothed seismicity and areal sources for regional seismic hazard assessment. The probabilistic seismic hazard for the North China Plain earthquake belt was investigated through sensitivity analysis related to the seismicity model and GMPEs. The analysis results indicate that the Hybrid model can produce a consistent result with the NSGM model in many cases. However, the NSGM model tends to overestimate hazard values in locations where no major events have occurred and underestimate hazard values in locations where major events have occurred. The Hybrid model can mitigate the degree of such biases. Compared to the modern GMPEs, the GMPE with epicentral distance measures significantly underestimate the seismic hazard under near-field and large-magnitude scenarios. In addition, a comparison of the uniform hazard spectra (UHS) obtained by the models, with China’s design spectrum, shows that the current design spectrum is more conservative than the calculated UHS.

Mapping Lithology with Hybrid Attention Mechanism–Long Short-Term Memory: A Hybrid Neural Network Approach Using Remote Sensing and Geophysical Data

Remote sensing provides an efficient roadmap in geological analysis and interpretation. However, some challenges arise when remote sensing techniques are integrated with machine learning in geological surveys. Factors including irregular spatial distribution, sample imbalance, interclass resemblances, regolith, and geochemical similarities impede geological feature diagnosis, interpretation, and identification across varied remote sensing datasets. To address these limitations, a hybrid-attention-integrated long short-term memory (LSTM) network is employed to diagnose, interpret, and identify lithological feature representations in a remote sensing-based geological analysis using multisource data fusion. The experimental design integrates varied datasets including Sentinel-2A, Landsat-9, ASTER, ALOS PALSAR DEM, and Bouguer anomaly gravity data. The proposed model incorporates a hybrid attention mechanism (HAM) comprising channel and spatial attention submodules. HAM utilizes an adaptive technique that merges global-average-pooled features with max-pooled features, enhancing the model’s accuracy in identifying lithological units. Additionally, a channel separation operation is employed to allot refined channel features into clusters based on channel attention maps along the channel dimension. The comprehensive analysis of results from comparative extensive experiments demonstrates HAM-LSTM’s state-of-the-art performance, outperforming existing attention modules and attention-based models (ViT, SE-LSTM, and CBAM-LSTM). Comparing HAM-LSTM to baseline LSTM, the HAM module’s integrated configurations equip the proposed model to better diagnose and identify lithological units, thereby increasing the accuracy by 3.69%.

Entity-Driven New Paradigm of Mine Data: Model Construction and Application

In the construction of a three-dimensional real scene (3D real scene) of a mine, we encounter challenges in organizing and correlating multi-source data. To surmount these challenges, we innovatively adopted the entity organization method. We proposed a definition of a “mining entity” and conducted an in-depth analysis of its main characteristics. Based on these characteristics, we classified and coded the main mining entities and further constructed an entity expression pattern layer that could fully reflect the characteristics of the mine. Based on multi-source heterogeneous mine data, we constructed a mining entity data layer. Furthermore, by adopting a graph database approach, we were able to create a graphical representation and correlation application of the complex relationships among mining entities. This work shows that compared to traditional data management methods, a mine data organization approach centered on entities can more effectively integrate and correlate multi-source mine data, providing strong support for the digital management of mines and the construction of 3D real scenes.

Assessment of the Synergistic Development Status of Nature-Based Tourism in the National Parks Agglomeration of the Qinghai–Tibet Plateau

The national parks agglomeration in China is presently undergoing a pivotal phase of planning and construction. Synergistic development of nature-based tourism within the Qinghai–Tibet Plateau National Parks Agglomeration is essential for optimizing the spatial configuration of regional functions and substantiating the necessity of the agglomeration development of national parks. This study develops an assessment framework for the synergistic development of nature-based tourism based on the interconnection of elements. By constructing a model of nature-based tourism elements’ interconnections and employing social network analysis, it investigates the characteristics and status of synergistic development in nature-based tourism within the Qinghai–Tibet Plateau National Parks Agglomeration, while also proposing pathways for optimization. The findings reveal that: (1) By integrating multi-source heterogeneous data on resource complementarity, product connectivity, tourist source sharing, and information linkage, a robust assessment framework for the synergistic development of nature-based tourism in the national parks agglomeration was established. (2) The synergistic development of nature-based tourism within the national parks agglomeration on the Qinghai–Tibet Plateau demonstrates spatial heterogeneity. Shangri-La National Park and Qinghai Lake National Park serve as the core of the tourism flow network, whereas Mount Kailash National Park, Gaoligong Mountain National Park, and Pamir–Kunlun Mountains National Park exhibit the least favorable states of coordinated development. (3) Utilizing social network analysis, this study classified the national parks into core and peripheral positions within the nature-based tourism synergy network and proposed corresponding optimization strategies. The findings of this research contribute to optimizing the overall layout of the Qinghai–Tibet Plateau national parks agglomeration. Additionally, they provide theoretical and methodological references for the nature-based tourism synergistic development in other national park agglomerations.

Online tensile strength prediction considering geometric and physical characteristics during FSW 2219-T8 aluminum alloy plate

Tensile strength of joints of friction stir welded (FSW) thick 2219-T8 aluminum alloy plates, is affected by multiple geometric and physical characteristics, such as axial force, welding temperature, gap, and mismatch of the butt face, which makes it difficult to predict the tensile strength with high precision on line. At present, simple welding process prediction based on temperature or force is difficult to achieve accurate control of complex welding process. In this paper, a multi-information fused one-dimensional Convolutional Neural Network (1DCNN), combining geometric and physical characteristics of welding process, was proposed to predict the tensile strength of joints. Firstly, the experiments of FSW thick 2219-T8 aluminum alloy plates were conducted and the axial force, welding temperature, gap and mismatch of the butt face are measured. Then, the measured multi-source data was fused through down-sampling technology. Before training, the Particle Swarm Optimization (PSO) and Long Short-Term Memory (LSTM) were adopted to optimize the initial learning rate and layer structure of the model. The experiment results showed that the proposed model can accurately and quickly predict the tensile strength of thick 2219-T8 aluminum alloy joints with the mean absolute percentage error less than 2%. In addition, metallographic analysis revealed that the joint tensile strength is interlocked with weld defects and fine grain strengthening. The research lays a foundation for welding quality control of FSW.

Unified representation aggregated from graph and global features for multi-source data clustering

Multi-source data has received increasing attention due to its excellent performance in clustering tasks. However, existing multi-source data clustering methods utilize shallow graph learning methods to model similarity graphs of multi-source data, and ignore the importance of weak connections between samples within the same cluster. Thus, these connections fail to be explored in the global similarity graph of multi-source data. In this paper, we proposed a unified representation aggregated from graph and global features (UGGF) method for multi-source data clustering. Specifically, it utilizes the cross-source graph diffusion process to preserve invariant connections in multi-source similarity graphs and preserve weak connections between samples through their connection relationships between different sources. Furthermore, inspired by self-attention, the encoder of transformers is used to learn multi-source global feature representation. Then, global features and similarity graphs are integrated to comprehensively explore the multi-source data, aiming to obtain a unified representation for the final clustering task. The proposed UGGF method is validated on four benchmark datasets. Extensive experimental results demonstrate the superiority of the proposed method compared with other state-of-the-art multi-source data clustering methods.

Analysis of Economic Vitality and Development Equilibrium of China’s Three Major Urban Agglomerations Based on Nighttime Light Data

Eliminating poverty, reducing inequality, and achieving balanced development are one of the United Nations Sustainable Development Goals. Objectively and accurately measuring regional economic vitality and development equilibrium is a pressing scientific issue that needs to be addressed in order to achieve common prosperity. Nighttime light (NTL) remote sensing data have been proven to be a good proxy variable for socio-economic development, and are widely used due to their advantages of convenient access and wide spatial coverage. Based on multi-source data, this study constructs an Economic Development Index (EDI) that comprehensively reflects regional economic vitality from two aspects, economic quality and development potential, combines the Nighttime Light Development Index (NLDI) as the evaluation indicators to measure the economic vitality and development equilibrium, analyzes the economic vitality and development equilibrium of 300 district and county units in China’s three major urban agglomerations from 2000 to 2020 and their temporal and spatial variation characteristics, and discusses the connotation of EDI and its availability. The results show the following: (1) From 2000 to 2020, the average growth rate of EDI in China’s three major urban agglomerations reached 36.32%, while the average decrease rate of NLDI reached 38.75%; both economic vitality and the development equilibrium have been continuously enhanced. Among them, the Yangtze River Delta (YRD) urban agglomeration experienced the fastest economic growth, while the Guangdong–Hong Kong–Macao Greater Bay Area (GBA) exhibited the strongest economic strength. (2) Both economic vitality and the development equilibrium in these three urban agglomerations exhibited distinct spatial agglomeration characteristics, namely center-surrounding distribution, coastal–inland distribution, and radial belt–pole distribution, respectively. (3) Over the past two decades, the economic development of these three urban agglomerations has progressed towards the pattern of regional coordinated development, pole-driven development and urban–rural integrated development. The research results can provide new research perspectives and scientific support for promoting regional balanced development, achieving sustainable development goals, and reducing inequality.

Change in aerosol optical depth in Chinese urban agglomerations during the early phase of dual-carbon actions

Dual-carbon policies were implemented by Chinese government to mitigate climate warming; however, changes in aerosol optical depth (AOD) during early phases of these actions (2020-2022) remain unclear. Thus, AOD variations during this period were investigated compared to the baseline (2015-2019 mean) across seven urban agglomerations (UAs) using multi-source data. Significant temporal variations in AOD anomalies (ΔAOD) were observed at annual and seasonal scales, with varying magnitudes. On average, AOD decrease by -0.210, -0.225, -0.180, and -0.038 in Beijing-Tianjin-Hebei (BTH), Yangtze River Delta (YRD), Guanzhong Plain (GZP), and Liaozhongnan (LZN), respectively. In contrast, increases of 0.041 and 0.057 were observed in Pearl River Delta (PRD) and Tianshan Mountains northern foothills (TSBP), while Lhasa metropolitan area (LS) remain stable (0.0001). In 2021, the most significant decrease in AOD was recorded in YRD, whereas the greatest declines in other urban areas were noted in 2022. The largest ΔAOD in LZN, BTH, and YRD occurred in autumn, while the peak ΔAOD in GZP was observed in spring. In contrast, the highest ΔAOD in PRD, TSBP, and LS occurred in summer. These temporal variations in ΔAOD can be attributed to dual-carbon policies, natural and meteorological anomalies, along with the impacts of stringent lockdown policies and abrupt dust events. Path analysis indicated that the effects of various drivers on ΔAOD were interconnected. When interactions among these factors were considered, their influences on ΔAOD were found to be either amplified or altered, highlighting the importance of accounting for interactions within the human-nature-meteorological system in formulation of atmospheric pollution policies. These findings contribute to a deeper understanding of ΔAOD variation mechanisms and provide valuable insights for the development of air pollution reduction strategies at the UA scale.