Search Results (23)

Against the backdrop of smart cities and digital cultural tourism, the accurate prediction of urban passenger flow is of great significance for public security management and resource allocation. However, existing studies mostly rely on single data sources or only perform a simple concatenation of multi-source features, lacking systematic indicator system design. Meanwhile, weekly or monthly data are commonly used with coarse temporal granularity, making it difficult to capture short-term fluctuations and lag effects. To overcome these limitations, this paper collects the daily passenger flow data of Hangzhou from 15 March 2024 to 15 March 2025; integrates multi-dimensional factors such as keyword search trends across platforms, holidays and major events, and online public opinion; and constructs three daily characteristic indicators: online search index, humanistic–meteorological index, and textual sentiment index. The data denoising, dimensionality reduction, and sentiment quantification are realized through methods including SSA, PCA, and SnowNLP. On this basis, a hybrid CNN-GRU model integrated with the attention mechanism is proposed. An improved artificial bee colony (ABC) algorithm is adopted for global hyperparameter optimization, and a weighted hybrid loss function (JQHL) is introduced to enhance the model’s adaptability to extreme values. The results show that the ABC-CNN-GRU-Attention model, incorporating multi-dimensional indicators, outperforms traditional methods on evaluation metrics, including MAE, RMSE, MAPE, R², and RPD, demonstrating a higher prediction accuracy and robustness. Full article

The alpine grassland ecosystem in the Source Region of the Yellow River (SRYR) faces the dual pressures of ecological protection and economic development. Its ecological fragility and climate sensitivity make local animal husbandry susceptible to meteorological disasters. To overcome adverse selection and moral hazard in traditional animal husbandry insurance, this study integrates 963 field sampling observation data, over 400 valid herdsmen survey data, and long-term environmental time series variables. A random forest model (R² = 0.59, RMSE = 65.84 g/m², superior to the artificial neural network in this paper) was used to estimate grass yield. Hodrick–Prescott (HP) filtering was used to separate meteorological yield per unit area and derive yield loss rate. A joint distribution model of meteorological indicators and loss rate was constructed using a Copula function to capture tail-dependent structures, providing a basis for determining trigger thresholds and actuarial pricing of pure insurance premiums. The study reveals the transmission mechanism of climate disasters to feeding costs and designs regional drought and snow disaster index insurance. The compensation standard is based on meteorological indicators falling below the trigger threshold and a yield reduction rate greater than 5%. Using 10,000 Monte Carlo simulations, the drought premium rates for zones I-IV are determined to be 2.03–6.03%, and the snow premium rates to be 2.25–5.42%, corresponding to a premium of RMB 5.21–9.61 per mu for drought and RMB 5.78–8.64 per mu for snow. This design reduces basis risk through zoning and composite triggering, providing a scientific tool for climate risk management in alpine grasslands. Full article

Globally, wetlands have suffered severe degradation due to natural environmental changes and human activities. The wetlands on the Qinghai–Tibet Plateau (QTP) play a unique and critical ecological role, making it essential to understand their spatiotemporal dynamics and driving forces for effective conservation. Based on multi-source remote sensing data and Partial Least Squares Structural Equation Modeling (PLS-SEM), this study comprehensively quantified the spatiotemporal changes in wetlands and their key driving factors on the QTP from 1990 to 2020. The results show a net increase in total wetland area (including both natural and artificial wetlands) of approximately 538.72 km² per year over the 30-year period. Spatially, wetland expansion was most pronounced in the central–western and northern parts of the plateau, primarily driven by the conversion of grasslands, barren lands, and snow/ice cover, while localized degradation persisted in eastern regions. The PLS-SEM demonstrated an excellent fit (R² = 0.962) and identified human activities—such as ecological restoration policies and infrastructure development—as the dominant direct driver of wetland expansion (path coefficient = 0.918). Climate change, improved vegetation cover, and cryospheric loss also contributed positively to wetland gains (path coefficients = 0.056, 0.044, and 0.138, respectively). This study provides a transferable framework for understanding complex wetland dynamics and their drivers in alpine regions under global environmental change, which is crucial for designing more effective wetland conservation strategies. Full article

Aircraft icing and snow accumulation are significant threats to flight safety and operational efficiency, necessitating rapid and accurate detection methods. The aim of this study was to develop and comparatively evaluate artificial intelligence (AI) models for the real-time detection of ice and snow on aircraft surfaces using vision systems. A custom dataset of annotated aircraft images under various winter conditions was prepared and augmented to enhance model robustness. Two training approaches were implemented: an automatic process using the YOLOv8 framework on the Roboflow platform and a manual process in the Google Colab environment. Both models were evaluated using standard object detection metrics, including mean Average Precision (mAP) and mAP@50:95. The results demonstrate that both methods achieved comparable detection performance, with final mAP50 values of 0.25–0.3 and mAP50-95 values around 0.15. The manual approach yielded lower training losses and more stable metric progression, suggesting better generalization and a reduced risk of overfitting. The findings highlight the potential of AI-driven vision systems to support intelligent de-icing decision-making in aviation. Future research should focus on refining localization, minimizing false alarms, and adapting detection models to specific aircraft components to further enhance operational safety and reliability. Full article

High-quality Earth Observation (EO) imagery is essential for accurate analysis and informed decision making across sectors. However, data scarcity caused by atmospheric conditions, seasonal variations, and limited geographical coverage hinders the effective application of Artificial Intelligence (AI) in EO. Traditional data augmentation techniques, which rely on basic parameterized image transformations, often fail to introduce sufficient diversity across key semantic axes. These axes include natural changes such as snow and floods, human impacts like urbanization and roads, and disasters such as wildfires and storms, which limits the accuracy of AI models in EO applications. To address this, we propose a four-stage data augmentation approach that integrates diffusion models to enhance semantic diversity. Our method employs meta-prompts for instruction generation, vision–language models for rich captioning, EO-specific diffusion model fine-tuning, and iterative data augmentation. Extensive experiments using four augmentation techniques demonstrate that our approach consistently outperforms established methods, generating semantically diverse EO images and improving AI model performance. Full article

The Tianshan Expressway plays a crucial role in China’s “Belt and Road” strategy, yet the extreme climate of the Tianshan Mountains poses significant traffic safety risks, hindering local economic development. Efficient detection of hazardous road surface conditions (RSCs) is vital to address these challenges. The complexity and variability of RSCs in the region, exacerbated by harsh weather, make traditional surveillance methods inadequate for real-time monitoring. To overcome these limitations, a vision-based artificial intelligence approach is urgently needed to ensure effective, real-time detection of dangerous RSCs in the Tianshan road network. This paper analyzes the primary structures and architectures of mainstream neural networks and explores their performance for RSC recognition through a comprehensive set of experiments, filling a research gap. Additionally, T-Net, specifically designed for the Tianshan Expressway engineering project, is built upon the optimal architecture identified in this study. Leveraging the split-transform-merge structure paradigm and asymmetric convolution, the model excels in capturing detailed information by learning features across multiple dimensions and perspectives. Furthermore, the integration of channel, spatial, and multi-head attention modules enhances the weighting of key features, making the T-Net particularly effective in recognizing the characteristics of snow-covered and icy road surfaces. All models presented in this paper were trained on a custom RSC dataset, compiled from various sources. Experimental results indicate that the T-Net outperforms fourteen once state-of-the-art (SOTA) models and three models specifically designed for RSC recognition, with 97.44% accuracy and 9.79% loss on the validation set. Full article

In the railway sector, rolling stock and infrastructure must be maintained in perfect condition to ensure reliable and safe operation for passengers. Climate change is affecting the urban and regional infrastructure through sea level rise, water accumulations, river flooding, and other increased-frequency extreme natural situations (heavy rains or snows) which pose a challenge to maintenance. In this paper, the use of artificial intelligence based on predictive maintenance implementation is proposed for the early detection of degraded conditions of a bridge due to extreme climatic conditions. For this prediction, continuous monitoring is proposed, with the aim of establishing alarm thresholds to detect dangerous situations, so restrictions could be determined to mitigate the risk. However, one of the main challenges for railway infrastructure managers nowadays is the high cost of monitoring large infrastructures. In this work, a methodology for monitoring railway infrastructures to define the optimal number of transductors that are economically viable and the thresholds according to which infrastructure managers can make decisions concerning traffic safety is proposed. The methodology consists of three phases that use the application of machine learning (Random Forest) and artificial cognitive systems (LSTM recurrent neural networks). Full article

For artificial snowfall, snow particle size can have a direct impact on snow quality. The operating conditions of the snow-makers and environmental factors will influence the atomization and crystallization processes of artificial snow making, which consequently affect snow particle size. This paper investigates the size distribution of snow particles during artificial snow making under different operating conditions and environmental parameters. For this purpose, an environmental chamber is designed and structured. The laser scattering method was used to measure the size distribution of snow under different parameters in the room. The results show that the distribution of snow crystal particle size aligns closely with the Rosin–Rammler (R-R) distribution. The higher the height of the snowfall, the longer the snow crystals grow and the larger the snow crystal particle size. It has been found that a higher air pressure favors atomization, while the opposite is true for water pressure, which results in a higher air–water pressure ratio, producing smaller snow particle sizes. Additionally, an ambient temperature in the range of −5 °C to −15 °C contributes to the snow crystal form transforming from plates to columns and then back to plates; the snow particle size first decreases and then increases. Snow crystal particles at −10 °C have the smallest size. Outdoor snow-makers should be operated at the highest possible air–water pressure ratio and snow height, and at a suitable ambient temperature. Full article

There are many mountain torrent disasters caused by melting icebergs and snow in Xinjiang, which are very different from traditional mountain torrent disasters. Most of the areas affected by snowmelt are in areas without data, making it very difficult to predict and warn of disasters. Taking the Lianggoushan watershed at the southern foot of Boroconu Mountain as the research subject, the key factors were screened by Pearson correlation coefficient and the factor analysis method, and the data of rainfall, water level, temperature, air pressure, wind speed, and snow depth were used as inputs, respectively, with support vector regression (SVR), random forest (RF), k-nearest neighbor (KNN), artificial neural network (ANN), recurrent neural network (RNN), and long short-term memory neural network (LSTM) models used to simulate the daily average water level at the outlet of the watershed. The research results showed that the root mean square error (RMSE) values of SVR, RF, KNN, ANN, RNN, and LSTM in the training period were 0.033, 0.012, 0.016, 0.022, 0.011, and 0.010, respectively, and in the testing period they were 0.075, 0.072, 0.071, 0.075, 0.075, and 0.071, respectively. The performance of LSTM was better than that of other models, but it had more hyperparameters that needed to be optimized. The performance of RF was second only to LSTM; it had only one hyperparameter and was very easy to determine. The RF model showed that the simulation results mainly depended on the average wind speed and average sea level pressure data. The snowmelt model based on machine learning proposed in this study can be widely used in iceberg snowmelt warning and forecasting in ungauged areas, which is of great significance for the improvement of mountain flood prevention work in Xinjiang. Full article

The accuracy of soil moisture retrieval based on traditional microwave remote sensing models in the Qinghai Tibet Plateau (QTP) is unstable due to its unique plateau climate. However, considering the impact of multiple multi-scale factors effectively improves the accuracy and stability of soil moisture inversion. This article uses Sentinel-1 and seasonal climate data to analyze factors and influencing mechanisms of soil moisture in the QTP. First, an artificial neural network (ANN) was used to conduct a significance analysis to screen significant influencing factors to reduce the redundancy of the experimental design and insert information. Second, the normalization effect of each factor on the soil moisture inversion was determined, and the factors with significant normalization influences were input to fit the model. Third, different fitting methods combined the semi-empirical models for soil moisture inversion. The decision tree Chi-square Automatic Interaction Detector (CHAID) analyzed the model accuracy, and the Pearson correlation coefficient between the sample and measured data was tested to further validate the accuracy of the results to obtain an optimized model that effectively inverts soil moisture. Finally, the influencing mechanisms of various factors in the optimization model were analyzed. The results show that: (1) The terrain factors, such as elevation, slope gradient, aspect, and angle, along with climate factors, such as temperature and precipitation, all have the greatest normalized impact on soil moisture in the QTP. (2) For spring (March), summer (June), and autumn (September), the greatest normalized factor of soil moisture is the terrain factor. In winter (December), precipitation was the greatest factor due to heavy snow cover and permafrost. (3) Analyzing the impact mechanism from various factors on the soil moisture showed a restricted relationship between the inversion results and the accuracy of the power fitting model, meaning it is unsuitable for general soil moisture inversion. However, among the selected models, the accuracy of the linear fit was generally higher than 79.2%, the Pearson index was greater than 0.4, and the restricted relationship between the inversion results and accuracy was weak, making it suitable for the general inversion of soil moisture in the QTP. Full article

Snow depth estimation with passive microwave (PM) remote sensing is challenged by spatial variations in the Earth’s surface, e.g., snow metamorphism, land cover types, and topography. Thus, traditional static snow depth retrieval algorithms cannot capture snow thickness well. In this study, we present a new operational retrieval algorithm, hereafter referred to as the pixel-based method (0.25° × 0.25° grid-level), to provide more accurate and nearly real-time snow depth estimates. First, the reference snow depth was retrieved using a previously proposed model in which a microwave snow emission model was coupled with a machine learning (ML) approach. In this process, an effective grain size (effGS) value was optimized by utilizing the snow microwave emission model, and then the nonlinear relationship between snow depth and multiple predictive variables, e.g., effGS, longitude, elevation, and brightness temperature (Tb) gradients, was established with the ML technique to retrieve reference snow depth data. To select a robust and well-performing ML approach, we compared the performance of widely used support vector regression (SVR), artificial neural network (ANN) and random forest (RF) algorithms over China. The results show that the three ML models performed similarly in snow depth estimation, which was attributed to the inclusion of effGS in the training samples. In this study, the RF model was used to retrieve the snow depth reference dataset due to its slightly stronger robustness according to our comparison of results. Second, the pixel-based algorithm was built based on the retrieved reference snow depth dataset and satellite Tb observations (18.7 GHz and 36.5 GHz) from Advanced Microwave Scanning Radiometer 2 (AMSR2) during the 2012–2020 period. For the pixel-based algorithm, the fitting coefficients were achieved dynamically pixel by pixel, making it superior to the traditional static methods. Third, the built pixel-based algorithm was verified using ground-based observations and was compared to the AMSR2, GlobSnow-v3.0, and ERA5-land products during the 2012–2020 period. The pixel-based algorithm exhibited an overall unbiased root mean square error (unRMSE) and R² of 5.8 cm and 0.65, respectively, outperforming GlobSnow-v3.0, with unRMSE and R² values of 9.2 cm and 0.22, AMSR2, with unRMSE and R² values of 18.5 cm and 0.13, and ERA5-land, with unRMSE and R² values of 10.5 cm and 0.33, respectively. However, the pixel-based algorithm estimates were still challenged by the complex terrain, e.g., the unRMSE was up to 17.4 cm near the Tien Shan Mountains. The proposed pixel-based algorithm in this study is a simple and operational method that can retrieve accurate snow depths based solely on spaceborne PM data in comparatively flat areas. Full article

More than half of the ski resorts in North America have closed since the early building booms—many facing a warming climate and pressures to find water to make artificial snow. Researching and documenting all resorts between 1969–2019, we find that 59% of all resorts in North America have closed since the resort boom of the 1960s and 70s (65% in the United States, 31% in Canada). This shift has left some states or provinces with only one or no resorts remaining. To proactively persevere with a variable climate, less water, and a need for more energy to make snow, we suggest mountains holistically plan for sustainability. Recommendations include third party environmental certification, commitment to sustainability at the management level, communication to customers about sustainability practices and implementing unique models for remaining open and competitive. These practices include resort consolidation, multi-mountain passes, and/or unique ownership models. We believe that ski resorts must focus on positive environmental practices, sustainability planning, and climate change adaptation if they want to remain viable and competitive in the coming decades. Full article

The VIIRS day/night band (DNB) high gain stage (HGS) pixel effective dwell time is in the range of 2–3 milliseconds (ms), which is about one third of the flicker cycle present in lighting powered by alternating current. Thus, if flicker is present, it induces random fluctuations in nightly DNB radiances. This results in increased variance in DNB temporal profiles. A survey of flicker characteristics conducted with high-speed camera data collected on a wide range of individual luminaires found that the flicker is most pronounced in high-intensity discharge (HID) lamps, such as high- and low-pressure sodium and metal halides. Flicker is muted, but detectable, in incandescent luminaires. Modern light-emitting diodes (LEDs) and fluorescent lights are often nearly flicker-free, thanks to high-quality voltage smoothing. DNB pixel footprints are about half a square kilometer and can contain vast numbers of individual luminaires, some of which flicker, while others do not. If many of the flickering lights are drawing from a common AC supplier, the flicker can be synchronized and leave an imprint on the DNB temporal profile. In contrast, multiple power supplies will throw the flickering out of synchronization, resulting in a cacophony with less radiance fluctuation. The examination of DNB temporal profiles for locations before and after the conversion of high-intensity discharge (HID) to LED streetlight conversions shows a reduction in the index of dispersion, calculated by dividing the annual variance by the mean. There are a number of variables that contribute to radiance variations in the VIIRS DNB, including the view angle, cloud optical thickness, atmospheric variability, snow cover, lunar illuminance, and the compilation of temporal profiles using pixels whose footprints are not perfectly aligned. It makes sense to adjust the DNB radiance for as many of these extraneous effects as possible. However, none of these adjustments will reduce the radiance instability introduced by flicker. Because flicker is known to affect organisms, including humans, the development of methods to detect and rate the strength of flickering from space will open up new areas of research on the biologic impacts of artificial lighting. Over time, there is a trend towards the reduction of flicker in outdoor lighting through the replacement of HID with low-flicker LED sources. This study indicates that the effects of LED conversions on the brightness and steadiness of outdoor lighting can be analyzed with VIIRS DNB temporal profiles. Full article

Road surface detection is important for safely driving autonomous vehicles. This is because the knowledge of road surface conditions, in particular, dry, wet, and snowy surfaces, should be considered for driving control of autonomous vehicles. With the rise of deep learning technology, road surface detection methods using deep neural networks (DNN) have been widely used for developing road surface detection algorithms. To apply DNN in road surface detection, the dataset should be large and well-balanced for accurate and robust performance. However, most of the images of road surfaces obtained through usual data collection processes are not well-balanced. Most of the collected surface images tend to be of dry surfaces because road surface conditions are highly correlated with weather conditions. This could be a challenge in developing road surface detection algorithms. This paper proposes a method to balance the imbalanced dataset using CycleGAN to improve the performance of a road surface detection algorithm. CycleGAN was used to artificially generate images of wet and snow-covered roads. The road surface detection algorithm trained using the CycleGAN-augmented dataset had a better IoU than the method using imbalanced basic datasets. This result shows that CycleGAN-generated images can be used as datasets for road surface detection to improve the performance of DNN, and this method can help make the data acquisition process easy. Full article

The switch and crossing (S&C) is one of the most important parts of the railway infrastructure network due to its significant influence on traffic delays and maintenance costs. Two central questions were investigated in this paper: (I) the first question is related to the feasibility of exploring the vibration data for wear size estimation of railway S&C and (II) the second one is how to take advantage of the Artificial Intelligence (AI)-based framework to design an effective early-warning system at early stage of S&C wear development. The aim of the study was to predict the amount of wear in the entire S&C, using medium-range accelerometer sensors. Vibration data were collected, processed, and used for developing accurate data-driven models. Within this study, AI-based methods and signal-processing techniques were applied and tested in a full-scale S&C test rig at Lulea University of Technology to investigate the effectiveness of the proposed method. A real-scale railway wagon bogie was used to study different relevant types of wear on the switchblades, support rail, middle rail, and crossing part. All the sensors were housed inside the point machine as an optimal location for protection of the data acquisition system from harsh weather conditions such as ice and snow and from the ballast. The vibration data resulting from the measurements were used to feed two different deep-learning architectures, to make it possible to achieve an acceptable correlation between the measured vibration data and the actual amount of wear. The first model is based on the ResNet architecture where the input data are converted to spectrograms. The second model was based on a long short-term memory (LSTM) architecture. The proposed model was tested in terms of its accuracy in wear severity classification. The results show that this machine learning method accurately estimates the amount of wear in different locations in the S&C. Full article