Xuemei Liu

Runoff prediction in ungauged basins remains a fundamental challenge due to the absence of discharge observations and the limited transferability of data-driven models across heterogeneous hydrological conditions. This study proposes a regime-aware framework that models hydrological processes from the perspective of shared driver-induced regimes. Instead of relying on runoff observations, a view-based contrastive self-supervised learning strategy is employed to extract transferable features from hydrometeorological drivers. Specifically, augmented temporal-window views generated through temporal masking and feature perturbation are contrasted to learn runoff-independent representations, enabling the identification of cross-basin hydrological regimes via unsupervised clustering. Regime-specific prediction models are then constructed and adaptively combined for target basins based on regime similarity. Experimental results show that the proposed method achieves a median Nash–Sutcliffe efficiency of 0.5 across ungauged test basins, demonstrating improved prediction performance under ungauged conditions. These findings highlight the effectiveness of learning shared hydrological regimes from driver space and provide a new perspective for robust runoff prediction in ungauged basins.

The regular detection of asphalt pavement damage is crucial for the safe operation of highway. However, the current detection method has the problem of low efficiency and accuracy. In addition, the image tangential distortion will be caused by the oblique photography of the acquisition equipment. Thus, the accuracy of extracted the pothole area is affected. In order to solve the above problems, a segmentation method of asphalt pavement pothole damage based on image correction and deep learning is proposed. Firstly, an image correction method called perspective change is proposed to solve the tangential distortion problem of asphalt pavement image. Secondly, an innovative detection and segmentation model called CMSegNet is developed and used for the first time for accurate segmentation of asphalt pavement pothole damage. In this model, the global feature extraction capability of the pothole target is improved by optimizing the Mamba structure. In addition, a Multi-Scale Attention Aggregation (MSAA) module is proposed to fuse multi-scale local and global features. The segmentation accuracy, IoU, F1-Score and segmentation efficiency of this model can reach 91.28%, 89.96%, 91.74% and 24.52FPS, respectively. Finally, on-site experiments of asphalt pavement pothole detection are carried out to verify the generality, power and scalability of the proposed correction method and segmentation model. The results show that the proposed image correction method based on perspective transformation can effectively reduce the error rate of quantization of pothole area. The error rate is only 6.84%, an improvement of 4.09%. In addition, the superiority of the proposed CMSegNet segmentation model is further proved.

The operational environment of water conservancy projects is complex, with numerous risk factors affecting engineering safety and exhibiting a chain-like propagation pattern, which poses significant challenges for risk prevention and control. This study proposes a knowledge-enhanced large language model to uncover the propagation patterns of risk chains and identify key risks impacting engineering safety. Integrating the large language model ChatGLM4 with the chain-of-thought strategy to identify risk entities and their relationships in inspection texts, thereby constructing a risk knowledge graph for engineering safety. Analyze the chain-like propagation patterns of risks in the knowledge graph and integrate them as structured knowledge into ChatGLM4, identifying triggering points, convergence points, and diffusion points in the risk chain to clarify key risks impacting engineering safety. The results show that the method proposed in this study achieves high accuracy in risk identification (with an average F1-score of 81.43%). Risks such as weed growth in the expansion joints of the lining panels and seepage in the slope or toe of the channel embankment exhibit strong transmissibility and a high degree of uncertainty. These studies can effectively reduce the impact of potential risks on engineering safety, enhancing the operational safety of projects.

This study investigates the role of Big Data Analytics Capabilities (BDACs) in ambidexterity explorative innovation (EXPLRI) and exploitative (EXPLOI) innovation for achieving a sustainable performance (SP) in the manufacturing sector of a resource-constrained developing economy. While a BDAC has been widely linked to innovation in developed economies, its effectiveness in developing contexts shaped by indigenous innovation practices like Jugaad remains underexplored. Anchored in the Resource-Based View (RBV) and Dynamic Capabilities (DC) theory, we propose a model where the BDAC enhances both EXPLRI and EXPLOI, which subsequently leads to an improved sustainable performance. We further examine the Jugaad capability as a cultural moderator. Using survey data from 418 manufacturing firms and analyzed via Partial Least Squares Structural Equation Modeling (PLS-SEM), results confirm that BDA capabilities significantly boost both types of innovations, which positively impact sustainable performance dimensions. Notably, Jugaad positively moderates the relationship between EXPLOI and financial, innovation, and operational performance but negatively moderates the link between EXPLRI and innovation performance. These findings highlight the nuanced influence of culturally embedded innovation practices in BDAC-driven ecosystems. This study contributes by extending the RBV–DC framework to include cultural innovation capabilities and empirically validating the contingent role of Jugaad in enhancing or constraining innovation outcomes. This study also validated the Jugaad capability measurement instrument for the first time in the context of Pakistan. For practitioners, aligning data analytics strategies with local innovative cultures is vital for sustainable growth in emerging markets.

Multibaseline (MB) phase unwrapping (PU), as the core step in MB InSAR, breaks the limitation of phase continuity assumption. However, it still suffers from insufficient noise robustness and low unwrapping efficiency. This paper transforms the MB PU problem into pure integer programming (PIP) problem and innovates a deep convolutional neural network (DCNN) named PIPNet to solve PU problem. This PIPNet is built on the U-Net framework, incorporating Transformer modules, upsampling, and dense connections mechanisms to achieve powerful feature extraction capabilities. Interferometric phase, fringe frequency, and intercept data are input into multi-ports of PIPNet, with multi-source features fused in the encoding-decoding stages to collectively act on PU. Then, we innovatively designed a new joint loss function and PIP constraints to guide the iterative optimization of model parameters. The attention mechanism enhances the model ability to disentangle interferometric phase details. Finally, the trained model can achieve real-time and accurate calculation of phase ambiguity number. Theoretical analysis and two experimental results indicate that PIPNet-PU has much lower execution time than MCF, CA, and TSPA algorithms, and the second-best unwrapping accuracy for short- baseline interferogram and the best unwrapping ability for long- baseline interferogram. PIPNet-PU can achieve good unwrapping effects even in areas with abrupt topographic change and dense interferometric fringes.

Traditional hydraulic structures rely on manual visual inspection for apparent integrity, which is not only time-consuming and labour-intensive but also inefficient. The efficacy of deep learning models is frequently constrained by the size of available data, resulting in limited scalability and flexibility. Furthermore, the paucity of data diversity leads to a singular function of the model that cannot provide comprehensive decision support for improving maintenance measures. This paper proposes an efficacious methodology for identifying diverse apparent damages in hydraulic structures to address the limitations of existing technologies. The advanced features of apparent damage in hydraulic structures were elucidated by fine-tuning the top-level parameters of the lightweight pre-trained model, thereby mitigating the data dependency issue inherent in the model. Ensemble learning algorithms are employed to classify high-dimensional samples to enhance the accuracy and stability of the classification. However, ensemble learning algorithms are subject to time consuming issues when applied to high-dimensional datasets. To this end, we propose a robust discriminative feature selection model to identify the most salient features, thereby enhancing the performance of apparent damage recognition in hydraulic structures while concurrently reducing the inference time. The results demonstrated that the accuracies of this method in identifying crack, fracture, hole and normal structures were 87.65%, 87.82%, 96.99%, and 95.25%, respectively. This methodology exhibits significant applicability and practical value for the intelligent inspection of hydraulic structures.

As the core step of multibaseline synthetic aperture radar interferometry (InSAR), multibaseline phase unwrapping (MBPU) always faces the noise challenges. Therefore, a new robust InSAR U-shaped MBPU algorithm with extended Margolus automata (U-EMA) is proposed. This algorithm is divided into two stages: U-shaped model solving and EMA noise robustness processing. Firstly, an N-dimensional standard discrete optimization model has been established for ambiguity numbers, and the N-dimensional discrete model, which has been maximally reduced in dimensionality, is solved in U-shaped manner. Then, EMA was introduced in phase unwrapping (PU) for the first time to process discrete solutions, thereby improving the noise robustness to spatially varying noise, and finally PU was achieved. Two set of PU experiments were organized under the conditions of interferogram signal-to-noise ratio of 3dB and 4dB, respectively, compared with minimum cost flow (MCF), clustering analysis, and two-stage programming approach (TSPA). The results show the effectiveness and noise robustness of the U-EMA method, the root mean square error of this algorithm is reduced by 63.7% and 14.6%, 63.2% and 31.3%, respectively, compared with the mainstream algorithms MCF for single-baseline PU and TSPA for MBPU.

During the joint extraction of entity and relationship from the operational management data of hydraulic engineering, complex sentences containing multiple triplets and overlapping entity relations often arise. However, traditional joint extraction models suffer from a single-feature representation approach, which hampers the effectiveness of entity relation extraction in complex sentences within hydraulic engineering datasets. To address this issue, this study proposes a multi-feature joint entity relation extraction method based on global context mechanism and graph convolutional neural networks. This method builds upon the Bidirectional Encoder Representations from Transformers (BERT) pre-trained model and utilizes a bidirectional gated recurrent unit (BiGRU) and global context mechanism (GCM) to supplement the contextual and global features of sentences. Subsequently, a graph convolutional network (GCN) based on syntactic dependencies is employed to learn inter-word dependency features, enhancing the model’s knowledge representation capabilities for complex sentences. Experimental results demonstrate the effectiveness of the proposed model in the joint extraction task on hydraulic engineering datasets. The precision, recall, and F1-score are 86.5%, 84.1%, and 85.3%, respectively, all outperforming the baseline model.

Agrometeorological hazards significantly impact agricultural production and rural economic development. The interdisciplinary nature of studying these hazards poses challenges such as poor data interoperability in research. This paper proposes a method for analyzing agrometeorological hazards using knowledge graphs to understand occurrence patterns and devise response strategies. The study involves classifying agricultural and meteorological knowledge and designing a hazard entity model based on the characteristics and influencing factors of agrometeorological hazards. Data mining and extraction techniques are used to extract relevant information from multiple sources, and a knowledge graph for knowledge fusion and storage is built. The retrieval and inference capabilities of the knowledge graphs are used to intelligently analyze agrometeorological hazards. Results indicate that analyzing agrometeorological hazards using knowledge graphs is an innovative method that offers new perspectives and ideas for agricultural meteorological hazard research, thereby promoting the sustainable development of agricultural production and the stable growth of the rural economy.

Due to the frequent occurrence of extreme weather in recent years, accurate runoff prediction is crucial for the rational planning and management of water resources. Addressing the high uncertainty and multiple influencing factors in runoff prediction, this paper proposes a runoff prediction method driven by multi-source data. Based on multivariate observed data of runoff, water level, temperature, and precipitation, a Time2Vec-TCN-Transformer model is proposed for runoff prediction research and compared with LSTM, TCN, and TCN-Transformer models. The results show that the Time2Vec-TCN-Transformer model outperforms other models in metrics including MAE, RRMSE, MAPE, and NSE, demonstrating higher prediction accuracy and reliability. By effectively combining Time2Vec, TCN, and Transformer, the proposed model improves the MAPE for forecasting 1–4 days in the future by approximately 7% compared to the traditional LSTM model and 4% compared to the standalone TCN model, while maintaining NSE consistently between 0.9 and 1. This model can better capture the periodicity, long-term scale information, and relationships among multiple variables of runoff data, providing reliable predictive support for flood forecasting and water resources management.