With ongoing climate warming, runoff in alpine catchments of Qilian Mountains (QLM) has exhibited significant variability over recent decades. Therefore, understanding runoff variations and the underlying driving mechanisms in QLM is helpful for regional water resources planning and management. In this study, three innovative attribution methods incorporating glacier meltwater were employed to quantitatively analyze the impacts of climate change, glacier meltwater and land surface change on runoff variations in three representative basins of the QLM. To improve the reliability of runoff attribution analysis, this study first adopted a coupled VIC-OGGM modeling framework to simulate glacier meltwater and evapotranspiration, which are difficult to directly observe at the basin scale, thus avoiding uncertainties introduced by inconsistent input data sources. The coupled model exhibited high robustness and simulation accuracy in reproducing hydrological processes, with NSE > 0.80 for streamflow and NSE > 0.78 for evapotranspiration. The integrated results from three methods indicated that climate change is the dominant driver of runoff variations in the XYRB and HRB, accounting for 46.6-60.1% and 60.3-68.1% of the increased runoff, respectively. In contrast, glacier meltwater plays the dominant role in the SLRB, contributing 39.0-45.2% of the increased runoff. Among the three attribution methods, the hydrological model is the most robust and reliable choice for runoff attribution studies. The modified double mass curve method and Budyko-based methods can serve as efficient and practical alternatives when detailed hydrological modeling is not feasible. This study incorporated the influence of glacier meltwater into the attribution analysis of runoff variations, providing important and valuable insights for the ecologically sustainable development of water resources in Hexi Inland River Basin.

This paper studies the problem of dynamic group nearest neighbor group query (DGNNQ for short) over streaming data, an important problem in the domain of streaming data management. Let 𝒮 be the set of streaming data. A DGNNQ, denoted as q(n, s, k, 𝒬) , monitors objects within the window that contains q(n) objects. Whenever q(s) objects pass, q returns q(k) objects with the smallest distance sum to the query point set 𝒬 . Although some efforts can support this kind of queries, they incur highly running cost, especially when query points are allowed to be inserted into, or remove from 𝒬 . In this paper, we propose a novel framework named Q2WP (short for Query points and Window-based Partition) over streaming data. We group nearby query points into k subsets, and form a group of k virtual points based on these subsets. We then propose a novel index named PM-Tree to organize these virtual points, as well as evaluate which objects have chance to become query result objects based on PM-Tree. We further propose a group of novel algorithms to support incremental maintenance when 𝒬 updates. Extensive performance studies on large real-world and synthetic datasets demonstrate that the proposed framework can efficiently support DGNNQ over streaming data.

Efficiently managing and querying large bipartite graphs necessitates powerful indexing structures. A fundamental challenge lies in supporting (α,β)-component (bi-component) search, a task that encompasses retrieving all bi-components (bi-core search) or identifying the specific bi-component containing a query vertex (bi-community search). Existing index-based solutions face a time-space trade-off: single-dimensional approaches incur significant storage redundancy, while holistic approaches suffer from repeated vertex retrievals, hindering their scalability. To overcome these limitations, in this paper, we address this trade-off by introducing a novel, finer-grained cohesive unit, the (α,β,γ)- cluster ( bi-cluster ). This unit inherently captures the nested and overlapping relationships among bi-components, enabling vertex deduplication at the index level. Building on this, we propose the SGL ( Summary Graph + Location ) index, which achieves zero-redundancy in vertex retrieval during query processing. It organizes bi-components into a compact Summary Graph (SG) of interconnected SNodes. This structure, coupled with a precise Location mapping, allows our search algorithm to retrieve results without redundant vertex accesses, effectively decoupling query cost from the graph's global scale. Furthermore, we provide efficient construction and maintenance algorithms for dynamic bipartite graphs. Extensive experiments on ten real and synthetic large bipartite graphs demonstrate that our method outperforms the state-of-the-art by up to two orders of magnitude in query speed for bi-component search, while maintaining low storage, thus bridging the gap between fast but space-inefficient and compact but slower existing approaches.

This paper researches the problem of Continuous Top-k Skyline Pairs Query (TKSPQ for short) over data stream, an important problem in the field of streaming data management. Let 𝒮 be the set of streaming data. Each TKSPQ, denoted as q(n, s, k), monitors q(n) objects within the window. When q(s) objects update, q returns q(k) pairs with the highest pair score in skyline pair set. To the best of our knowledge, it is the first effort to support TKSPQ over data stream. In this paper, we propose a novel framework named PSPS (short for Partition-based Skyline Pair Search) over data stream. We partition objects within the window into a set of partitions, and then select a group of high-quality pairs within each partition as candidate pairs. Then, we introduce a novel index called PQ-Tree to organize objects, which helps us efficiently select high-quality pairs. Finally, we propose a set of novel algorithms for supporting incremental maintenance when handling newly arrived/expired objects. Extensive experiments on both real and synthetic datasets demonstrate that PSPS effectively supports TKSPQ over data stream.

Highlights What are the main findings? What are the implications of the main findings?Highlights What are the main findings? What are the implications of the main findings?Abstract Studying the impact of extreme climate events on vegetation dynamics is crucial for maintaining ecosystem stability. Based on ERA5-Land temperature and precipitation data, as well as MOD13C2 data, this study employs Pearson correlation analysis, wavelet analysis, and the XGBoost-SHAP model to analyze the response of the Normalized Difference Vegetation Index (NDVI) in China to extreme climate variations. The findings reveal the following. (1) NDVI increased steadily at a rate of 0.021/10 yr from 2001 to 2024. Extreme temperature and precipitation indices show an increasing trend in most regions. (2) NDVI was positively correlated with most extreme temperature and precipitation indices, and showed a significant negative correlation exclusively with Frost Days FDO (Frost Days) and CDD (Consecutive Dry Days). (3) R25 (Number of Heavy Precipitation Days) and SDII (Simple daily intensity index) are the primary drivers of nationwide NDVI changes, with contribution rates of 31.6% and 12.5%. Extreme climate indices can significantly affect vegetation growth when surpassing certain thresholds. For instance, the thresholds for R25 and SDII are 1.85 days and 4.35 mm/day. This study provides a scientific basis for understanding and managing vegetation responses to increasing climate extremes.

Community search in heterogeneous information networks (HINs) is a fundamental task in graph management and analysis. Existing relational community search models, which rely on simple connection requirements between vertices, often fall short of capturing users’ complex needs. To address this limitation, we introduce the concept of complex relational constraints based on meta-paths. These constraints enable users to specify refined connection requirements by limiting the number of meta-paths between vertex pairs, thereby capturing higher-order semantic relationships. Based on this, we formally define the complex relational community and formulate the problem of Minimum Complex Relational Community Search (MCRCS), which identifies the minimal such community containing a given query vertex, and prove its NP-hardness. To solve the MCRCS problem, we propose both exact and approximate algorithms. We first develop two exact algorithms: a global vertex-removal method (Global) and a local search method (Ls). Given the problem’s complexity, we then introduce two efficient approximation algorithms, Greedy and LsApp, which employ vertex prioritization strategies to find high-quality communities with significantly reduced computational cost. Comprehensive experiments on four real-world HIN datasets demonstrate the effectiveness and efficiency of our proposed algorithms.

Study region: Northwest China Study focus: Understanding the propagation mechanism from meteorological to groundwater drought is crucial for groundwater management and drought early warning. However, scant research exists for mechanism of the unseen groundwater drought propagation. This study applied drought indices including the Standardized Precipitation Index (SPI) and Groundwater Drought Index (GDI), and utilized methods such as run theory, convergent cross mapping (CCM), Copula function, and Bayesian network, as well as several open-source data sources to analyze the drought characteristics, propagation rule, threshold, and recovery time of meteorologicalgroundwater drought in Northwest China (NWC) from 1960 to 2024. Specifically, the 'compound meteorological-groundwater drought event' is defined as the period from the onset of groundwater drought to the end of meteorological drought, aiming to highlight the full system response time from the initiation of deep water deficit to shallow water recovery. The contributions of driving factors were further quantified using XGBoost-SHAP, a game theory-based approach for interpreting model outputs and quantifying feature importance. New hydrological insights for the region: The number of meteorological-groundwater drought events is lower than that of meteorological droughts but higher than that of groundwater droughts, with the shortest average duration (2.29 months) and the lowest severity (3.94). The propagation time (PT) of meteorological-groundwater drought is 4.69 months. The average probabilities of the meteorological drought triggering mild, moderate, severe, and extreme groundwater droughts are 30.27 %, 20.60 %, 9.63 %, and 5.50 %. The propagation threshold is dominated by extreme meteorological drought, accounting for 55.69 %. The recovery time for compound meteorological-groundwater droughts reached up to 3.05 months, exceeding that of individual meteorological or groundwater drought events. ENSO has the strongest influence on the

Trajectory-User Linking (TUL) task aims to accurately match anonymous trajectories to their corresponding users. As a critical task in mobility data mining, its resolution is crucial for a wide range of downstream applications, including personalized recommendations, urban planning, and public safety. However, existing methods primarily focus on point-level data while neglecting the holistic travel semantics embedded within urban road network. Moreover, they overlook the higher-order relationships among trajectories of different users. Consequently, we propose a Multiscale Semantics-Relationships Fusion Representation Model for TUL, namely MSRTUL. Specifically, we first propose a multiscale destination-oriented trajectory semantics encoder, which captures destination-oriented spatial-temporal semantics from both road-level and zone-level trajectories to encode the holistic travel semantics. Subsequently, we design a multiscale trajectory high-order relationships encoder, which jointly models higher-order relationships among trajectories, points, and categories through the hypergraph. To effectively combine the travel semantics and higher-order relationships, we design a dual-view fusion layer. We conduct extensive experiments on three real-world datasets, demonstrating that MSRTUL achieves significant improvements over multiple baselines in the TUL.

This paper studies the problem of DMCTQ (Dynamic Multiple Continuous Top-k Queries) over streaming data, a fundamental problem in the domain of streaming data management. Let S be the set of streaming data, and Q be the query workload. It contains a set of queries with different query parameters. Each query qi(n, s, k, F) in Q monitors objects generated in the last qi(n) time units, and returns qi(k) objects with the highest scores to the system whenever qi(s) time units pass. Some existing methods support DMCTQ but they incur high computational costs, especially when queries within Q are allowed to be changed. In this paper, we propose a novel framework named Skyline-based Top-k Query Framework (STKQF for short) over streaming data. It is based on the following observation. For two queries q1 and q2 in the query workload Q, if q1(n) > q2(n) and q1(k) > q2(k), the meaningful objects of q2 must be meaningful objects of q1. Therefore, we can support DMCTQ based on partial queries. Based on this, we propose a grid based index to maintain these queries, form a small number of virtual queries, support DMCTQ via monitoring meaningful objects under these virtual queries. Finally, we conduct extensive performance studies on large real and synthetic datasets, which demonstrate that our new framework could efficiently support DMCTQ over streaming data.

Vegetation drought is a critical manifestation of ecosystem vulnerability in high-altitude, water-limited regions under climate change. The Yellow River Water Conservation Area (YRWC), as the core water source of the Yellow River Basin, is highly sensitive to variations in hydrothermal conditions. In this study, a Temperature–Vegetation–Precipitation Drought Index (TVPDI) was constructed to characterize the spatio-temporal dynamics of vegetation drought in the YRWC for 2003, 2012, and 2019. The XGBoost–SHAP framework was further employed to quantitatively analyze the nonlinear response characteristics and relative contributions of key factors within the TVPDI framework. Scenario-based spatial simulations of vegetation drought for 2035 are then conducted based on the GeoSOS-FLUS model. The results indicate that vegetation drought in the YRWC exhibits a relatively stable spatial pattern, with drought severity gradually intensifying from southeast to northwest and moderate drought as the dominant type. Precipitation is the key variable of TVPDI, followed by land surface temperature, while NDVI mainly plays a nonlinear regulatory role. Among external factors, atmospheric moisture conditions show relatively higher explanatory relevance, whereas topographic and human activity factors exert comparatively weaker influences. Scenario-based simulation results suggest that vegetation drought may be alleviated under low-emission pathways, whereas high-emission scenarios substantially exacerbate drought severity and associated risks. This study presents an interpretable, index-based analytical framework combined with scenario-based spatial simulation for characterizing vegetation drought in the YRWC, thereby providing scientific support for ecological management and climate adaptation strategies in the Yellow River Basin.

The continuous identification of top-k maximal sum intervals using a sliding window over a data stream is a critical operation for applications in IoT and beyond. A maximal sum interval is a non-overlapping, contiguous subsequence with the maximal sum in a sequence of signed values. Existing algorithms are ill-suited for streaming contexts: they either exhaustively enumerate all intervals even for small k values, or depend on indexes that require frequent and costly restructuring. We propose a novel partition-based strategy. Our core insight is a partitioning scheme that guarantees that any maximal sum interval is fully contained within a single partition, enabling independent and parallel processing. This design provides two key advantages: it enables safe pruning of partitions that cannot contribute to top-k results, drastically narrowing the search space, and it enables efficient, incremental maintenance of the maximal sum intervals in each partition. We develop algorithms for partition construction, incremental partition updates, and partition-based top-k maximal sum interval search. Extensive experiments on real and synthetic datasets demonstrate that our approach significantly improves efficiency.

Continuous distance-based outlier detection in streaming data poses significant challenges and has a wide range of practical applications. Traditional threshold-based methods perform well under stable streaming conditions, where fixed parameters remain effective. However, they often struggle with dynamic data distributions and high stream speeds, leading to suboptimal performance, limited control over the number of returned outliers, and failure to meet real-time detection requirements. To address these issues, this paper introduces a novel R ecall and P roportion- A ware O utlier D etection (RPA-OD) query. In RPA-OD, ρ defines a distance relaxation that enables real-time outlier detection. Specifically, objects with fewer than k neighbors within the distance threshold r' ∈ ( r , (1+ρ) ⋅ r ] might be classified as ρ-inliers. This relaxation introduces a trade-off between recall and efficiency, allowing the system to adapt under varying streaming conditions. We propose efficient algorithms to support RPA-OD in data streams, leveraging several novel data structures developed as part of this study. Extensive experiments on five real-world datasets show that RPA-OD significantly improves data throughput, provides precise control over the number of outliers detected, and consistently ensures real-time processing performance.

Complex Event Processing (CEP) is a powerful technique for detecting event patterns within event streams. Traditional CEP matching methods rely on rigid selection policies that focus on the continuity of events in queries. However, real-world applications often generate event streams with varying time granularities. In such cases, rigid selection policies tend to produce either too many redundant matches or too few meaningful matches, lacking the flexibility required for fine-tuning according to specific application needs. To address this challenge, this paper introduces a novel complex event query that integrates a parameter-driven selection policy. This policy allows users to specify a parameter for an event instance e, which constrains the maximum number of matches that e can generate. Additionally, we propose a matching algorithm that supports this parameter-driven selection policy by leveraging potential domination relationships among complex event matches. Finally, we conduct experiments on both synthetic and real-world datasets to demonstrate the effectiveness and efficiency of the proposed methods.

R-tree is a multi-dimensional spatial index that organizes objects using Minimum Bounding Rectangles (MBRs) to support similarity search. Existing studies on R-Tree optimization can be classified into two categories: heuristic-based approaches that refine node insertion and splitting strategies, and learning-based approaches that use machine learning models to discover adaptive construction strategies. However, these methods ignore the impact of query workload, resulting in redundant multibranch traversals during query and degraded query efficiency. In this paper, we propose a novel query-aware learned R-tree for multi-dimensional object management and similarity search. Specifically, we first design a greedy-based index construction algorithm to group objects or nodes with high query relevance based on a query correlation graph from historical queries. Then, we propose a learning-based algorithm to determine query hotspot regions. Finally, extensive experiments on both real and synthetic datasets demonstrate that our method significantly reduces the query costs and exhibits strong query adaptability.

Similarity search in metric spaces is a fundamental problem in data management with many applications. While numerous indices have been proposed to support similarity search, most existing approaches rely on pivot-based strategies that suffer from critical limitations. Traditional single-pivot methods offer limited pruning power, while multi-pivot techniques often become inefficient as data evolves, since pivot updates incur substantial computational overhead. In this paper, we propose LM-Tree (short for Learned M-Tree), a hybrid learned index that combines pivots and learning models with M-Tree to address the above problems. LM-Tree's key innovation lies in its self-adaptive node architecture, where each node dynamically selects an appropriate number of pivots and incorporates lightweight learning models to enhance pruning efficiency. This design enables each node to maintain a relatively large number of child nodes (or objects for leaf nodes) while consistently delivering strong pruning performance, thereby enabling LM-Tree to use a small number of nodes to index objects in metric spaces. Furthermore, we develop an efficient maintenance algorithm that handles dynamic updates, including pivot adjustments, model reforms, node splits, and merges with low overhead. Extensive experiments on both real and synthetic datasets demonstrate that LM-Tree significantly outperforms state-of-the-art methods.

Complex event processing faces significant challenges in real-world scenarios where noisy event streams containing missing events, misordered sequences, and erroneous attribute values are prevalent. Existing approaches address only specific aspects of this problem, with approximate matching techniques handling missing events and out-of-order processing methods addressing sequence misalignment, but none provide a comprehensive solution. This paper introduces K-NFA, a novel fault-tolerant automaton that systematically handles all major error types in complex event matching. Our approach features a compact automaton structure with specialized transitions for event deletion and swap operations, coupled with a dynamic matching algorithm that efficiently processes noisy streams while respecting temporal and attribute constraints. Experimental results show our method improves fault-tolerant matching capacity by 32.91

Recently, community search in directed graphs has garnered significant attention, particularly with the introduction of the D-truss, also known as the (k_c, k_f) -truss, which serves as a robust subgraph structure in directed graphs. In this structure, each edge must form cycle (flow) triangles with at least k_c ( k_f ) vertices. The goal of maximal D-truss community search is to identify the largest D-truss for a given query vertex in a directed graph. However, existing methods often lack both efficiency and effectiveness when applied to large-scale and dynamic directed graphs. To overcome these limitations, we investigate the problem of maximal triangle-connected D-truss community search (MDTCS) in dynamic directed graphs in this paper. First, we introduce the concept of triangle connectivity. Then, we explore an efficient index, named PartialTruss, which effectively captures the partial correlation of edges within a D-truss community based on triangle connectivity. Next, an effective search algorithm that utilizes the PartialTruss index is proposed. Furthermore, we propose a novel method for efficiently maintaining the index in response to dynamic updates in directed graphs. Finally, we conduct extensive experiments on real-world networks, which demonstrate that our community search method, leveraging the PartialTruss index, achieves an improvement in efficiency by 1–2 orders of magnitude compared to state-of-the-art methods.

In the domain of route planning, the critical concern for travel efficiency has shifted towards optimizing travel time over distance due to the rise in congestion and the dynamic nature of modern road networks. Addressing this shift, we introduce Dynamic Route Optimization with Multi-Category Constraints (DROMC) for Point-of-Interest (POI) visits, which seeks to find the most time-efficient path that satisfies a user-defined list of requirements, considering both the spatial and temporal dimensions. This paper proposes a novel approach that leverages a path enumeration algorithm, which iteratively validates the k-fastest paths until all user queries are satisfied, ensuring adherence to time constraints and POI availability. To enhance the algorithm's efficiency, we employ several key methodologies. First, we adapt the kSP algorithm to account for POI-focused path enumeration. We also introduce a Shared Prefix Tree (SPFT) with binary encoding, which significantly improves the storage and retrieval of path information. Moreover, we integrate a grid-based heuristic for quicker computation and implement strategic pruning methods to circumvent redundant calculations and manage POI business hours effectively. Our extensive experiments on real-world networks demonstrate the algorithm's superiority in finding more efficient paths in shorter time frames compared to existing methods.

In the context of global warming,it is of great significance to clarify the change characteristics and trends of hydrological elements in the mountainous areas of inland river basins to ensure water resource security.In this study,the spatiotemporal variation of hydrological elements under four scenarios,SSP1-2.6,SSP2-4.5,SSP3-7.0 and SSP5-8.5,was predicted with the SWAT model and the collective average data of five global cli-mate models in CMIP6.The results showed that:(1)The evaluation coefficient of the SWAT model was higher in the calibration period(NSE=0.92,R2=0.93,PBIAS=-7.09％)and validation period(NSE=0.89,R2=0.91,PBIAS=4.74％),indicating that SWAT had good applicability in the simulation of runoff in the upper reaches of the Heihe River Basin.(2)Under the four scenarios,the runoff from the mountains in the future will increase by 12.2％,8.1％,10.4％and 19.2％respectively compared with the base period,and the runoff will increase sig-nificantly in autumn and winter.In the near and far future,the increase in the average total water yield in the basin is between 6.2～25.4 mm(22.2～35.7 mm),and the increase in the average underground flow is 1.6～7.4 mm(7.4～12.1 mm),and the increase of each hydrological element is greater in the far future.(3)In terms of the spatial distribution of hydrological elements,the spatial distribution of precipitation,evapotranspiration,surface production and underground runoff increased from northwest to southeast,while the total water yield and lateral flow were high in the middle and low in the north.(4)The spatial distribution pattern of the changes of each hy-drological element is quite different,and the spatial distribution difference of the change of hydrological ele-ments between different scenarios in the near future(2021-2060)period is relatively small,and the difference in the temporal and spatial distribution of the change in the far future(2061-2100)is even greater.In conclu-sion,the SWAT model can better describe the temporal and spatial changes of hydrological processes and hydro-logical elements in the mountainous areas of inland river basins.