Self-organizing Map Research Articles

Mn-Fe bimetallic oxide catalytic oxidation combined with ultrafiltration for treating secondary effluent: EfOM molecular transformation and membrane fouling control mechanism

Despite extensive application and research of membrane technology in secondary effluent treatment, fouling induced by effluent organic matter (EfOM) remains a challenge. Herein, a pretreatment strategy involving Mn-Fe bimetallic oxide-activated peroxymonosulfate (MnFeO/PMS) was used before ultrafiltration to decompose EfOM and tackle membrane fouling. Results showed that membrane fouling was effectively alleviated, with reductions in reversible and irreversible fouling resistances of 88.3 % and 58.9 %, respectively. Theoretical analyses illustrated a more pronounced adsorption energy and decomposition tendency of PMS on the MnFeO surface. The generated OH and SO4− played a key role in the degradation of EfOM. The parallel factor analysis along with self-organizing maps revealed the preferential oxidative degradation of humic acids and fulvic acids over tryptophan-like substances, which contributed to the reversible fouling mitigation. In addition, MnFeO/PMS transformed high and middle molecular weight organics into low molecular weight compounds. The primary molecular reactions involved in the degradation of EfOM included oxygen addition, dealkylation and deamination, which changed the properties of EfOM and contributed to the mitigation of membrane fouling. Morphology analysis further indicated that the fouling layer on the membrane surface was inhibited when filtering the MnFeO/PMS-pretreated sample. This study offers revelations for integrating oxidation with ultrafiltration in secondary effluent treatment to alleviate EfOM-induced membrane fouling.

Uncertainty assessment in unsupervised machine-learning methods for deepwater channel seismic facies using outcrop-derived 3D models and synthetic seismic data

Unsupervised machine-learning (ML) techniques have been widely applied to analyze seismic reflection data, including the identification of seismic facies and structural features. However, interpreting the resulting clusters often relies on geoscientists’ expertise, necessitating a robustness assessment of these methods. To evaluate their reliability, synthetic data generated from an actual outcrop model are used to demonstrate how two unsupervised methods, self-organizing maps (SOMs) and generative topographic maps (GTMs), cluster deepwater channel-related seismic facies and then measure the associated error. Six seismic attributes, comprising root-mean-square amplitude, instantaneous envelope, peak magnitude, and spectral decomposition frequencies at 20, 40, and 55 Hz, served as input variables. Geobodies are assigned to each cluster formed, and the error in facies clustering is quantified by comparing the actual 3D model with the facies grouped by ML methods on a voxel-by-voxel basis. This allowed for error quantification and the computation of metrics such as F1 score and accuracy through correlation matrices. Key findings reveal that (1) GTM and SOM exhibit similar performance, with a clustering configuration of 81 for GTM slightly outperforming others. (2) Error rates are approximately 2% for the predominant facies (background shale) but significantly higher for individual channel-related facies, suggesting that channel clusters might represent multiple facies. (3) Resolution and imbalanced data distribution impact seismic facies predictability, resulting in nonuniqueness in cluster generation. (4) Using synthetic seismic data proved valuable for experimenting with different unsupervised MLs, highlighting the need for assessing uncertainty in these methods, given their implications for crucial economic decisions reliant on reservoir interpretation, modeling, and volumetric estimations.

Integrating self-organizing feature map with graph convolutional network for enhanced superpixel segmentation and feature extraction in non-Euclidean data structure

Integrating self-organizing feature map with graph convolutional network for enhanced superpixel segmentation and feature extraction in non-Euclidean data structure

Sieci neuronowe do mapowania archeologicznych artefaktów litowych

An analytical non-destructive strategy to chemically characterize lithic artefacts has been developed. Around 100 archaeological lithic materials found in Neolithic-Chalcolithic sites in the Mediterranean region of the Iberian Peninsula and nowadays stored in different museums of the Valencian Community (Spain), were studied. The materials belong to different typologies of rock (diabase, sillimanite, ophite and amphibolite) and were analysed employing portable energy dispersive X-ray fluorescence spectroscopy (pXRF) directly in the rock surface. The obtained data were processed through neural networks protocol, specifically the so-called Kohonen networks or Self Organised Maps (SOM), to map the geologic samples. This self-organized topological feature maps are suitable to deal with multidimensional representations and map them in a two-dimensional space of neurons, following an unsupervised learning protocol. SOM is used to reduce multidimensional data onto lower-dimensional spaces and clustering procedures. As a result, SOM create spatially organized representations, which enhance the discovery of correlations between data. In this case the method has enabled the evaluation of elemental markers related to each rock type behaving as a fine hidden pattern detector and so understand the possible advantages and disadvantages of the analytical method employed to define provenance issues. The attribution suggested by statistics is mainly driven by the composition of rocks essential minerals which are linked to the different petrogenetic conditions. The results showed that in most of the cases the distribution and dispersion of the chemical profile depend of the kind of rock, and clearly suggest that a good way to identify stone tools raw material procurement is to look for elemental markers, being the prior step to create an approximation to ancient exchange networks and their evolution in a diachronic axis.

Multi-isotopes (δD, δ18Owater, 87Sr/86Sr, δ34S and δ18Osulfate) as indicators for groundwater salinization genesis and evolution of a large agricultural drainage lake basin in Inner Mongolia, Northwest China

Groundwater salinization, a major eco-environmental problem in arid and semi-arid areas, can accelerate soil salinization, reducing crop productivity and imbalances in ecosystem diversity. This study classified water samples collected from the Ulansuhai Lake basin into five clusters using self-organizing maps (SOM). On this basis, multiple isotopes (δ18Owater, δD, 87Sr/86Sr, δ18Osulfate and δ34S) and isotopic models (Rayleigh fractionation and Bayesian isotope mixing models) were used to identify and quantify the genesis and evolution of groundwater salinization. The results showed that the samples were brackish or saline water, and the hydrochemical types were dominated by Na + K-Cl (SO4). It has been proved that the processes associated with groundwater salinization in the Ulansuhai Lake basin were dominated by water-rock interaction and human inputs. Among them, evaporite dissolution contributed substantially to groundwater salinity. Furthermore, salt inputs from human activities cannot be negligible. Based on the model calculations, evaporite dissolution accounted for the most significant proportion of all sources, with a mean value of 53 %. In addition, human inputs from regular agricultural activities (28 % from sewage and manure and 8 % from fertilizers) constituted another vital source of groundwater salinization associated with extensive agricultural activities in the study area. This study's results can deepen our understanding of the genesis of groundwater salinization and the evolution of the agricultural drainage lake basin. This knowledge will assist the Environmental Protection Department in developing effective policies for groundwater management in the Yellow River Basin.

A Soft-Tissue Driven Bone Remodeling Algorithm for Mandibular Residual Ridge Resorption Based on Patient CT Image Data.

The role of the biomechanical stimulation generated from soft tissue has not been well quantified or separated from the self-regulated hard tissue remodeling governed by Wolff's Law. Prosthodontic overdentures, commonly used to restore masticatory functions, can cause localized ischemia and inflammation as they often compress patients' oral mucosa and impede local circulation. This biomechanical stimulus in mucosa is found to accelerate the self-regulated residual ridge resorption (RRR), posing ongoing clinical challenges. Based on the dedicated long-term clinical datasets, this work develops an in-silico framework with a combination of techniques, including advanced image post-processing, patient-specific finite element models and unsupervised machine learning Self-Organizing map algorithm, to identify the soft tissue induced RRR and quantitatively elucidate the governing relationship between the RRR and hydrostatic pressure in mucosa. The proposed governing equation has not only enabled a predictive simulation for RRR as showcased in this study, providing a biomechanical basis for optimizing prosthodontic treatments, but also extended the understanding of the mechanobiological responses in the soft-hard tissue interfaces and the role in bone remodeling.

A regional examination of the footprint of agriculture and urban cover on stream water quality

Freshwater systems in cold regions, including the Laurentian Great Lakes, are threatened by both eutrophication and salinization, due to excess nitrogen (N), phosphorus (P) and chloride (Cl−) delivered in agricultural and urban runoff. However, identifying the relative contribution of urban vs. agricultural development to water quality impairment is challenging in watersheds with mixed land cover, which typify most developed regions. In this study, a self-organizing map (SOM) analysis was used to evaluate the contributions of various forms of land cover to water quality impairment in southern Ontario, a population-dense, yet highly agricultural region in the Laurentian Great Lakes basin where urban expansion and agricultural intensification have been associated with continued water quality impairment. Watersheds were classified into eight spatial clusters, representing four categories of agriculture, one urban, one natural, and two mixed land use clusters. All four agricultural clusters had high nitrate-N concentrations, but levels were especially high in watersheds with extensive corn and soybean cultivation, where exceedances of the 3 mg L−1 water quality objective dramatically increased above a threshold of ‍∼30 % watershed row crop cover. Maximum P concentrations also occurred in the most heavily tile-drained cash crop watersheds, but associations between P and land use were not as clear as for N. The most urbanized watersheds had the highest Cl− concentrations and expansions in urban area were mostly at the expense of surrounding agricultural land cover, which may drive intensification of remaining agricultural lands. Expansions in tile-drained corn and soybean area, often at the expense of mixed, lower intensity agriculture are not unique to this area and suggest that river nitrate-N levels will continue to increase in the future. The SOM approach provides a powerful means of simplifying heterogeneous land cover characteristics that can be associated with water quality patterns and identify problem areas to target management.

Data oversampling and feature selection for class imbalanced datasets

Introduction: Significant advancements and modifications have been implemented in data classification (DC) in the past few decades. Due to their infinite quantity and imbalance, data becomes challenging for classification. The biggest concern in DM (Data Mining) is Class Imbalance (CI). To avoid these issues in recent work proposed map reduce based data parallelization of class imbalanced datasets. Methods: A novel Over Sampling (OS) technique called Minority Oversampling in Kernel Canonical Correlation Adaptive Subspaces (MOKCCAS) has been suggested with the objective to minimize data loss throughout (FSP) Feature Space Projections. This technique takes advantage of the constant Feature Extraction (FE) capability of a version of the ASSOM (Adaptive Subspace Self-Organizing Maps) that is derived from Kernel Canonical Correlation Analysis (KCCA). And in classification, Feature Selection (FS) plays an important role because the acquired dataset might contain large volume of samples, utilizing all features of samples from the dataset for classification will decrease the classifier performance. And then data parallelization will be done by using map reduce framework to solve this computation requirement problem.Result: Then proposes a feature selection model using Mutated whale optimization (MWO) methods and produces features and reduces the time consumption. Finally proposed class balancing model will be tested using uniform distribution based enhanced adaptive neuro fuzzy inference system (UDANFIS). Test outcomes validate the efficiency of the suggested technique by precision, recall, accuracy and Error Rate (ER).Conclusion: The study subsequently suggests a novel OS approach called MOKCCAS to lessen the loss of data throughout feature space projection.

Construction of regional multi-scenario spatially balanced ecological security pattern based on self-organizing feature map: A case study of Guangzhou, China

Constructing ecological security patterns (ESPs) is the ecological basis for achieving sustainable regional development. This study explored a framework of spatially balanced ESPs suitable for rapidly urbanizing areas to address the issues of overreliance on ecological land and subjective parameter settings in traditional ESP construction methods. Three ecosystem service bundles (product supply, water conservation, and habitat maintenance) were identified in Guangzhou using the self-organizing feature map method. These bundles were then used to identify multifunctional sources. The minimum size of the ecological source was determined to be 10 km2 through the “structure-pattern-function” framework, and a total of 19 ecological sources from three ecological functions were extracted. In these three scenarios, the 102 ecological corridors were classified into five categories based on their connected sources. Ecological node parameters were set based on local planning and dominant species characteristics, enhancing the explanatory power of the circuit theory. The multi-scenario ESP framework proposed in this study achieved a spatially balanced distribution of the entire region at the point, line, and surface levels, as confirmed by the ecological network assessment results. Future planning in Guangzhou should value the ecological functions of non-conventional ecological land, such as cropland and urban green spaces, while considering the impact of human activities on the environment and promptly modifying crucial node areas.

Self Organizing Maps (SOM) and statistical methods for describing the physicological profile of undergraduates students of engineering

To meet a Brazilian guideline for engineering courses, which advocates the use of active learning in applied disciplines, this research studies new students entering in the Metallurgical Engineering Course in Escola Politécnica of Universidade de São Paulo (PMT/USP). It was applied to them a traditional MTBI (Myers–Briggs Type Indicator) questionary with the aim of finding out if there is recurrence of psychological profiles and to understand the characteristics of the freshmen to learning. This information was subjected to statistical analysis to observe the recurrence of the psychological characteristics of freshmen over the years (2017 to 2021). The data was also subjected to the Self Organizing Maps (SOM) machine learning technique, to delineate and group the psychological profiles observed in the students, enabling the analysis of the psychological characteristics and the comparison with the results obtained via statistics. After that, the results from the MBTI Traditional and the SOM clustering were submitted to comparison methods through multivariate statistics and Multiple Correspondence Analysis (MCA). It was found that there is recurrence of psychological characteristics between the years of collection, what were the psychological profiles of these freshmen as to how they learned. It was found that the freshmen surveyed mostly belong to Generation Z (or iGen) and characteristics pertinent to this group were observed. A scenario is obtained to point out the best active learning techniques for these students with the verification of recurrence of profiles and other correlated psychological characteristics, aiming to provide the most effective teaching in the engineering course.

Coupled zoning and spatial heterogeneity of human activities and natural endowments based on self-organizing map and random forest: A case study of the agro-pastoral ecotone in Gansu, China

Regional zoning is an important way for humans to understand geographical space，but the existing research on regional zoning focuses on natural conditions, ignoring the interdependent characteristics of human and nature, and the driving mechanism of regional differentiation is not sufficiently explained. In this study, using algorithms such as self-organizing feature map, clustering quality index, structural similarity index measure (SSIM) and random forest, the human activity (HA)–natural endowment (NE) coupled zoning in the agro-pastoral ecotone in Gansu, China (AEGC), was conducted to explore the driving mechanism of regional differentiation. Results show the following: (1) The HA-NE coupled zoning has the best clustering effect when implementing the partitioning scheme with a classification number of 4. Accordingly, the AEGC can be divided into four regions with significant differences between HAs and NEs. (2) From the perspective of spatial distribution, HA–NE coupled zoning, climate zoning, geographic zoning, and vegetation zoning are similar, whereas HA zoning is different. The results of SSIM showed that HA–NE coupled zoning takes all types of factors into consideration (SSIM mean 0.708), and the zoning results are better than single type zonings. HAs have strong independence (SSIM mean 0.576), and geographic conditions are closely related to all other factors (SSIM mean 0.671). (3) Elevation is the most important driver of regional differentiation in the AEGC, with a contribution degree of 22.36%; other important drivers include land use intensity, precipitation, and normalized difference vegetation index, with contribution degree distributions of 17.38%, 16.34%, and 15.79%. Furthermore, the dominant factors of regional differentiation in each sub-region are different. This study emphasizes that regional characteristic factors and uneven spatial distribution factors are important drivers of regional differentiation, and land use intensity has become an important force influencing geographic space. Policy recommendations for zonal governance are made based on the inherent conditions of different regions. This study may provide a reference for scientific regional zoning and cognitive regional differentiation.

Hybrid Approach for the Financial Assessment of Companies using Fuzzy Multi-Criteria Decision-Making and Self-Organizing Maps

Hybrid Approach for the Financial Assessment of Companies using Fuzzy Multi-Criteria Decision-Making and Self-Organizing Maps

Fuzzy Logic-Based Fragility Curve Development for Steel Moment-Resisting Frames Considering Uncertainties in Seismic Response

In this paper, a comprehensive range of uncertainties is considered to assess the seismic abilities of a moment-resisting system. To incorporate the parameter of construction quality, which has a descriptive nature, a suitable fuzzy logic engine has been developed. This engine, for the first time, addresses the quantitative assessment of construction quality parameters based on linguistic variables, including map accuracy, worker skills, material quality, and site supervision conditions. Instead of using random selection, a self-organizing map (SOM) algorithm is employed to carefully select strong ground motion records, reducing time costs. By applying incremental dynamic analysis (IDA) results, analytical equations are derived for the response surface method. These equations determine the collapse fragility’s mean and standard deviation. The material quality is modeled using the fuzzy inference engine, with the coefficient of logarithm response surface. Collapse fragility curves are created by taking into account many of their material quality values and utilizing the fuzzy model to estimate the modeling parameter based on the logarithm regression coefficients. These curves take into consideration various sources of uncertainty. In countries with inadequate material quality control, it is important to take cognitive uncertainty into account when developing fragility curves. This will help improve the overall risk management strategy.

A dynamic correction method for the optimal value settings of the solution purification process at multiple time scales

The solution purification process includes multiple continuous reactors. Setting the key technical indicators of each reactor through global optimization is the prerequisite for realizing the optimal operation of the entire process. Affected by fluctuations in inlet conditions, adjustments of operating parameters, and random disturbances, the operating status of the solution purification process will change accordingly, causing the optimal value settings based on global optimization to become no longer applicable. To ensure the applicability of the optimal value settings as the process changes and considering that the production data collected at different time scales contain different process information, this study proposes a dynamic correction method for the optimal value settings of the solution purification process at multiple time scales. First, considering the low-frequency testing data that can reflect the operation effect, the low-frequency correction is realized by combining mechanism knowledge and expert experience. Second, based on the characteristic that the high-frequency detection data can reflect the changing operating status in time, a supervised self-organizing map method is proposed to classify the changing trends in the operating status. Finally, an integrated, spatiotemporal, just-in-time learning method (with multiple changing trends in the operating status) is proposed to realize high-frequency correction. The experimental results show that the proposed method can dynamically correct the optimal value settings and reduce resource consumption while ensuring product quality.

Explainable Histopathology Image Classification with Self-organizing Maps: A Granular Computing Perspective

The automatic analysis of histology images is an open research field where machine learning techniques and neural networks, especially deep architectures, are considered successful tools due to their abilities in image classification. This paper proposes a granular computing methodology for histopathological image classification. It is based on embedding tiles of histopathology images using deep metric learning, where a self-organizing map is adopted to generate the granular structure in this learned embedding space. The SOM enables the implementation of an explainable mechanism by visualizing a knowledge space that the experts can use to analyze and classify the new images. Additionally, it provides confidence in the classification results while highlighting each important image fragment, with the benefit of reducing the number of false negatives. An exemplary case is when an image detail is indicated, with small confidence, as malignant in an image globally classified as benign. Another implemented feature is the proposal of additional labelled image tiles sharing the same characteristics to specify the context of the output decision. The proposed system was tested using three histopathology image datasets, obtaining the accuracy of the state-of-the-art black-box methods based on deep learning neural networks. Differently from the methodologies proposed so far for the same purpose, this paper introduces a novel explainable method for medical image analysis where the advantages of the deep learning neural networks used to build the embedding space for the image tiles are combined with the intrinsic explainability of the granular process obtained using the clustering property of a self-organizing map.

A biologically inspired computational model of human ventral temporal cortex

Our minds represent miscellaneous objects in the physical world metaphorically in an abstract and complex high-dimensional object space, which is implemented in a two-dimensional surface of the ventral temporal cortex (VTC) with topologically organized object selectivity. Here we investigated principles guiding the topographical organization of object selectivities in the VTC by constructing a hybrid Self-Organizing Map (SOM) model that harnesses a biologically inspired algorithm of wiring cost minimization and adheres to the constraints of the lateral wiring span of human VTC neurons. In a series of in silico experiments with functional brain neuroimaging and neurophysiological single-unit data from humans and non-human primates, the VTC-SOM predicted the topographical structure of fine-scale category-selective regions (face-, tool-, body-, and place-selective regions) and the boundary in large-scale abstract functional maps (animate vs. inanimate, real-word small-size vs. big-size, central vs. peripheral), with no significant loss in functionality (e.g., categorical selectivity and view-invariant representations). In addition, when the same principle was applied to V1 orientation preferences, a pinwheel-like topology emerged, suggesting the model's broad applicability. In summary, our study illustrates that the simple principle of wiring cost minimization, coupled with the appropriate biological constraint of lateral wiring span, is able to implement the high-dimensional object space in a two-dimensional cortical surface.

Proactive cyber fraud response: a comprehensive framework from detection to mitigation in banks

PurposeThis study addresses the ever-increasing cyber risks confronting the global banking sector, particularly in India, amid rapid technological advancements. The purpose of this study is to de velop an innovative cyber fraud (CF) response system that effectively controls cyber threats, prioritizes fraud, detects early warning signs (EWS) and suggests mitigation measures.Design/methodology/approachThe methodology involves a detailed literature review on fraud identification, assessment methods, prevention techniques and a theoretical model for fraud prevention. Machine learning-based data analysis, using self-organizing maps, is used to assess the severity of CF dynamically and in real-time.FindingsFindings reveal the multifaceted nature of CF, emphasizing the need for tailored control measures and a shift from reactive to proactive mitigation. The study introduces a paradigm shift by viewing each CF as a unique “fraud event,” incorporating EWS as a proactive intervention. This innovative approach distinguishes the study, allowing for the efficient prioritization of CFs.Practical implicationsThe practical implications of such a study lie in its potential to enhance the banking sector’s resilience to cyber threats, safeguarding stability, reputation and overall risk management.Originality/valueThe originality stems from proposing a comprehensive framework that combines machine learning, EWS and a proactive mitigation model, addressing critical gaps in existing cyber security systems.

A transdiagnostic examination of cognitive heterogeneity in children and adolescents with neurodevelopmental disorders

ABSTRACT Children and adolescents with neurodevelopmental disorders demonstrate extensive cognitive heterogeneity that is not adequately captured by traditional diagnostic systems, emphasizing the need for alternative assessment and classification techniques. Using a transdiagnostic approach, a retrospective cohort study of cognitive functioning was conducted using a large heterogenous sample (n = 1529) of children and adolescents 7 to 18 years of age with neurodevelopmental disorders. Measures of short-term memory, verbal ability, and reasoning were administered to participants with attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), comorbid ADHD/ASD, and participants without neurodevelopmental disorders (non-NDD) using a 12-task, web-based neurocognitive testing battery. Unsupervised machine learning techniques were used to create a self-organizing map, an artificial neural network, in conjunction with k-means clustering to identify data-driven subgroups. The study aims were to: 1) identify cognitive profiles in the sample using a data-driven approach, and 2) determine their correspondence with traditional diagnostic statuses. Six clusters representing different cognitive profiles were identified, including participants with varying forms of cognitive impairment. Diagnostic status did not correspond with cluster-membership, providing evidence for the application of transdiagnostic approaches to understanding cognitive heterogeneity in children and adolescents with neurodevelopmental disorders. Additionally, the findings suggest that many typically developing participants may have undiagnosed learning difficulties, emphasizing the need for accessible cognitive assessment tools in school-based settings.

Spatiotemporal Evolution and Driving Factors of Ecosystem Supply and Demand Bundles: A Case Study in the Sichuan-Yunnan Ecological Buffer Area, China

Identifying the spatial characteristics of ecosystem service (ES) supply and demand is crucial for effective ecosystem management and restoration. Past related studies have primarily focused on balancing ES supply and demand and supply clustering, with less attention focused on the drivers of demand clustering and their spatial evolution. This study explored the spatiotemporal supply–demand dynamics in four crucial ecosystem services (ESs) in the Sichuan-Yunnan ecological buffer area region between 2005 and 2019, namely water yield, net primary production, soil conservation, and habitat quality. Self-organizing maps and geographical detectors were used to classify supply–demand ES bundles as their main drivers, respectively. The main results of the study included: (1) A decline in habitat quality, whereas net primary productivity, water yield, and soil conservation increased. However, there were increasing demands for habitat quality, water yield, and net primary productivity, despite the decrease in demand for soil conservation. (2) Demand for habitat quality was met by supply, whereas there were deficits in soil conservation, water yield, and net primary productivity, which contributed to the demand in the east exceeding that elsewhere. (3) The proportion of ES bundle 2 increased, whereas those of the remaining ES bundles declined. Similarly, the areas of ES demand bundles (ESDBs) 1 and 4 decreased, whereas those of 2 and 3 increased. While the spatial extent of the ESBs remained relatively stable, those of the ESDBs in the northern regions increased. Key factors influencing the spatial distribution of ES supply include human activities, population density, and precipitation, whereas land use, population density, and the human activity index primarily affect demand distribution. The results of this study can act as a reference for comprehensive regional ecosystem management.

Drought and its ecological risk bundle from the perspective of watershed hydrological cycle

The mechanisms underlying the impacts of climate change and vegetation dynamics on hydrological drought in humid regions are still lacking. In this study, we connected the four components of meteorology-soil-vegetation-runoff to investigate the spatio-temporal response relationship between vegetation growth and different drought types. Based on the Variable Infiltration Capacity model and the Self-organizing Map Algorithm, we proposed ecological risk bundles at the grid scale to characterize the potential impacts of different types and levels of drought on vegetation. Furthermore, we quantified the driving impact of temporal and spatial changes in vegetation coverage on the propagation of meteorological-hydrological drought. The study found that the centers of gravity for the occurrence frequencies of extreme and mild drought shifted towards regions where vegetation growth was influenced by climate change. In certain regions of the watershed, vegetation exhibits significant spatial and temporal heterogeneity in its response to stress caused by different drought types. From 2004 to 2014, the stress on vegetation caused by moderate and mild meteorological droughts weakened, while soil moisture stress intensified after 2014. Simultaneously, the impacts of climate change and vegetation growth on runoff reached 48.25 % and 35.13 % respectively, and their synergistic effects triggered changes in the risk of co-concurrent return periods for hydrological drought events. Under the 100-year design return period, the co-occurrence return period of runoff shifted from its natural state of 162.9 years to 52.8 years, and the joint return period reversed its scenario, becoming shorter than the co-occurrence return period.