Optimizing Anchorage Safety Under Typhoons: Key Factor Identification and Dynamic Tiered Management via SEM–fsQCA Hybrid Modeling

Abstract

Identifying and optimizing core factor configurations for anchorage operational safety under typhoon scenarios is critical to enhancing anchorage operational resilience and sustainable port development. This study develops a complementary hybrid SEM–fsQCA framework: key factors are identified via literature review and expert interviews; SEM quantifies factor correlations and contribution weights and corrects expert-evaluated anchorage capacity; six core factors are extracted, three typhoon types (heavy-rainfall, strong-wind, complex-track) are defined, and a coupled anchorage–typhoon case dataset is constructed. Subsequently, fsQCA performs necessary condition analysis and identifies causal configurations driving safety effectiveness. Based on these configurations, we establish a dynamic three-tier risk classification framework for refined anchorage management. Validated using 36 coupled cases (12 anchorages × 3 typhoon types) from Huizhou Port, a core hub in the Guangdong–Hong Kong–Macao Greater Bay Area, this framework enables adaptive vessel traffic scheduling throughout the entire typhoon cycle through dynamic tiered management. The proposed “identification-intervention-feedback” closed-loop governance model delivers theoretical rigor and operational implementation ability for coastal port typhoon risk mitigation.

1. Introduction

Against the backdrop of intensifying global climate change and rising frequency of extreme weather events, ports—as critical nodes in global trade and supply chain networks—face mounting challenges to operational safety and resilience [1,2]. Port sustainable development is a core priority in the contemporary maritime sector [3]. A sustainable port requires not only efficient operations and green technologies but also resilience to maintain functional continuity during extreme climate events [4,5]. However, frequent typhoon strikes severely test this resilience [6]. As the final refuge for vessels during typhoons, anchorages directly determine ports’ ability to maintain basic functions during disasters, thereby affecting supply chain stability [7]. Integrating anchorage typhoon safety capacity into port sustainable development frameworks thus holds significant theoretical and practical implications. Frequent landfalls of severe typhoons damage port infrastructure through extreme winds, waves, and storm surges, triggering cascading disruptions including operational suspensions, vessel congestion, and deteriorated emergency response—all of which undermine port continuity and regional economic stability [8]. Historical events highlight these risks: Typhoon Hato (2017) sank over 10 merchant vessels in the Pearl River Estuary, causing 16 fatalities and missing persons [9]; Super Typhoon Mangkhut (2018), with maximum sustained winds of Beaufort force 14–17, forced major Pearl River Estuary ports to close completely for over 48 h and shelter hundreds of vessels [10]. These events have caused substantial casualties, infrastructure damage, economic losses, and global supply chain disruptions. Accordingly, enhancing anchorage safety during typhoons by transitioning from reactive response to proactive risk governance is imperative for sustainable port development.

Anchorage safety during typhoons is a systemic issue encompassing engineering, management, and operational dimensions, and related research has evolved substantially. A literature review identifies two primary paradigms:

The first focuses on scenario-specific practical countermeasures. For example, Hu et al. verified strong wind defense effectiveness using the single-point mooring practice of the Hui Xiang Hai at Taihu Port during Typhoon Muifa [11]; Zhu et al. quantified mooring forces for Qiongzhou Strait ro-ro passenger ships and proposed a route-specific typhoon prevention guide [12]; Dong Baojian and Wang Dechun specialized in typhoon prevention measures for non-self-propelled engineering vessels such as barges, floating cranes, and drilling platforms [13]; Li Meiyun extended the research to the local government level, optimizing the typhoon emergency response process in District Y of Shenzhen [14]. These practical studies involve various technical factors (e.g., mooring equipment: chain length/scope, breaking strength, anchor types; ship dynamics: drift force, yawing motion, cable tension, windage area) as well as management and human factors (e.g., real-time monitoring, alert thresholds, emergency protocols). These regional and scenario-specific studies provide operational management solutions and empirical evidence, highlighting the urgency and complexity of anchorage scheduling and typhoon preparedness for vessels. However, they mostly rely on qualitative analysis and case-based induction, making it difficult to quantify the impact of factors, generalize conclusions, or support the formulation of dynamic strategies.

The second paradigm focuses on macro-scale modeling and assessment frameworks to enhance theoretical generalizability and interpretability, falling into three subcategories: First, high-precision hazard driver simulation—e.g., Xiong et al.’s high-resolution typhoon wind field model incorporating complex surface features for improved catastrophe risk inputs [15]. Second, systematic assessment of impacts and cascading risks—e.g., Wang Wenjing’s enhanced DSAEEF-LTD model (via new hazard parameters and algorithm optimization) for more physically sound and efficient coastal typhoon pre-assessment [16]; Tan Xinru et al.’s typhoon–storm surge–exposure network model, which quantifies propagation pathways and vulnerabilities in socio-economic systems and reveals cascading amplification of secondary/derivative disasters [17]; and Liu Hongguang et al.’s coupled wind–rain–landslide model identifying critical thresholds and spatial patterns for co-triggered collapses [18]. Together, these advance macro-level understanding of multi-hazard coupling and risk transmission. Third, port resilience quantification—e.g., Hui Tianyong et al.’s Bayesian-network-based probabilistic model covering absorption, adaptation, and recovery [19]. These macro-scale hazard models primarily focus on external stressors, including typhoon parameters (wind speed/direction, wave height, tidal current) and anchorage conditions (water depth, seabed sediment type, shelter geometry). However, their macro-outputs rarely translate into operational decisions like anchorage scheduling, and linear/independence-assumption models cannot capture nonlinear configurational causal mechanisms—a core limitation in identifying “multiple equivalent paths” or “necessary conditions”.

In summary, extant research has two critical research gaps: (1) micro-scale response measures lack systematic modeling support, limiting precise decision-making; (2) macro-scale models fail to adequately capture concurrent, nonlinear, and configurational causal mechanisms. To address these gaps, this study constructs a hybrid analytical framework combining structural equation modeling (SEM) and fuzzy-set qualitative comparative analysis (fsQCA), as shown in Figure 1. SEM quantifies factor effects and structural relationships, while fsQCA identifies multiple sufficient condition configurations for high safety performance. The findings are translated into a three-tier (A/B/C) dynamic anchorage management framework tailored to three typical typhoon scenarios: heavy-rainfall, strong-wind, and complex-track events. This work advances theoretical understanding of safety effectiveness generation in complex port systems, and provides an operational tool to transition anchorage management from static assessment to dynamic adaptive response.

Empirically, this study focuses on Huizhou Port, a core hub in the Guangdong–Hong Kong–Macao Greater Bay Area [20]. Located on the Pearl River Estuary’s east bank, it is among China’s most typhoon-prone ports (4–6 direct/major impacts annually) [21]. A total of 36 anchorage–typhoon coupled configurations are established by pairing its 12 planned anchorages with the three aforementioned typhoon scenarios. Key outputs include an optimized anchorage carrying capacity inventory and an fsQCA-derived three-tier dynamic classification system, providing a scientific, actionable decision basis for typhoon-period anchorage safety management. The remainder of the paper is organized as follows: Section 2 will elaborate on the specific steps of the hybrid SEM–fsQCA methodology, including questionnaire design, SEM path analysis, anchorage capacity optimization, and fsQCA calibration and analysis; Section 3 presents the empirical results: SEM weights, optimized capacities, and fsQCA configurational paths; Section 4 conducts a typological interpretation of the six configurational paths, constructs a dynamic tiered anchorage management framework applied to Huizhou Port, and discusses the limitations and generalizability of this framework; Section 5 concludes with research findings and managerial implications.

2. A Hybrid SEM–fsQCA Methodology for Port Anchorage Safety Under Typhoon Scenarios

This study adopts a hybrid SEM–fsQCA methodology underpinned by sequential synergistic logic [22,23]. Specifically, confirmatory factor analysis (CFA) is first applied within the SEM framework to estimate factor loadings of each observed indicator on the corresponding latent constructs. The structural model then estimates the standardized influence weights of each latent variable on anchorage typhoon safety performance (TSP). Based on these results, the three latent variables with the strongest total effects—meteorological and oceanographic conditions (MOC), inherent anchorage properties (IAP), and management support effectiveness (MSE)—are selected as the theoretical basis for fsQCA condition selection. For each latent variable, the two observed indicators with the highest factor loadings (six in total) are extracted as fsQCA conditions to ensure empirical validity and representativeness. Subsequently, fsQCA identifies multiple equivalent sufficient configurational pathways leading to high safety effectiveness. In this integrated framework, SEM addresses how much each factor independently affects the outcome, whereas fsQCA clarifies how factor combinations generate equivalent high safety outcomes across distinct configurations [24]. Together, these complementary approaches reveal the nonlinear, conjunctural causal mechanisms governing anchorage typhoon safety.

2.1. Structural Equation Modeling (SEM)

To quantify the correlations, relative weights, and structural relationships among key factors influencing anchorage typhoon safety effectiveness [25], and to correct the expert-assessed anchorage capacity, a structural equation model (SEM) was employed for empirical analysis using AMOS 24.0.

2.1.1. Questionnaire Design and Data Collection

Based on relevant port regulations, the current typhoon preparedness status at target port anchorages and related academic research, a questionnaire was designed to assess factors influencing safety effectiveness at target port anchorages (design process shown in Figure 2) The questionnaire consisted of 26 items: 2 demographic variables, 21 core scale items, and 3 multiple-choice open-ended questions. Core scale items employed a five-point Likert format [26], a simplified version of classic scale designs, where respondents selected options ranging from “strongly agree” to “strongly disagree”. The respondents of the questionnaire are the personnel working on the ship during the typhoon preparedness of the target port, the port management and supervision personnel and other relevant personnel [27]. We distributed 400 questionnaires and obtained 301 valid responses. Among the respondents, 15.28% were captains, 47.51% were crew members, 16.94% were port management personnel, 9.63% were maritime supervisors, and the remainder were relevant personnel (pilots, mooring teams, etc.). The sample coverage is accurate, and the sample size is sufficient.

2.1.2. Reliability Analysis

Reliability of the measurement model was evaluated using Cronbach’s α, with values ranging from 0 to 1 and higher values indicating better internal consistency. The reliability results (

Table 1. Reliability Analysis of the Anchoring Safety and Effectiveness Scale.

Dimension	Cronbach’s α	Number of Items
Meteorological and Oceanographic Conditions (MOC)	0.808	3
Inherent Anchorage Properties (IAP)	0.81	3
Vessel Maneuvering Status (VMS)	0.837	3
Ship Hardware Configuration (SHC)	0.802	3
Management Support Effectiveness (MSE)	0.898	6
Typhoon Safety Performance (TSP)	0.839	3
Influencing factors of safety efficiency of typhoon prevention in the anchorage area	0.908	21

) indicate that all dimensions had Cronbach’s α values greater than 0.8, and the overall scale α was 0.908, indicating excellent internal consistency and reliability.

2.1.3. Validity Analysis

① Confirmatory Factor Analysis (CFA)—CFA is a core statistical method within the Structural Equation Modeling (SEM) framework, primarily used to test the validity of the measurement model. Through CFA, we verify whether the theoretical factor structure of the measurement scale is supported by the data in the specific context and sample of this study. As illustrated in Figure 3, which presents CFA model fit results, $CMIN/DF=1.409$ falls within the excellent range (1–3), while $RMSEA=0.037$ remains below the critical threshold of 0.05. Additionally, IFI, TLI, and CFI all exceeded 0.90, indicating excellent fit.

② Convergent Validity—Convergent validity is assessed by calculating the Average Variance Extracted (AVE) and Composite Reliability (CR) to evaluate whether there are significant distinctions between different dimensions and whether the observed variables measuring the same dimension exhibit high inter-variable correlation. As presented in

Table 2. Convergent and discriminant validity of the scale.

Latent Variable	CR	AVE	MOC	IAP	VMS	SHC	MSE	TSP
MOC	0.809	0.585	0.765
IAP	0.810	0.587	−0.466	0.766
VMS	0.8385	0.635	−0.445	0.55	0.797
SHC	0.802	0.575	−0.43	0.475	0.448	0.758
MSE	0.902	0.606	−0.517	0.555	0.526	0.563	0.778
TSP	0.839	0.636	−0.528	0.559	0.555	0.532	0.594	0.797

, all dimensions had AVE values > 0.5 and CR values > 0.7, confirming sufficient convergent validity for each latent construct.

③ Discriminant Validity Test—Discriminant validity is used to verify whether different latent variables in the measurement model are statistically distinct, ensuring that the items measuring different constructs can be differentiated from one another and capture different dimensions. This study employs the Fornell–Larcker criterion for testing [28]. This criterion requires that the square root of the Average Variance Extracted (AVE) for any latent variable must be greater than the absolute value of the correlation coefficient between that latent variable and all other latent variables. Results in Table 2 show that the square root of AVE for each dimension exceeded the absolute value of its standardized correlation with all other latent variables, verifying adequate discriminant validity across all constructs.

2.2. Methodology for Anchorage Capacity Optimization Based on SEM Weights

To scientifically determine ultimate typhoon carrying capacity for each Huizhou Port anchorage, this study builds upon expert judgment, further optimizing and calibrating initial recommended capacity (IRC) using objective standardized weights derived from SEM analysis. The aim is to generate more precise and theoretically grounded capacity estimates.

2.2.1. Data Integration and Preprocessing

• Basic Capacity Data

Basic capacity data for each anchorage were sourced from the official report titled Research Report on Planning of Typhoon Shelter Waters for Huizhou Port (details in Appendix

Table A1. Basic characteristics and recommended capacity of Huizhou Port anchorages.

Number	Planned Water Area (104 m2)	Sediment Type	Representative Vessel Type	Anchorage Capacity (Vessels)
No. 1	1166	Fluid mud, Fine sand, Silt	General vessels with draft ≤4.3m	150–160
No. 2	302	Fluid mud, Fine sand, Silt	General vessels, 1000–3000 DWT	10
No. 3	635	Fluid mud, Silt, Silty sand	General vessels, ≤10,000 DWT	14
No. 4	506	Fluid mud, Fine sand, Silt	General vessels, ≤10,000 DWT	11
No. 5	2679	Silt	General vessels, 1000–3000 DWT	95
No. 6	1672	Silt	General vessels, 1000–70,000 DWT	55
No. 7	1903	Fluid mud, Silt, Silty sand	General vessels, 10,000–70,000 DWT	1–2
No. 8	1256	Fluid mud, Silt, Silty sand	Hazardous chemical vessels, 10,000–70,000 DWT	1
No. 9	1350.7	Silt	General vessels, 30,000–150,000 DWT	/
No. 10	473.8	Silt	General vessels, 30,000–100,000 DWT	/-
No. 11	5819.5	Silt	Hazardous chemical vessels, 50,000–300,000 DWT	/
No. 12	981	Silt	General vessels (10,000 DWT) or electrical construction platforms	21

). These data are defined in this study as the initial recommended capacity (IRC), which serves as the foundational input for subsequent optimization and calibration using SEM-derived weights.

• Anchorage Attributes and Weight Data

Core independent variable data were sourced from the SEM questionnaire dataset. For each anchorage, the average score of all sample items under each latent variable was calculated to obtain dimension-specific factor scores. Negatively scaled factors were reverse-coded using Formula (1) to ensure calculation consistency:

$$\mathit{\text{Positive Score}}=6-\mathit{\text{Raw Score}}$$

(1)

• Theoretical Capacity Score

Theoretical capacity scores (TCS) comprehensively reflect objective safety potential of anchorages. Calculation is based on scores of five factors (after normalization) for each anchorage and corresponding standardized weights determined through SEM analysis, derived using weighted summation:

$${{\displaystyle TCS}}_i={\displaystyle \sum _{K=1}^5({{\displaystyle F}}_{ik}\times {{\displaystyle w}}_k})$$

(2)

where ${{\displaystyle TCS}}_i$ represents theoretical capacity score for anchorage $\mathrm{i}$; ${{\displaystyle F}}_{ik}$ is the score for anchorage $i$ on factor $k$; and ${{\displaystyle w}}_k$ is the SEM standardized weight for the corresponding factor, satisfying ${\displaystyle \sum {{\displaystyle w}}_k}=1$. Calculation results are presented in Section 3 (Results Analysis).

2.2.2. Processing by Logarithmic Transformation

To address the significant order-of-magnitude variation (1 to 155) in capacity recommendations provided by experts, this study employs a decimal logarithmic transformation to convert exponential differences into linear variation [29]. This enables the construction of a stable, sensitive computational framework preserving inherent capacity differentials between anchorages. The calculation procedure is as follows:

• Calculate Logarithmic Benchmark Capacity (LBC)

To convert vast absolute differences in expert capacities into relative differences, base-10 logarithmic transformation is applied.

$${{\displaystyle LBC}}_i={{\displaystyle \log }}_{10}({{\displaystyle IRC}}_i)$$

(3)

where ${{\displaystyle IRC}}_i$ represents the expert-recommended capacity for the i-th anchorage. This reflects the order of magnitude of anchorage capacity.

• Determine the Adjustment Coefficient (AC)

Selection of the benchmark anchorage determines the reliability of model output. In safety-critical domains, slightly conservative yet credible results hold far greater value than potentially overestimated risk outcomes. Adhering to the principle of safety conservatism, this study establishes the highest-scoring Anchorage 1 as benchmark, setting a high-standard excellence benchmark. Calculation formula:

$${{\displaystyle ALC}}_i={{\displaystyle LBC}}_i\times {{\displaystyle AC}}_i$$

(4)

where ${{\displaystyle TCS}}_{base}$ is the score of the benchmark anchorage (No. 1). This coefficient reflects relative performance of the i-th anchorage compared with the benchmark across multiple factors.

• Calculate the Adjusted Logarithmic Capacity (ALC)

The logarithmic benchmark capacity is multiplied by the adjustment coefficient to achieve micro-level correction of macro-level capacity.

$${{\displaystyle ALC}}_i={{\displaystyle LBC}}_i\times {{\displaystyle AC}}_i$$

(5)

• Calculate the Final Recommended Capacity (FRC)

Adjusted logarithmic capacity is converted back to linear space of vessel quantities via the inverse logarithmic function, yielding optimized maximum carrying capacity.

$${{\displaystyle FRC}}_i={{\displaystyle 10}}^{{{\displaystyle ALC}}_i}$$

(6)

2.3. Fuzzy-Set Qualitative Comparative Analysis (fsQCA)

Anchorage typhoon safety is not the result of a single linear factor but rather a product of complex, multidimensional interplay, which exhibits equifinality and conjunctural causation [30]. While previous structural equation modeling effectively identified key influencing factors, it struggled to reveal how these factors interact and substitute one for another. This study applies fsQCA to transcend the examination of single-factor effects and delve deeper into the diverse pathways underlying typhoon resilience at Huizhou Port anchorages [31]. This provides direct scientific foundations for formulating differentiated and refined safety management strategies [32].

2.3.1. Selection of Typical Cases

fsQCA has no strict sample size requirements and is applicable to 15–80 cases [32]. To ensure the reliability and practicality of fsQCA in this study, this study follows the principles of theoretical relevance and purposive sampling. Based on key influencing factors identified through prior SEM analysis, we constructed typical cases centered on the extreme weather event “typhoon” as the core dimension [33]. This approach focuses the analysis on how anchorage attributes [34] and management conditions combine under different typhoon scenarios to jointly affect anchorage typhoon safety effectiveness [35]. The case set includes 12 anchorages in Huizhou Port and three representative 2023 typhoons (Doksuri, Saola, Haikui) that affected the port, which are classified into three types based on core disaster-inducing factors [36]: heavy-rainfall type, strong-wind type, and complex-track type (

Table 3. Key attributes of the three typical typhoons.

Name	Type	Primary Hazard-Inducing Factor	Prevailing Wind Direction	Prevailing Wind Force	Number of Sheltering Vessels
Doksuri	Heavy-Rainfall Type	Extreme precipitation and associated secondary hydrometeorological effects	NNE to N	6–10	305
Saola	Strong-Wind Type	Extreme wind stress	E to NE	7–11	333
Haikui	Complex-Track Type	Prolonged and multidirectional uncertain loads due to anomalous track	NE to E	6–8	328

Note: The prevailing wind direction and force are data recorded during the typhoon’s impact period on Huizhou Port.

). These three types cover the primary physical stress spectrum of typhoon hazards to anchorage safety, ensuring the representativeness and diversity of the case set.

The heavy-rainfall type is defined by extreme precipitation and associated secondary hydrometeorological effects as primary hazard drivers, with objective criteria of extreme cumulative or short-term rainfall intensity triggering severe flooding, landslides, or hydrological anomalies in port waters. Typhoon Doksuri (Figure 4) [37] is a representative case: despite concurrent strong winds, its impacts were dominated by torrential rainfall and subsequent waterlogging, which drastically reduced visibility and impaired vessel navigation and port traffic management. This type primarily tests port systems’ monitoring, early warning, navigation management, and disturbance resistance under coupled hydrometeorological effects in non-peak wind pressure environments.

The strong-wind type is defined by extreme wind stress as the key disaster-inducing factor, characterized by the typhoon’s maximum sustained wind speed near its center or persistent winds in its outer circulation reaching destructive levels, with wind load constituting the dominant contributor to total disaster losses. Typhoon Saola (Figure 5) [38] exemplifies this category, featuring a compact circulation and rapid intensification in a high-intensity, rapid-impact pattern. Huizhou Port experienced an abrupt 180° wind direction shift and a wind force surge from Beaufort 6 to 12 within 6 h; this drastic wind load vector change critically tested the anchorages’ effective shelter capacity, while the sharply compressed time window between early warning response and vessel evacuation directly examined the decision-making precision and execution agility of the emergency management system under extreme conditions.

The complex-track type is defined by prolonged, multi-directional and uncertain impacts from anomalous track characteristics, with key discriminant criteria including looping, stagnation, meandering, or multiple abnormal directional changes. Typhoon Haikui (Figure 6) [39] typifies this pattern: its highly sinuous track and extremely long lifespan subjected Huizhou Port to a dynamic compound hazard mode, with repeated multi-phase impacts of wind, rainfall and storm surge from varying directions and fluctuating intensities over several days, rather than a single peak event. Repeated wind direction shifts imposed stringent requirements on the multi-directional adaptability of anchorage sheltering, while the extended emergency period severely tested the sustained resilience of the port management system.

In summary, the selected typhoon events systematically cover the primary physical stress spectrum of typhoon hazards to port anchorage safety and fully reflect the performance characteristics of Huizhou Port’s anchorage typhoon preparedness system across different spatiotemporal scales and disaster mechanisms, ensuring the representativeness and diversity of the case set.

We constructed the analysis dataset by individually pairing each anchorage with each of the three typical typhoons, forming distinct “anchorage–typhoon” pairs as the unit of analysis. This resulted in a final dataset comprising 36 anchorage–typhoon preparedness cases, as presented in Appendix

Table A2. Case dataset.

Case	Anchorage	Typhoon	WFA	SWH	EWS	AST	RAT	DLD	SEA
1	No. 1	Doksuri	7	1.5	140	0.8	24.5	0.935	9.1
2	No. 1	Saola	8	1.5	110	0.8	34.5	0.994	8.3
3	No. 1	Haikui	6	1	120	0.8	17	0.961	7.8
4	No. 2	Doksuri	8	2.5	125	0.8	24.5	0.556	8.9
5	No. 2	Saola	9	2.5	95	0.8	34.5	0.889	7.3
6	No. 2	Haikui	6	2	105	0.8	17	0.667	7.9
7	No. 3	Doksuri	8	2.5	100	0.5	24	0.769	8.6
8	No. 3	Saola	10	3	130	0.5	32.5	0.923	8.2
9	No. 3	Haikui	6	2	125	0.5	16	0.923	8.9
10	No. 4	Doksuri	8	3	135	0.8	24	0.500	8.3
11	No. 4	Saola	10	3	150	0.8	32.5	1.100	9
12	No. 4	Haikui	7	2	145	0.8	16	0.600	9.1
13	No. 5	Doksuri	8	2.5	110	0.2	23	0.908	8.9
14	No. 5	Saola	9	3	75	0.2	32.5	0.989	7.2
15	No. 5	Haikui	6	2	80	0.2	16	0.943	7.5
16	No. 6	Doksuri	8	3	90	0.2	24	1.061	8.4
17	No. 6	Saola	9	3	85	0.2	32.5	1.143	7.3
18	No. 6	Haikui	7	2	90	0.2	16	1.122	7.7
19	No. 7	Doksuri	8	3	65	0.5	20.5	2.000	7.5
20	No. 7	Saola	10	4	115	0.5	29.5	2.000	8.3
21	No. 7	Haikui	8	2.5	110	0.5	14	3.000	7.5
22	No. 8	Doksuri	10	4	55	0.5	21.5	2.000	5.4
23	No. 8	Saola	11	5	95	0.5	30.5	1.000	5.4
24	No. 8	Haikui	8	2.5	90	0.5	14	2.000	7.4
25	No. 9	Doksuri	10	4	60	0.2	21.5	1.000	5.4
26	No. 9	Saola	11	4	100	0.2	30.5	1.000	6.9
27	No. 9	Haikui	8	2.5	95	0.2	14	1.000	6.9
28	No. 10	Doksuri	10	4	45	0.2	20.5	1.000	5.5
29	No. 10	Saola	11	5	80	0.2	29.5	1.000	4.9
30	No. 10	Haikui	8	2.5	75	0.2	14	1.000	5.2
31	No. 11	Doksuri	10	4	50	0.2	20.5	1.000	4.9
32	No. 11	Saola	11	4.5	85	0.2	29.5	1.000	4.6
33	No. 11	Haikui	8	2	80	0.2	14	1.000	5.8
34	No. 12	Doksuri	9	2.5	130	0.2	24.5	0.250	7.6
35	No. 12	Saola	10	2.5	105	0.2	34.5	0.150	7.8
36	No. 12	Haikui	7	2	115	0.2	17	0.250	8.3

Notes: Values represent the effective wind shelter (EWS) in degrees for a 180° sector facing the typhoon’s approach direction; descriptions in parentheses detail the primary geographical or man-made features providing the shelter. Emergency Response Timeline. During Typhoon Doksuri: Level IV emergency response activated at 0930 h, 25 July; upgraded to Level III at 2200 h, 25 July; response terminated at 1030 h, 28 July. During Typhoon Saola: Typhoon preparedness Level IV response activated at 0830 h, 29 August; upgraded to Level III at 1200 h, 30 August; further upgraded to Level II at 0800 h, 31 August; escalated to Level I at 2000 h, 31 August; downgraded to Level III at 0930 h, 2 September; further downgraded to Level IV at 1200 h, 2 September; typhoon response terminated at 1600 h, 2 September. During Typhoon Haikui: Level IV response activated at 1900 h, 3 September; terminated at 0900 h, 5 September.

.

2.3.2. Condition Specification and Model Construction

This study strictly adheres to the principle of integrating theoretical guidance and empirical support for condition specification and research model construction. Causal conditions were derived directly from SEM results, which both identified key latent variables affecting anchorage safety and quantified the contribution of each observed indicator to its corresponding latent construct via factor loadings. First, we identified the three core latent variables with the strongest explanatory power for the outcome: MOC, IAP, and MSE. Subsequently, to ensure fsQCA conditions were operational, measurable, and representative of core latent constructs, the two observed indicators with the highest factor loadings were extracted from each core latent variable (higher loadings indicate stronger indicator representativeness and reliability). As shown in Figure 2, the selected indicators with the highest standardized factor loadings are MOC1 (strong wind threat), MOC2 (high wave threat), IAP1 (anchorage geographic shelter), IAP2 (anchorage seabed holding power), MSE6 (emergency response dispatch), and MSE3 (anchorage allocation and scheduling). Given data availability, their strong representativeness, and relevance to each typhoon event, we quantified the six observed indicators.

(1)

The threat of strong winds: Quantified by the actual wind force at the anchorage during each typhoon impact period to reflect its effect on anchorage safety effectiveness [40].

(2)

The threat of huge waves: Quantified by the significant wave height at the anchorage during each typhoon impact period to reflect its hazard level to anchorage typhoon safety.

(3)

Geographical shelter of anchorage: To capture dynamic cross-case differences under risk scenarios, effective wind shelter for a specific typhoon was adopted as the measurement indicator. The prevailing wind direction of each typhoon was obtained from meteorological data, and a 180° sector centered on the anchorage and facing the typhoon’s approach direction was defined. Effective wind shelter was calculated as the sum of azimuth angles occupied by all sheltering objects within this sector (detailed results in Appendix Table A2).

(4)

Anchorage seabed holding power: A static inherent attribute of an anchorage with a constant value across typhoon cases for a given anchorage. It was quantified by anchorage sediment type, with ordinal values assigned to three substrate types: 0.8 for fluid mud–fine sand–silt composites, 0.5 for fluid mud–silt–silty sand composites, and 0.3 for silt-dominated substrates (detailed sediment types in Appendix Table A1).

(5)

Emergency response dispatch: This refers to the system that organizes stakeholders to complete risk avoidance and emergency response tasks via unified command, resource allocation, and process coordination within limited timeframes [41]. Sufficient lead time is the core premise of effective emergency dispatch; thus this indicator was quantified by response advance time, defined as Formula (7):

$$RAT={{\displaystyle T}}_{impact}-{{\displaystyle T}}_{response}$$

(7)

where ${{\displaystyle T}}_{impact}$ denotes the start time of significant typhoon impact, ${{\displaystyle T}}_{response}$ denotes time of first emergency response activation.

The results are provided in Appendix Table A2.

(6)

The most direct observable outcome of this dimension is vessel number and concentration in each anchorage within a specific time window. Considering the order-of-magnitude differences in Huizhou Port’s anchorage areas, dispatch load distribution during each typhoon impact period was adopted to quantify individual anchorage operational load pressure. For anchorages with non-zero recommended capacity, it was calculated as Formula (8):

$${{\displaystyle DLD}}_i={\displaystyle \frac{{{\displaystyle N}}_{peak}}{{{\displaystyle FRC}}_i}}$$

(8)

where ${{\displaystyle DLD}}_i$ denotes dispatch load distribution for the i-th anchorage; ${{\displaystyle N}}_{peak}$ denotes the peak number of vessels within the anchorage during the typhoon event; and FRC_i is optimized recommended capacity for the i-th anchorage.

For anchorages with zero recommended capacity (typically indicating major safety deficiencies such as extremely poor sheltering or inadequate management), a value of 1 was directly assigned regardless of actual vessel number, representing full membership in the high-load set (calculated results in Appendix Table A2).

(7)

Anchorage typhoon safety effectiveness: To ensure reliable, comparable measurement, this outcome variable was comprehensively assessed across wind/wave resistance, inherent waterway properties, and emergency support capability to evaluate overall vessel safeguarding performance under different typhoon scenarios. It was quantified via a 1–10 expert scoring scale, where 1 indicates extremely poor effectiveness and 10 indicates optimal effectiveness. Experts assigned a safety effectiveness score to each anchorage under each typhoon based on actual conditions and observed safety performance.

In summary, the fsQCA research model is established as shown in Figure 7.

2.3.3. Data Processing and Calibration

• Variable Nomenclature

Wind force at anchorage is denoted as “WFA” (~WFA for negation); significant wave height as “SWH” (~SWH); effective wind shelter as “EWS” (~EWS); anchorage sediment type as “AST” (~AST); response advance time as “RAT” (~RAT); dispatch load distribution as “DLD” (~DLD); and the outcome variable safety effectiveness assessment as “SEA” (~SEA).

• Anchor Setting and Variable Calibration

The core principle of fsQCA is to convert raw data into set-theoretic concepts via calibration—a process that transforms measured variables into continuous fuzzy-set membership scores (0–1) to assign interpretable set membership to the original data. Calibration is standardly performed using three sequential qualitative anchors: full membership, crossover point, and full non-membership. For variables without well-established theoretical or empirical calibration standards, the 90th, 50th, and 10th percentiles of the data distribution were adopted as the thresholds for the three corresponding anchors to minimize subjective bias (the calibration anchors and descriptive statistics for all conditions and the outcome variable are presented in

Table 4. Data calibration and descriptive statistics for the outcome and conditions.

Calibration Anchors				Descriptive Statistics
Variable	Full Membership	Crossover Point	Full Non-Membership	Mean	Standard Deviation	Maximum	Minimum
WFA	10.5	8	6.5	8.556	1.520	11	6.000
SWH	4	2.5	2	2.847	0.977	5	1.000
EWS	132.5	97.5	62.5	98.889	26.836	150	45
AST	0.8	0.35	0.2	0.425	0.252	0.8	0.200
RAT	32.5	23.5	14	23.361	7.039	34.5	14.000
DLD	2	0.943	0.4	1.045	0.551	3	0.150
SEA	8.9	7.55	5.3	7.325	1.372	9.1	4.6

Note: Since the calibrated membership score for the causal condition DLD was exactly 0.5 in some cases, a constant of 0.001 was added to this condition following the principle of conservative analysis.

). This approach transformed all condition and outcome variables into fuzzy sets with 0–1 bounded membership scores, where the score magnitude represents the degree of a case’s membership in the target set. To avoid logical contradictions (cases with an exact membership score of 0.5 are automatically excluded by fsQCA software), a trivial constant (±0.001) [42] was added to the raw values of relevant conditions as needed to eliminate this exact value. All calibrations were performed using fsQCA 4.1 software, and resulting membership scores (rounded to two decimal places) were used as the input for subsequent fsQCA calculations.

2.3.4. fsQCA Analysis and Robustness Check

• Necessity analysis of single conditions

Necessity analysis was conducted to identify whether any single condition constitutes a necessary condition for the outcome. In standard QCA procedures, the necessity of a single variable is determined by its consistency score [43], calculated as Formula (9):

$$Consistency({{\displaystyle X}}_i\le {{\displaystyle Y}}_i)={\displaystyle \sum [\min ({{\displaystyle X}}_i},{{\displaystyle Y}}_i)]/{\displaystyle \sum {{\displaystyle X}}_i}$$

(9)

For a condition X (single or combinatorial) to be sufficient for the outcome Y, its fuzzy-set membership score must be less than or equal to that of Y, with a consistency threshold set at 0.8. For necessity, the threshold is set at 0.9: conditions with consistency exceeding 0.9 are identified as necessary conditions for the outcome. Following necessity and sufficiency assessment, the explanatory power of X for Y is quantified by coverage, calculated as Formula (10):

$$Coverage({{\displaystyle X}}_i\le {{\displaystyle Y}}_i)={\displaystyle \sum [\min ({{\displaystyle X}}_i},{{\displaystyle Y}}_i)]/{\displaystyle \sum {{\displaystyle Y}}_i}$$

(10)

Coverage indicates the empirical explanatory scope of condition X for the outcome, with higher values indicating stronger explanatory power.

• Truth table construction

From a configurational perspective, a truth table was constructed to analyze the combined effects of causal conditions. Consistent with well-established QCA methodological guidelines [44], the raw consistency threshold was set at 0.8, and the proportional reduction in inconsistency (PRI) consistency threshold at 0.65. Given that there were only 36 valid cases, the case frequency threshold was set at 1. Configuration analysis was conducted using fsQCA 4.1 software, with SEA as the outcome variable. To obtain theoretically meaningful and parsimonious solutions, this study primarily reports the intermediate solution (Appendix

Table A3. Intermediate solution for high anchorage typhoon safety effectiveness.

Path	Raw Coverage	Unique Coverage	Consistency
~WFA * ~SWH * EWS * ~DLD	0.488901	0.0586681	0.987193
~WFA * ~SWH * EWS * AST	0.44556	0.0919663	0.95362
~SWH * EWS * RAT * ~DLD	0.362579	0.0232558	0.988473
~WFA * ~SWH * AST * RAT * ~DLD	0.227273	0.00105715	0.966292
WFA * SWH * AST * RAT * ~DLD	0.207717	0.0570825	0.895216
solution coverage	0.718288
solution consistency	0.927012

Notes: “*” denotes logical AND, meaning that all conditions connected by * must be satisfied simultaneously.

), which is derived based on theoretical assumptions for counterfactual cases [45]. Based on theoretical logic, practical feasibility, and the expected impact of each condition on the outcome, counterfactual assumptions were set as follows: conditions with a clear positive effect on anchorage safety (EWS, AST, RAT) were assumed present; conditions with a clear negative effect (DLD) were assumed absent; typhoon-induced hazard conditions (WFA, SWH) were set as present or absent with no prior assumptions, to fully explore safety configurations under diverse meteorological and oceanographic scenarios.

• Robustness check

Given the case-oriented nature of QCA and the limited sample size, two robustness checks were performed following well-established methodological protocols [46]:

(1)

The raw consistency threshold was increased from 0.8 to 0.85.

(2)

One case (Anchorage 8) was randomly removed from the sample.

Both checks yielded conditional configurations substantially consistent with the original results, confirming the reliability of the findings for subsequent analysis. Detailed robustness check results are presented in Appendix

Table A4. Intermediate solution for high anchorage typhoon safety effectiveness (robustness check).

Path	Raw Coverage	Unique Coverage	Consistency
~WFA * ~SWH * EWS * ~DLD	0.490196	0.0588236	0.987193
~WFA * ~SWH * EWS * AST	0.446741	0.09221	0.95362
~SWH * EWS * RAT * ~DLD	0.36354	0.0233175	0.988473
~WFA * ~SWH * AST * RAT * ~DLD	0.227875	0.00105995	0.966292
WFA * SWH * AST * RAT * ~DLD	0.205617	0.0572338	0.984772
solution coverage	0.717541
solution consistency	0.95285

Notes: “*” denotes logical AND, meaning that all conditions connected by * must be satisfied simultaneously.

.

2.4. Ethics Statement

This study involved no human participants, clinical interventions, or vulnerable populations. The questionnaire-based surveys were conducted as part of routine port and maritime industry safety and operational assessments—specifically examining anchorage safety under typhoon conditions, current operational practices, and management effectiveness. No experimental manipulations were performed; no sensitive personal information, biological samples, or personally identifiable data were collected. All data were anonymized and used exclusively for academic research and statistical analysis, in full compliance with international academic integrity standards and institutional research ethics guidelines. Accordingly, formal ethical review approval and written informed consent were not required for this study.

3. Results Analysis

3.1. SEM Path Analysis and Weight Quantification

A structural equation model (SEM) was established to validate the research hypotheses (Figure 8). Statistical significance was evaluated via the critical ratio (C.R.), where a C.R. > 1.96 indicates statistical significance at the p < 0.05 level. The actual impact and standardized weights of each dimension on typhoon safety effectiveness were calculated based on path relationship test results.

① Structural Equation Model Fit Test—As shown in Figure 8, the structural equation model for factors influencing anchorage typhoon safety effectiveness demonstrates adequate overall fit.

② SEM Path Test—This study proposed 11 hypotheses. Test results, presented in

Table 5. Results of path relationship tests in the SEM for factors influencing anchorage typhoon safety effectiveness.

Structural Path			Estimate	S.E.	C.R.	p
IAP	<---	MOC	−0.47	0.078	−6.381	*
MSE	<---	MOC	−0.334	0.079	−4.684	*
MSE	<---	IAP	0.398	0.075	5.558	*
SHC	<---	VMS	0.173	0.067	2.319	0.02
SHC	<---	MSE	0.406	0.078	4.836	*
SHC	<---	IAP	0.194	0.084	2.264	0.024
TSP	<---	MOC	−0.199	0.082	−2.578	0.01
TSP	<---	MSE	0.223	0.081	2.64	0.008
TSP	<---	IAP	0.191	0.082	2.335	0.02
TSP	<---	VMS	0.228	0.064	3.265	0.001
TSP	<---	SHC	0.174	0.079	2.256	0.024

Note: “*” indicates p < 0.001.

and Figure 7, indicated that the statistical data support all hypothesized paths, thus validating all research hypotheses.

③ Weight Quantification—Total effects, defined as the sum of direct and all indirect effects, were calculated for each factor. The results are presented in

Table 6. Analysis of the influence weights of dimensions on anchorage typhoon safety effectiveness.

Influencing Factor	Total Effect (Influence Weight)	Standardized Weight (Relative Proportion)	Effect Direction
MOC	−0.458	0.300	Negative
IAP	0.342	0.224	Positive
MSE	0.294	0.193	Positive
VMS	0.258	0.169	Positive
SHC	0.174	0.114	Positive

.

3.2. Anchorage Capacity Optimization Based on SEM Weights

Applying the logarithmic transformation procedure described in Section 2.2.2, the final recommended capacities (FRC) for each anchorage were obtained. Calculation results are shown in

Table 7. Logarithmically normalized capacity optimization table.

Anchorage	IRC	TCS	LBC	AC	ALC	FRC
No. 1 (benchmark)	155.0	3.803	2.190	1.000	2.190	155.00
No. 2	10.0	3.703	1.000	0.974	0.974	9.41
No. 3	14.0	3.702	1.146	0.974	1.116	13.05
No. 4	11.0	3.768	1.041	0.991	1.032	10.76
No. 5	95.0	3.739	1.978	0.983	1.944	87.99
No. 6	55.0	3.711	1.740	0.976	1.698	49.91
No. 7	1.5	3.280	0.176	0.862	0.152	1.42
No. 8	1.0	3.766	0.000	0.990	0.000	1.00
No. 9	/	3.697	/	/	/	/
No. 10	/	3.752	/	/	/	/
No. 11	/	3.702	/	/	/	/
No. 12	21.0	3.796	1.322	0.998	1.320	20.88

Note: Experts recommend using the midpoint value for range-based capacities, keeping integers unchanged, and temporarily excluding “/” from calculations.

.

Optimization results in Table 7 indicate that Anchorage No. 1 can accommodate up to 155 vessels; Anchorage No. 2, up to 9; Anchorage No. 3, up to 13; Anchorage No. 4, up to 10; Anchorage No. 5, up to 87; Anchorage No. 6, up to 49; Anchorage No. 7, up to 1; Anchorage No. 8, up to 1; and Anchorage No. 12, up to 20.

3.3. fsQCA Results Analysis

3.3.1. Necessity Analysis Results

Following the necessity analysis procedure described in Section 2.3.4, this study employed fsQCA 4.1 software to perform tests for necessary conditions. The results are presented in

Table 8. Results of necessity analysis for single conditions.

Condition	High Anchorage Typhoon Safety Effectiveness		Non-High Anchorage Typhoon Safety Effectiveness
	Consistency	Coverage	Consistency	Coverage
WFA	0.362051	0.494942	0.624707	0.770954
~WFA	0.832452	0.710740	0.590749	0.455325
SWH	0.538055	0.548787	0.812061	0.747709
~SWH	0.752643	0.816046	0.509953	0.499140
EWS	0.845137	0.859677	0.432084	0.396774
~EWS	0.406977	0.442529	0.847190	0.831609
AST	0.659620	0.779026	0.360656	0.384519
~AST	0.478858	0.453453	0.792740	0.677678
RAT	0.575053	0.621714	0.550937	0.537714
~RAT	0.572410	0.585406	0.612412	0.565405
DLD	0.555497	0.477510	0.894028	0.693776
~DLD	0.643763	0.870622	0.326698	0.398856

Note: “~” denotes the negation of a condition.

.

The results indicate that none of the causal conditions qualify as necessary conditions for the outcome, as the consistency scores for all individual conditions are below the 0.9 threshold. This suggests that anchorage typhoon safety effectiveness is not determined by any single factor.

3.3.2. Sufficiency Analysis and Configuration Paths

Following the truth table construction procedure described in Section 2.3.4, the configuration analysis yielded six paths leading to high anchorage typhoon safety effectiveness. The configuration paths were further analyzed and discussed in depth according to the logical solution scheme proposed by Ragin (

Table 9. Configuration analysis for high anchorage typhoon safety effectiveness.

Condition	Configuration Paths
	H1	H2	H3	H4	H5	H6
WFA	⊗	⊗			⊗	⬤
SWH	⊗	⊗	⊗		⊗	⬤
EWS	⬤	$⬤$	$⬤$	$⬤$		$⬤$
AST		$⬤$	$⬤$	$⬤$	⬤	$⬤$
RAT			⬤	⬤	⬤	⬤
DLD	$\otimes$		$\otimes$	$\otimes$	$\otimes$
consistency	0.987	0.954	0.988	0.922	0.966	0.895
raw coverage	0.489	0.446	0.363	0.267	0.227	0.208
unique coverage	0.059	0.092	0.023	0.008	0.001	0.057
solution consistency	0.927
solution coverage	0.718

Note: $⬤$ indicates the presence of a core condition; $\otimes$ indicates the absence of a core condition; ⬤ indicates the presence of a peripheral condition; ⊗ indicates the absence of a peripheral condition; blank cells indicate that the condition is irrelevant (i.e., its presence or absence does not substantially affect the outcome).

).

Combining the intermediate and parsimonious solutions, a configuration table for fuzzy-set qualitative comparative analysis was constructed. Specifically, under the joint influence of the six causal conditions, six configuration paths for high anchorage typhoon safety effectiveness emerged (labeled H1–H6). The overall solution coverage for these six high-safety-effectiveness configurations was 0.718, and the overall solution consistency was 0.927. This indicates that 71.8% of the cases achieving high safety effectiveness can be explained or covered by these configurations, and among the cases fitting these configurations, 92.7% indeed achieved high safety effectiveness. These two metrics collectively underscored the significance and reliability of the configuration paths for high anchorage typhoon safety effectiveness.

4. Analysis and Discussion of Configuration Results

The six configuration paths identified in Section 3.3.2 (Table 9) reveal multiple pathways to high anchorage typhoon safety effectiveness. This section interprets these paths by synthesizing them into a typological framework, followed by empirical application at Huizhou Port.

4.1. Typological Interpretation of Configuration Paths

The interpretation of configurational paths in this study does not rely solely on the presence or absence of conditions in statistical tables (as shown in Table 9), but is grounded in domain-specific practical knowledge of anchorage safety operations. In assessing whether a particular path possesses practical plausibility, we draw upon professional expertise: effective wind shelter angle reflects the physical protection capacity of an anchorage against prevailing wind directions; sediment type is directly related to anchor holding effectiveness; and response advance time represents the early warning and organizational efficiency of the management system, among others. Guided by this professional knowledge, the study synthesizes the six paths into a tripartite framework with clear practical implications based on their core driving logics, condition configurations, and managerial implications, thereby ensuring that the interpretative results are comprehensible and actionable for port management personnel. This framework not only categorized the surface manifestations of the paths but also revealed the underlying mechanisms that generate safety effectiveness.

4.1.1. Type 1—Intrinsic-Property-Dominant

Type 1 pathways are primarily driven by superior intrinsic anchorage properties (EWS and AST). These anchorages maintain high safety performance even with suboptimal managerial factors (e.g., RAT) or unfavorable natural conditions, reflecting the engineering principle that site selection and construction define the upper limit of anchorage safety capacity [47]. This type includes three pathways:

(1)

Path H1 (~WFA * ~SWH * EWS * ~DLD): Characterized by favorable meteorological and oceanographic conditions with low WFA and low SWH, where the anchorage itself possesses high effective wind shelter (EWS). Concurrently, effective port scheduling management controls DLD, avoiding congestion risks and thereby achieving anchorage safety. It illustrates the robust state achievable by combining basic infrastructure and management protocols under low typhoon intensity.

(2)

Path H2 (~WFA * ~SWH * EWS * AST): Building on H1, this configuration is driven by the synergistic combination of EWS and AST [48]. It maintains safety via intrinsic attributes under low wind-wave conditions, with higher tolerance to variations in emergency response timeliness and vessel density, but remains sensitive to meteorological hazard intensification, requiring supplementary Type 2 management strategies beyond certain thresholds.

(3)

Path H6 (WFA * SWH * EWS * AST * RAT): This is the most resilient configuration, sustaining high safety even under extreme MOC conditions through the synergy of superior intrinsic properties and timely emergency response (RAT), representing the upper bound of a port’s typhoon resistance capacity. Huizhou Port should identify and upgrade anchorages with such potential to serve as “Strategic Emergency Anchorages” for large, high-risk vessels unable to evacuate during typhoons.

In summary, Type 1 pathways cover the full functional spectrum of intrinsic properties, from leveraging baseline advantages to withstanding extreme hazards, with EWS and AST as fundamental determinants of safety effectiveness. Port managers should conduct structural performance assessments via geotechnical surveys and hydrodynamic modeling to formulate differentiated anchorage utilization strategies.

4.1.2. Type 2—Management-Support-Dominant

Type 2 pathways compensate for intrinsic property deficiencies or challenging natural conditions through efficient operational management and response systems. When anchorages lack optimal inherent attributes or face complex meteorological scenarios, warning response timeliness (RAT) and scheduling rationality (~DLD) become decisive factors. This type includes three pathways:

(1)

Path H3 (~SWH * EWS * RAT * ~DLD): This path addresses scenarios with high wind force but low significant wave height. Safety is achieved through the synergy of high EWS, early RAT, and controlled DLD. Success depends on anticipating wind intensification, promptly directing vessels to sheltered anchorages, and preventing overload, emphasizing targeted matching between warning response and scheduling allocation.

(2)

Path H4 (EWS * AST * RAT * ~DLD): This path exemplifies the synergistic alliance between intrinsic properties and management conditions. This configuration is indifferent to immediate WFA and SWH conditions. When anchorages possess adequate EWS and AST, combined with early warning (RAT) and load control (~DLD), high safety effectiveness is achievable regardless of immediate meteorological conditions. This “responding to variability with invariability” mode relied not only on the port’s deep understanding of its anchorages’ intrinsic properties but also on powerful forecasting and organizational coordination capabilities [49], representing an ideal operational state for port typhoon management.

(3)

Path H5 (~WFA * ~SWH * AST * RAT * ~DLD): This path integrates multiple favorable conditions—low wind and waves, favorable AST, early response, and controlled load distribution. Notably, the AST-RAT combination can partially substitute for high EWS, indicating substitutability among safety drivers. This configuration serves as a benchmark for evaluating typhoon preparedness plan completeness and execution fidelity, informing the optimization of warning criteria, allocation rules, and load-monitoring protocols.

In summary, Type 2 pathways highlight the central role of the warning response–scheduling allocation loop. The consistent co-occurrence of RAT and ~DLD indicates that isolated early warning or simplistic load restrictions are insufficient; warning information must translate into executable scheduling actions ensuring capacity compliance to achieve genuine safety.

4.1.3. Type 3—Load-Control-Dominant

The type 3 paths stem from a profound insight derived from the parsimonious solution. By stripping away all redundant conditions, the parsimonious solution points directly to the most fundamental driving factors. Notably, ~DLD (low dispatch load distribution) serves as a fundamental driver with raw coverage of 0.644 and unique coverage of 0.230. This indicates that rational load control independently explains 23% of high-effectiveness cases—unmatched explanatory power among single conditions. While this finding partly reflects calibration rules (zero-capacity anchorages assigned full high-load membership), it undeniably underscores the critical importance of scheduling and allocation. Load control is a foundational prerequisite: even Type 1 anchorages with superior properties or Type 2 scenarios with robust management suffer compromised safety under overcrowding. Thus, rational DLD control is the most cost-effective intervention for enhancing port typhoon preparedness.

4.2. Empirical Application of the Evaluation Model at Huizhou Port Anchorage

The three configuration types are not mutually exclusive but form a three-dimensional hierarchical framework for optimizing the typhoon safety performance of Huizhou Port anchorages. Hierarchically, Type 1 (intrinsic properties) constitutes the physical infrastructure layer, which defines the long-term upper limit of the port’s typhoon resistance capacity and resource base; Type 2 (management support) acts as the operational execution layer, which determines the efficiency and resilience of the port’s typhoon response under given infrastructure conditions; Type 3 (load control) serves as the baseline prerequisite layer, which is a non-negotiable bottom line for all typhoon preparedness strategies. Based on this framework and case reviews, we classified the anchorages of Huizhou Port into three categories (A, B, and C). Tier A anchorages integrate core advantages of Type 1 and Type 2 configurations. As the core typhoon-resistant anchorages, they shall be designated as critical port emergency resources, prioritized for immediate deployment upon typhoon alert issuance, with strict dynamic capacity control and safe spacing reservations to avoid overload. Tier B anchorages only partially match the characteristics of Type 1 or Type 2 configurations, with certain deficiencies in either inherent properties or management support. They are conditionally usable anchorages and shall only be deployed when the forecasted wind and wave levels are low. Tier C anchorages have significant deficiencies in inherent properties or management support, particularly critical scheduling flaws leading to high risks of vessel collision, anchor dragging, and secondary accidents. They will require strict load control and mandatory usage restrictions when necessary. Specific classification criteria are presented in

Table 10. Anchorage classification criteria.

Anchorage Category	Anchorage Characteristics	Management Strategy
A: Core Typhoon-Resistant Anchorage	Superior inherent properties and highly effective management support	Priority use with strict capacity control
B: Conditionally Usable Anchorage	Certain deficiencies exist in either inherent properties or management support	Enhanced management and conditional use
C: Restricted-Use Anchorage	Significant deficiencies in inherent properties or management support	Reasonable scheduling, planning for upgrades, and restricted use when necessary

.

Based on the aforementioned classification criteria and the performance of each anchorage under different typhoons, a dynamic classification of the anchorages was conducted. The results are presented in

Table 11. Dynamic classification of anchorages in Huizhou Port.

Anchorage Number	Typhoon Name and Type
	Doksuri (Heavy-Rainfall Type)	Saola (Strong-Wind Type)	Haikui (Complex-Track Type)
No. 1	A	B	B
No. 2	A	B	B
No. 3	A	A	B
No. 4	A	B	B
No. 5	B	C	C
No. 6	C	C	C
No. 7	C	B	B
No. 8	C	C	C
No. 9	C	C	C
No. 10	C	C	C
No. 11	C	C	C
No. 12	B	B	B

.

The analysis reveals that anchorage safety effectiveness varies significantly across typhoon types. For example, Typhoon Saola—characterized by rapid intensification and sudden wind shifts—exhibited a high-intensity, rapid-impact profile that severely tested the emergency management system’s agility. The swift influx of many vessels within a short timeframe, combined with suboptimal scheduling and allocation, led to the downgrade of some anchorages from Class A to Class B. Therefore, conducting a dynamic classification of anchorages based on specific typhoon types is essential. This approach enables the implementation of the most effective, targeted measures for different anchorage classes when confronted with varying typhoon scenarios, ultimately maximizing anchorage safety effectiveness.

4.3. Limitations and Generalizability

Based on the above configurational analysis, it is necessary to further discuss the limitations and generalizability of this study. This study is based on a single case—Huizhou Port—which possesses unique geographical and operational characteristics. Therefore, the specific anchorage capacity values and dynamic tiering results presented in this paper (e.g., Table 11) should not be directly extrapolated to open anchorages or other ports with significantly different sediment types and vessel compositions. However, the methodological framework constructed in this paper—the hybrid SEM–fsQCA model and dynamic tiered management process—exhibits good transferability. Provided that it is recalibrated based on local data (e.g., questionnaire surveys, measured anchorage characteristics, historical typhoon scenarios), this framework can be applied to other coastal ports. Furthermore, the three typhoon types identified in this study (heavy-rainfall, strong-wind, complex-track) represent the typical hazard spectrum faced by most coastal regions. Therefore, the configurational logics identified (e.g., ‘intrinsic-property-dominant’ and ‘management-support-dominant’) possess high conceptual generalizability across different ports. Future research will apply the same framework to multiple ports for comparative analysis to further test the cross-case generalizability of these configurational types and explore whether new configurational pathways exist.

5. Conclusions and Managerial Insights

This study establishes a hybrid SEM–fsQCA framework to address typhoon-period anchorage safety challenges in Guangdong Province, validated using 36 anchorage–typhoon coupled cases from typhoon-prone Huizhou Port. We find that anchorage safety effectiveness stems from synergistic multi-factor interactions with conjunctural causation and equifinality, rather than single-factor linear effects. SEM identifies meteorological and oceanographic conditions, inherent anchorage properties, and management support as the three core safety drivers, with the resultant weights subsequently employed to optimize anchorage capacity estimates. fsQCA further reveals three dominant configurational paths to high safety performance: intrinsic-property-dominant, management-support-dominant, and load-control-dominant types. High safety can be achieved via either superior infrastructure or robust operational management, while rational load control is a mandatory prerequisite for all safety pathways.

Grounded in these mechanistic insights, this study proposes a dynamic tiered anchorage management paradigm applicable to all coastal ports in Guangdong Province. Based on configurational pathway analysis, the paradigm dynamically classifies anchorages into three tiers according to their performance across different typhoon scenarios (A: core typhoon-resistant, B: conditionally usable, C: restricted-use). This enables a transition from static one-size-fits-all management to dynamic precision governance. A phased implementation roadmap is provided for Guangdong port management authorities: In the short-term (within 1 year), prioritize rational anchorage scheduling and strict load control via real-time vessel monitoring and dynamic capacity management systems, with mandatory enforcement of typhoon type-specific maximum capacity thresholds. In the medium term (1–3 years), enhance management support systems by developing an intelligent dispatch decision-support platform, which enables automated matching and simulation of typhoon warning information and anchorage allocation plans, to improve the precision of early warning-dispatch closed-loop management. In the long term (beyond three years), upgrade anchorage inherent properties through comprehensive structural safety assessments, engineering retrofits, and functional repositioning of anchorages with inadequate shelter or suboptimal seabed conditions, to fundamentally improve the overall typhoon resilience of Guangdong’s port anchorage systems.

This dynamic tiered anchorage management model enhances typhoon resilience at the operational level while also aligning with the strategic goals of port sustainable development. Specifically, through rational scheduling and load control, it reduces unnecessary vessel shifting and congestion, thereby lowering fuel consumption and greenhouse gas emissions (environmental sustainability); by improving anchorage safety effectiveness, it minimizes the risks of collisions, anchor dragging, and other accidents, safeguarding the lives of crew members and port communities (social sustainability); by avoiding anchorage overload and forced port closures, it reduces direct and indirect economic losses from supply chain disruptions (economic sustainability). These insights provide a reference for coastal ports seeking to coordinate safe operations with sustainable development goals in the context of climate change.