A Study on the Site Selection of Offshore Photovoltaics in the Northwest Pacific Coastal Waters Based on GIS and Fuzzy-AHP

Abstract

The scarcity of land resources has become a bottleneck restricting the development of photovoltaic (PV) energy, and it is imperative to expand PV layout into the ocean. However, existing studies lack a refined site selection framework for large-scale sea areas. This study takes the Northwest Pacific coastal waters as the research area and constructs a three-stage evaluation framework for the suitability of offshore PV site selection, which includes “resource potential–spatial exclusion–multi-criteria assessment”. The results show that the theoretical power generation potential is generally “higher in the south and lower in the north”, with some deviations in local areas due to differences in temperature and wind speed. Only 4.3% of the sea area is feasible for development. The high-suitability areas are concentrated in the southeast coast of Vietnam and the northwest side of Taiwan Island. The South China Sea has superior development potential, while the Bohai Sea and the Yellow Sea are relatively less suitable. This study generates the first offshore PV site selection map covering the research area, providing a scientific basis for the formulation of differentiated development strategies for regional offshore PV. It has important practical value for promoting the sustainable development of blue energy.

1. Introduction

In the era of global energy transition towards low-carbon and clean systems, the vigorous development of renewable energy has become a global consensus to combat climate change and ensure energy security [1]. Within this process, the large-scale and efficient exploitation of renewable energy serves as a critical pathway for many nations to drive energy transformation [2]. Solar energy, characterized by its limitless resources and widespread distribution, has emerged as a main force in reshaping the global energy structure. However, in economically developed coastal regions where land resources are scarce and ecological constraints are increasingly stringent, the development of traditional land-based photovoltaic (PV) power stations faces a severe “land bottleneck” [3]. Expanding PV systems into marine spaces—developing offshore PV—is not only an innovative direction for the sustainable development of the PV industry but also a strategic measure to optimize energy layout and foster the blue economy [4]. The Northwest Pacific coastal waters, possessing vast sea areas and extensive coastlines, along with superior nearshore solar resource endowments, offer immense potential for the large-scale development of OFPVs [5].

Despite the numerous advantages of OFPV power stations, their site selection is a complex spatial decision-making problem involving multiple disciplines and objectives, with complexity and challenges far exceeding those of land-based projects [6]. The core challenges stem from the distinctiveness of the marine environment and the intertwining of multiple spatial interests. Physical geography and resource conditions constitute the fundamental constraints, including solar radiation intensity, wind speed, water temperature, water depth, wave forces, tidal currents, and seabed geology. These factors directly determine the technical feasibility, structural safety, construction costs, and operational and maintenance complexity of the power station [7]. Ecological environment and spatial-planning constraints act as critical limitations; site selection must strictly avoid sensitive areas such as marine protected areas (MPAs), important fishery waters, shipping lanes, submarine cables, and military zones to minimize disturbance to marine ecosystems and avoid conflicts with existing marine activities [8]. Furthermore, socio-economic and techno-economic factors—such as proximity to grid connection points and power load centers, construction and transmission costs, and local policy support—collectively constitute the decisive factors for project economic feasibility [9]. Therefore, constructing a scientific, systematic, and transparent site selection methodology is a primary prerequisite for promoting the orderly development of the OFPV industry.

In recent years, the integration of Geographic Information Systems (GISs) and Multi-Criteria Decision-Making (MCDM) methods has become the mainstream paradigm for renewable energy project site selection and has been widely applied [8,10]. While this methodology is relatively mature in the offshore wind sector, related research in the field of OFPVs is gradually emerging. Existing studies typically assess suitability by constructing indicator systems that include dimensions such as resources, environment, and economy [3]. However, a systematic review of the literature reveals notable deficiencies in research specifically targeting the Northwest Pacific coastal waters. First, most research frameworks either focus on macroscopic resource assessment or are limited to specific case studies, failing to organically integrate the three key stages of resource-potential calculation, rigid constraint exclusion, and flexible criteria assessment into a coherent decision-making framework [11]. Second, regarding the construction of criteria systems, specific constraints unique to the Northwest Pacific—such as safety distances from oil and gas platforms, specific shipping lane regulations, and risks of extreme weather like typhoons—have not been considered with sufficient refinement and systematicity [12]. Third, in terms of assessment methods, handling uncertainty in the decision-making process (e.g., the subjectivity of criteria weights and data ambiguity) remains to be deepened. Classical methods like the FAHP allow for the integration of subjective judgments and expert opinions in such complex spatial decisions, providing a structured framework to prioritize factors influencing OFPV site selection and assess their relative importance [13].

To bridge these research gaps, this study aims to construct a comprehensive “resource potential–spatial exclusion–multi-criteria assessment” suitability evaluation framework for OFPV site selection, focusing on the Northwest Pacific coastal waters. The main content and innovations of this study are as follows. (1) Construction of a systematic three-stage decision framework: The first stage quantitatively assesses solar power generation potential based on data such as solar radiation, wind speed, and temperature; the second stage employs a set of explicit exclusion criteria to rapidly identify and eliminate areas completely unsuitable for development; and the third stage establishes an assessment system containing seven core criteria (e.g., distance to shore, wave conditions, and power generation potential), uses the FAHP method to determine weights, and generates the final suitability zoning map through GIS weighted overlay. (2) Integration of macroscopic screening and microscopic assessment: This framework overcomes the limitations of single methods, ensuring both the comprehensiveness of the assessment scope and the refined ranking of potential sites through progressive analysis, providing decision-makers with full-chain support from strategic planning to project siting. (3) Provision of an empirical case for the Northwest Pacific coastal waters: By applying the proposed framework, the first high-resolution OFPV site selection suitability map covering the Northwest Pacific coastal waters is generated. This identifies priority development zones and provides a scientific basis for differentiated development strategies across different sea areas, holding significant practical reference value for promoting the rational layout of the OFPV industry.

2. Materials and Methods

2.1. Study Area Overview

The study area comprises the coastal waters of the Northwest Pacific Ocean, geographically spanning from 104° E to 135° E and 3° N to 44° N, covering an area of approximately 8.4 million km² (Figure 1). This region encompasses the vast continental shelf of the Northwest Pacific, providing broad prospects for the construction and development of the offshore photovoltaic industry.

2.2. Data Sources and Preprocessing

The data used in this study for spatial exclusion analysis, such as marine protected areas, active faults, and ports, are the latest available data up to 2024. The key environmental driving data used in this study, including total solar radiation, wind speed, 10 m wind fields, and wave height, are all derived from the ERA5 reanalysis dataset released by the European Centre for Medium-Range Weather Forecasts (ECMWF) [14]. This dataset, which assimilates data from advanced numerical models and the global observation system, has a high spatial resolution and temporal continuity. It has become an important data source for global and regional climate characteristic analysis and renewable energy potential assessment, with its data accuracy verified in numerous studies. For this study, the multi-year average values from 2020 to 2024 were selected as the assessment baseline [14]. The aim is to simultaneously depict the current spatial pattern of resource and environmental conditions within a stable recent window of ≥5 years and to maintain temporal consistency with the fuzzy AHP weights based on current expert knowledge, thereby balancing the timeliness of macro-site screening and the scientific nature of the evaluation results. All data underwent standardization of geographic coordinate systems and spatial resolutions to ensure consistency in subsequent spatial analyses.

Table 1. List of data used in this study, their sources, and characteristics.

Data Category	Specific Parameters	Data Source	Original Resolution (Spatial/Temporal)	Application in This Study
Resource data	Total solar radiation, Wind speed (10 m), Sea surface temperature (SST)	ERA5 reanalysis dataset (ECMWF) [14]	0.25° × 0.25°, Hourly	Stage 1: Calculating solar power generation potential.
Exclusion criteria	Water depth (Bathymetry)	GEBCO	15 arc-seconds (~500 m)	Stage 2 (exclusion): Exclude areas with depth < 10 m and >50 m.
Exclusion criteria	Marine protected areas (MPAs)	European Environment Agency	Vector polygons	Stage 2: Exclude areas <1 km from MPAs.
Exclusion criteria	Active faults	National Earthquake Data Center	Vector lines	Stage 2: Exclude areas <2 km from active faults.
Exclusion criteria	Offshore oil and gas platforms	Global Energy Monitor	Vector points	Stage 2: Exclude areas <0.5 km from platforms.
Exclusion criteria	Airports	GitCode/open-source toolkit	Vector points	Stage 2: Exclude areas <15 km from airports.
Exclusion criteria	Major ports	OpenStreetMap (OSM)/Ministry of Transport data	Vector points	Stage 2: Exclude areas <3 km from ports.
Exclusion criteria	Coastline	GSHHG (NOAA)	Vector lines	Stage 2: Calculate distance to shore; exclude areas <10 km and >50 km.
Exclusion criteria	Power generation potential	Generated from Stage 1 resource data	0.25° × 0.25°	Stage 3: Used as a continuous evaluation factor.
Exclusion criteria	Water depth	GEBCO	15 arc-seconds (~500 m)	Stage 3: Used as a continuous evaluation factor.
Exclusion criteria	Wave parameters (Significant wave height)	ERA5 reanalysis dataset (ECMWF) [14]	0.25° × 0.25°	Stage 3: Used as “wave conditions” factor.
Exclusion criteria	Distance to shore	Calculated via GIS Euclidean distance from coastline	Raster (derived)	Stage 3: Used as “Distance to shore” factor.
Exclusion criteria	Distance to ports	Calculated via GIS Euclidean distance from ports	Raster (derived)	Stage 3: Used as “distance to ports” factor.
Exclusion criteria	Distance to airports	Calculated via GIS Euclidean distance from airports	Raster (derived)	Stage 3: Used as “distance to airports” factor.
Exclusion criteria	Distance to MPAs	Calculated via GIS Euclidean distance from MPAs	Raster (derived)	Stage 3: Used as “distance to MPAs” factor.

details the data used, their sources, and characteristics.

2.3. Methodology

This study adopts a three-stage decision framework integrating Geographic Information Systems (GISs) and the Fuzzy Analytic Hierarchy Process (FAHP) [13] to assess the site suitability of OFPV power stations in the study area. The logical flow of the framework includes (1) an assessment of solar power generation potential, (2) a spatial exclusion analysis based on constraint conditions, and (3) a suitability evaluation based on multi-criteria FAHP. The methods for each stage are detailed below.

2.3.1. Assessment of Solar Power Generation Potential

The purpose of this stage is to quantitatively assess the initial power generation potential of various locations within the study area from the perspective of natural resource endowment. A simplified yet practical physical model was employed for calculation. Annual power generation potential is a critical indicator for assessing OFPV stations, calculated as the average power potential multiplied by the number of hours in a year. The theoretical power generation potential per unit area of an OFPV station is calculated using Equation (1) [5]:

P_T = SPD × η,

(1)

where P_T represents the power output per unit area of the OFPV station, SPD is the Solar Power Density, and η represents the conversion efficiency of the OFPV modules.

The formula for Solar Power Density (SPD) is as follows [15]:

SPD = SSRD/(3600 × 1000),

(2)

where SSRD is the Surface Solar Radiation Downwards (J/m²). These values need to be normalized over a specified time period (in seconds). In this study, considering the 1 h temporal resolution, the SPD value in kilowatts per square meter (kW/m²) is equal to the SSRD value divided by 3600 s and then by 1000.

The conversion efficiency of the OFPV modules is calculated using Equation (3) [16]:

η = η_ref(1 − δ)(T_c − T_ref),

(3)

where η_ref is the electrical efficiency of the offshore solar PV system, set at 0.2 according to widespread usage; δ is the temperature coefficient of the PV module (approximately 0.004 °C⁻¹); and T_ref is the reference temperature (typically 25 °C) [17]. T_c represents the operating temperature of the PV module, calculated as follows [18]:

$$T_c = 2.0458 + 0.9458T_{\omega } + 0.0215 \times {\displaystyle \frac{\mathit{SSRD}}{3600}}-1.2376V_{\omega }$$

(4)

where T_ω and V_ω represent the ambient temperature and wind speed, respectively, which can be downloaded from the ERA5 dataset. This study assumes that the solar PV modules include a solar tracking system that dynamically adjusts the panel angle to maximize solar absorption; therefore, panel orientation was not considered. Additionally, other factors, including wind direction, soiling (e.g., bird droppings), waves, and distance to load centers, were excluded from the power output estimation [19,20]. While these factors may affect the actual power generation potential, they should be considered in future comprehensive project planning and assessments.

2.3.2. Spatial Exclusion Analysis

Based on international common practices and relevant studies, rigid exclusion criteria were established (

Table 2. Selection of standard thresholds for identifying exclusion areas.

Exclusion Criterion	Threshold (Buffer Range)	Basis for Determination
Marine protected areas	<1 km	Referring to the general environmental management practices in maritime engineering construction [21], a safety buffer zone is established to minimize the potential disturbance to the ecosystem of the protected area during installation and maintenance activities [3,22].
Active faults	<2 km	The distance of 2 km is a general measure to avoid potential risks to fixed or floating structures from geological disasters (such as tsunamis or seafloor instability caused by earthquakes) [23,24].
Oil and gas platforms	<0.5 km	The threshold of 0.5 km is to avoid conflicts with the operational areas of offshore oil and gas facilities (such as safety zones and helicopter landing areas), and it complies with the standard safety distances for offshore facilities [1,13].
Major ports	<3 km	The threshold of 3 km is intended to avoid spatial conflicts with the busy shipping lanes in and out of ports, ensuring the safety of navigation and the normal operation of port activities. This distance refers to the typical range of port safety operation zones and high-density navigation areas [5,24].
Airports	<15 km	Based on the regulations of the International Civil Aviation Organization and national airspace management provisions, large-scale reflection from photovoltaic systems may cause visual interference to pilots. The threshold of 15 km refers to the general safety distance recommendations for similar renewable energy projects (especially large-scale photovoltaic and solar thermal power stations) and complies with strict aviation safety standards [20].
Water depth	<10 m or >50 m	The lower limit of 10 m is intended to avoid ecologically sensitive intertidal zones, seagrass beds, and coral reef areas; the upper limit of 50 m is based on the economic threshold of the anchoring systems and mooring technologies of mainstream floating photovoltaic platforms. Construction and maintenance costs will significantly increase beyond this depth [23,25].
Distance to shore	<10 km or >50 km	The lower limit of 10 km is designed to reduce impacts on coastal scenery, fishing activities, and tourism. The upper limit of 50 km is based on an analysis of the balance between submarine cable transmission losses and costs [1,5,25].

). Any area violating one of these criteria is deemed unsuitable, thereby macroscopically eliminating areas completely unfeasible for development. Buffer zones for the exclusion criteria were generated in ArcGIS 10.5. By overlaying these buffers, the unsuitable areas in the study region that violate the exclusion criteria were identified.

2.3.3. FAHP Weight Determination and Comprehensive Suitability Evaluation

Basis for the Selection of Evaluation Indicators

The construction of the evaluation indicator system in this study adheres to the principles of “systematicness, scientificity, and operability”, with the objective of encompassing the three core dimensions that influence the site selection of offshore floating photovoltaic (OFPV) projects: technical feasibility, environmental compatibility, and economic rationality. The selection of indicators is primarily grounded in the following two aspects:

(1)

Literature Review and Best Practices

Through a systematic review of the existing literature on offshore wind power, onshore photovoltaics [1], and emerging offshore photovoltaic site selection (e.g., Reference [3]), key constraints and evaluation factors that are widely recognized were identified. These include resource potential [5], water depth [6], and wave conditions [1]. These factors have been extensively validated in both theoretical and practical contexts, providing a robust foundation for the selection of indicators in this study.

(2)

Characteristics of the Northwest Pacific Region

The general indicators were localized to reflect the specific characteristics of the study area. The Northwest Pacific Region is characterized by active geological structures, dense shipping traffic, and frequent typhoons. Consequently, indicators such as “distance to major ports/routes” [6] were specifically included to capture the unique risks and constraints of the area. This localization ensures that the evaluation system is tailored to the specific environmental and operational challenges of the region.

The seven core indicators finally determined can be divided into three categories:

Resource and technology (driving factors): Power generation potential (C1), water depth (C2), and wave conditions (C5) are the physical driving factors that determine the basic performance and engineering feasibility of the project.

Environment and safety (constraint factors): Distance to marine protected areas (C3), distance to airports (C6), and distance to active faults (already addressed in the exclusion stage) are policy red lines and safety baselines that the project must comply with to ensure ecological compatibility and operational safety.

Economy, operation, and maintenance (cost factors): Distance to shore (C4) and distance to ports (C7) directly affect the construction costs, grid connection, and economic viability of subsequent operation and maintenance.

Evaluation Criteria System

(1)

Power Generation Potential

Power generation potential is the core indicator measuring the economic feasibility of an OFPV project, directly determining energy output and return on investment. While primarily influenced by total solar irradiance, factors such as operating temperature, salt spray attenuation, and micro-tilting of floating bodies caused by waves in the marine environment also significantly affect final efficiency [26,27]. Therefore, this indicator comprehensively reflects the solar resource endowment and actual generation potential of a specific sea area and is typically assigned the highest priority in decision-making [28].

(2)

Wave Conditions

Wave conditions are decisive environmental factors affecting the structural safety and system reliability of OFPV floating bodies [26]. Continuous wave loading causes fatigue damage to structures and poses severe challenges to mooring system stability; extreme waves can lead to catastrophic failure. Additionally, wave-induced rocking and tilting reduce the optimal light-receiving angle, thereby lowering efficiency [29]. Thus, selecting areas with mild wave conditions is key to ensuring engineering safety and controlling maintenance costs.

(3)

Water Depth

Water depth is a fundamental techno-economic indicator determining the technical scheme and foundation cost [25,30]. Different depth ranges correspond to different foundation structures, ranging from pile-fixed types in shallow waters to floating types in deep waters, with vast differences in technical complexity and cost [22]. Generally, increased depth leads to a sharp rise in material, installation, and investment costs.

(4)

Distance to Shore

Distance to shore is a critical logistics indicator affecting construction and O&M economics [1,25]. Greater distances imply increased submarine cable lengths and transmission losses, as well as higher time and financial costs for daily O&M and personnel transport [24]. Although offshore areas may offer better resources, the surging grid connection costs weaken economic attractiveness. A balance must be sought to keep the distance within an economically reasonable range [13].

(5)

Distance to Marine Protected Areas

This indicator assesses ecological compatibility. Maintaining sufficient buffer distances from sensitive areas like nature reserves effectively avoids disturbance to rare species habitats and migration channel [30]. This is essential for complying with environmental regulations, enhancing social acceptance, and fulfilling corporate social responsibility [22,24].

(6)

Distance to Ports

Ports serve as core logistical bases for construction, installation, and O&M. Proximity to suitable ports directly relates to the speed and cost of equipment transport, vessel scheduling, and supply replenishment [24]. A nearby, well-equipped port significantly improves construction efficiency and ensures rapid response during emergencies.

(7)

Distance to Airports

This indicator is primarily for aviation safety. Large-scale PV panels may produce glare at specific angles [24], posing potential risks to aircraft during takeoff and landing. Furthermore, PV arrays may interfere with airport radar signals [11]. Therefore, sites must avoid airport clearance protection zones to ensure aviation safety.

The “Technology–Environment–Economy” seven-dimensional evaluation index system constructed in this study closely follows the macro-strategic assessment objectives and achieves the necessary balance between systematicness, relevance, and operability. However, it needs to be critically examined in terms of its completeness, measurement methods, and application boundaries: although the framework covers the core constraints of resources, technology, and environment, it does not yet include other external dimensions that are equally critical to project feasibility, such as grid connection, regional policies, social acceptance, and risks of extreme events like typhoons [1]. Moreover, the measurement of indicators has been reasonably simplified (for example, wave conditions are measured by average significant wave height instead of extreme values, and distance calculations use Euclidean distance) [7,10]. These decisions are mainly due to the reality of data availability at the regional scale and the difficulty of establishing a universal model. This constitutes certain limitations, which can be further deepened in the feasibility studies at the specific project scale [3,27].

Determination of Criteria Weights Based on FAHP

To scientifically quantify the relative importance of each criterion, this study employs the FAHP method. FAHP is a systematic Multi-Criteria Decision-Making method that decomposes complex problems into hierarchies and converts subjective judgments into objective weights using fuzzy numbers to handle uncertainty [13]. The decision process is divided into three levels: goal, criteria, and indicators.

Seven evaluation criteria were used. Pairwise comparisons were conducted using a scale of 1 to 9 (

Table 3. Fundamental scale for pairwise comparisons.

AHP Normal Scale	AHP Reciprocal Scale	Definition	FAHP TFN Scale	Explanation
1	1	Equal importance	(1, 1, 1)	Criteria are equally important.
3	1/3	Moderate importance	(2, 3, 4)	One criterion is slightly more important.
5	1/5	Strong importance	(4, 5, 6)	One criterion is strongly more important.
7	1/7	Very strong importance	(6, 7, 8)	One criterion is very strongly more important.
9	1/9	Extreme importance	(9, 9, 9)	One criterion is extremely more important.
2, 4, 6, 8	1/2, 1/4, 1/6, 1/8	Intermediate values	(1,2,3); (3,4,5)…	Used to compromise between two judgments.

), where 1 represents equal importance, and 9 represents the highest importance [13]. The linguistic terms are converted into Triangular Fuzzy Numbers (TFNs). Each TFN consists of a triplet (l, m, and u), representing the minimum, most likely, and maximum possible values, respectively [31,32]. The triplet structure is shown in Figure 2.

First, experts provided assessment values to create the general fuzzy comparison matrix, as shown in Equation (5):

$$\stackrel{\mathrm{~}}{A }={\left({\stackrel{\mathrm{~}}{a}}_{\mathit{ij}}\right)}_{n\times n}=\left(\begin{array}{ccc}(1,1,1)& \dots & (l_{1n},m_{1n},u_{1n})\\ ⋮& \ddots & ⋮\\ (l_{n1},m_{n1},u_{n1})& \dots & (1,1,1)\end{array}\right)$$

(5)

Consistency checks were performed using the consistency ratio (CR). A CR < 0.10% indicates satisfactory consistency [32]. The principal eigenvalue (${\lambda }_{\max }$) and Consistency Index (CI) were calculated, followed by the CR using the Random Index (RI):

$$\mathit{CR}={\displaystyle \frac{\mathit{CI}}{\mathit{RI}}}$$

(6)

$$\mathit{CI}={\displaystyle \frac{{\lambda }_{\mathit{max}}-n}{n-1}}$$

(7)

Calculated based on the Analytic Hierarchy Process (AHP) method, and after passing the consistency test (CR < 0.1), the Fuzzy Analytic Hierarchy Process (FAHP) was adopted for weight calculation. FAHP extends the traditional AHP by integrating fuzzy set theory to mitigate the uncertainty associated with human judgments in the decision-making process. Converting crisp values into fuzzy judgments enables FAHP to reduce ambiguity in criterion evaluation and improve decision-making.

Subsequently, the fuzzy comparison matrix was transformed into a crisp comparison matrix using the centroid defuzzification method [33].

$${\mathit{a}}_{\mathit{ij}}({\stackrel{\mathrm{~}}{\mathit{a}}}_{\mathit{ij}})={\displaystyle \frac{{\mathit{l}}_{\mathit{ij}}+{\mathit{m}}_{\mathit{ij}}+{\mathit{u}}_{\mathit{ij}}}{3}},\forall \mathit{i},\mathrm{j}\in [1,\mathit{n}]$$

(8)

Fuzzy weights were then derived. The fuzzy geometric mean value is calculated as follows:

$$\stackrel{\mathrm{~}}{\mathit{R}{\mathit{S}}_{\mathit{i}}}={\sum }_{\mathit{j}=1}^{\mathit{n}}{\stackrel{\mathrm{~}}{\mathit{a}}}_{\mathit{ij}}=({\sum }_{\mathit{j}=1}^{\mathit{n}}{\stackrel{\mathrm{~}}{\mathit{l}}}_{\mathit{ij}},{\sum }_{\mathit{j}=1}^{\mathit{n}}{\stackrel{\mathrm{~}}{\mathit{m}}}_{\mathit{ij}},{\sum }_{\mathit{j}=1}^{\mathit{n}}{\stackrel{\mathrm{~}}{\mathit{u}}}_{\mathit{ij}}),\mathit{i}\in [1,\mathit{n}]$$

(9)

$$\begin{gathered} \widetilde{{\mathit{S}}_{\mathit{i}}}={\displaystyle \frac{\stackrel{\mathrm{~}}{\mathit{R}{\mathit{S}}_{\mathit{i}}}}{{\sum }_{\mathit{j}=1}^{\mathit{n}}\stackrel{\mathrm{~}}{\mathit{R}{\mathit{S}}_{\mathit{j}}}}}=({\displaystyle \frac{{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{l}}_{\mathit{ij}}}{{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{l}}_{\mathit{ij}}+{\sum }_{\mathit{k}=1,\mathit{k}\ne \mathit{i}}^{\mathit{n}}{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{u}}_{\mathit{ij}}}},{\displaystyle \frac{{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{m}}_{\mathit{ij}}}{{\sum }_{\mathit{k}=1}^{\mathit{n}}{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{m}}_{\mathit{ij}}}},{\displaystyle \frac{{\sum }_{\mathit{j}=1}^{\mathit{n}}{\stackrel{\mathrm{~}}{\mathit{u}}}_{\mathit{ij}}}{{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{u}}_{\mathit{ij}}+{\sum }_{\mathit{k}=1,\mathit{k}\ne \mathit{i}}^{\mathit{n}}{\sum }_{\mathit{j}=1}^{\mathit{n}}{\mathit{l}}_{\mathit{ij}}}}) \\ =({\mathit{l}}_{\mathit{i}},{\mathit{m}}_{\mathit{i}},{\mathit{u}}_{\mathit{i}}),\mathit{i}\in [1,\mathit{n}] \end{gathered}$$

(10)

Afterwards, the local weights (ω_i) are obtained through the conversion of fuzzy weights as follows:

$${\omega }_i=S_i(\widetilde{S_i}) ={\displaystyle \frac{l_i+m_i+u_i}{3}},i\in [1,n]$$

(11)

However, the sum of these weights may not be exactly equal to 1; therefore, the weights need to be normalized. The normalized weight vector (W) is as follows:

$$W= ({\omega }_1,{\omega }_2,{\omega }_3,{\omega }_i),i\in [1,n]$$

(12)

Finally, the global evaluation criterion weights can be obtained by multiplying the local weights of the criterion layer with the weights of the corresponding indicator layer (

Table 4. Results of evaluation criteria weights for OFPV site selection.

Criterion	Symbol	Final Weight	Rank	Type
Power potential	C1	0.3498	1	Core Driving Factor
Water depth	C2	0.1880	2	Key Technical Factor
Distance to MPAs	C3	0.1430	3	Compliance Factor
Distance to shore	C4	0.0853	4	Cost-Sensitive Factor
Wave height	C5	0.0832	5	Risk-Sensitive Factor
Distance to airports	C6	0.0782	6	Safety Factor
Distance to ports	C7	0.0724	7	Cost Factor

). The global weight of indicators at the highest level is equal to their local weight [34].

To ensure the professionalism and effectiveness of weight acquisition, this study invited five experts with over ten years of front-line work experience or senior research titles in marine engineering, renewable energy planning, and marine policy to participate in the consultation. The selection of experts followed the principles of relevance, authority, and industry representation to ensure their in-depth understanding and professional judgment of the technical, environmental, and economic dimensions of offshore floating photovoltaic (OFPV) site selection.

Data Processing and Quality Control Steps:

Consistency ratio test: Each expert’s pairwise comparison matrix was individually subjected to a rigorous consistency ratio test to ensure the logical consistency of their judgments. Upon calculation, all experts’ CR values were found to be less than 0.10, meeting the acceptable standard proposed by Saaty [35]. This indicates that the experts’ judgments are internally consistent and do not generate contradictory priority assignments.

Inter-expert variability analysis: To measure the consistency of the expert group’s opinions, this study calculated the Spearman rank correlation coefficient between individual expert weights and the final aggregated weights (geometric mean). The preliminary analysis revealed a high positive correlation (Spearman’s ρ > 0.85) in the experts’ perception of the importance ranking of each criterion. This suggests that, despite minor differences in numerical judgments, the experts reached a high consensus on the relative importance of core driving factors (such as power generation potential C1) and secondary constraint factors (such as distance to ports C7), enhancing the robustness of the group weights.

Handling and transparency of aggregated opinions: The final fuzzy pairwise comparison matrix was aggregated using the geometric mean method from the fuzzy judgment matrices of all five experts. This method is less sensitive to extreme values and better represents group consensus [31]. To enhance research transparency, the anonymous original judgment matrices of the experts will be provided in the Supplementary Materials for peer review.

Sensitivity analysis: The weights of the two highest criteria, “power generation potential” (0.3498) and “water depth” (0.1880), were adjusted by ±10%, and the weights of the other criteria were re-normalized. The MCDM-GIS analysis was then re-performed. The results showed that the spatial distribution and classification of the key areas remained highly stable, with the area change rate of the highly suitable zones being less than 5%. This indicates that the site selection results are highly robust to minor changes in weights as long as the relative importance order of the core driving and constraining factors remains unchanged.

The final suitability score calculation involves three steps: (1) threshold classification of raw data based on expert opinions and the literature [36]; (2) scoring of different intervals on a 0–100 scale to form reclassified raster layers (

Table 5. Valuation scoring of evaluation criteria.

Criterion	Highly Suitable (90–100)	Suitable (80–90)	Moderately Suitable (70–80)	Low Suitability (60–70)	Unsuitable (<60)
Distance to shore (km)	10–20	20–30	30–40	40–50	<10, >50
Waves (m)	0–1	1–1.5	1.5–2	2–2.5	>2.5
Power potential (kwh/m2)	>350	300–350	250–300	200–250	<200
Distance to marine protected areas (km)	>15	10–15	5–10	1–5	<1
Distance to airports (km)	>30	25–30	20–25	15–20	<15
Distance to ports (km)	20–35	15–20; 35–40	10–15; 40–45	3–10; 45–50	<3, >50
Water depth (m)	10–20	20–30	30–40	40–50	<10, >50

); and (3) weighted overlay analysis in ArcGIS 10.5 to generate the final suitability map.

The rationale for determining the scoring thresholds of each indicator in Table 5 is as follows:

Power generation potential, water depth, and distance to shore: The thresholds mainly refer to the commonly used ranges in the technical and economic studies of offshore wind power and similar marine energy projects [30,37]. For example, the water depth range of 10–50 m balances ecological sensitivity (shallow water areas) with engineering costs (deep water areas) [6], and the distance to shore range of 10–50 km balances the ecological impact near the shore with transmission costs [1].

Distance to marine protected areas, airports, and ports: The buffer-zone thresholds are based on international practices, relevant management regulations (such as regulations on aviation obstacles and port water safety management), and suggested safety distances in the literature. For example, the 15 km buffer around airports refers to the requirements for avoiding glare from photovoltaic panels and potential radar interference in terms of aviation safety.

Wave conditions: The thresholds are set according to the engineering standards for floating structures (including floating photovoltaics) and research on tolerable sea conditions [32,35]. For example, areas with wave heights > 2.5 m are rated as “unsuitable”, considering that typical floating photovoltaic structures face significant stability and fatigue risks under moderate-to-severe sea conditions.

The scoring adopts a linear assignment from 0 to 100, with an arithmetic division according to the suitability degree within different threshold ranges, to achieve continuous and differentiated spatial expression in the GIS weighted overlay.

3. Results

3.1. Distribution of Photovoltaic Power Generation Potential

Based on satellite remote-sensing and meteorological data from 2020 to 2024, the annual total solar radiation per unit area and the theoretical annual photovoltaic power generation potential per unit area (unit: kWh/m²) for the study waters were calculated. The results show that the solar energy resource distribution in the study area exhibits a distinct macroscopic pattern of “high in the south and low in the north”.

In terms of spatial distribution, total solar radiation presents a core area of high values (>1600 kWh/m²) in the Northern South China Sea, decreasing towards the northern waters, with the northern part of the study area being a relatively low-value zone (<1400 kWh/m²; Figure 3). The spatial pattern of theoretical annual power generation potential is roughly consistent with that of total solar radiation, but discernible regional differences exist (Figure 4). Specifically, the advantage of theoretical power generation potential per unit area in parts of the southern high-radiation zone is relatively weakened, while in parts of the northern waters (such as the Bohai Sea), the theoretical potential performance is superior to what the radiation resource levels might suggest.

Temporally, the annual total solar radiation remained overall stable during the study period (2020–2024), with minor inter-annual fluctuations. Analysis reveals that while the trend of theoretical annual power generation potential is generally stable, it is not entirely synchronous with changes in total solar radiation, highlighting the modulating effect of other meteorological factors. Taking 2023 as a typical example, although total solar radiation decreased to some extent, the occurrence of more favorable temperature and wind speed conditions—specifically lower average temperatures and higher wind speeds—effectively enhanced the cooling effect on PV modules, ultimately maintaining an upwards trend in the theoretical annual power generation potential per unit area.

3.2. Spatial Exclusion Results

Following the systematic exclusion based on the seven criteria described above (Figure 5), approximately 95.7% of the initial study area was designated as restricted zones, leaving the remaining 4.3% (approximately 358,000 km²) identified as “feasible sea areas” to proceed to the suitability evaluation stage. The results indicate that “water depth” and “distance to shore” are the two most critical exclusion constraints, fundamentally defining the potential geographic scope for OFPV development from technical and economic feasibility perspectives. The remaining feasible areas appear primarily as strip-like zones located approximately 10 to 50 km from the coastline, with water depths between 10 and 50 m (Figure 6).

3.3. Comprehensive Suitability Evaluation Results

The highest suitability values appear along the southeast coast of Vietnam and the waters northwest of Taiwan Island; these regions combine optimal solar resources, favorable marine conditions, and moderate environmental constraints. The second-highest value regions are widely distributed along the coast of Malaysia and the periphery of the South China Sea, demonstrating the superior overall development potential of the South China Sea. Moderately suitable areas are widely distributed across most of the East China Sea and parts of the Yellow Sea. Although the solar resource endowment and marine conditions in these areas are inferior to those in the South China Sea, they avoid major ecological red lines and shipping lane conflicts, possessing feasible development conditions. Areas with poor suitability are mainly concentrated in the Bohai Sea and Yellow Sea. These regions are primarily limited by relatively low solar radiation intensity and a higher density of marine protected areas, resulting in lower comprehensive suitability scores (Figure 7).

Based on the calculated OFPV site selection scores, the Reclassify tool in ArcGIS 10.5 was used to categorize the scores into five levels: Very Unsuitable (Cyan), Unsuitable (Green), Suitable (Yellow), Relatively Suitable (Orange), and Highly Suitable (Red). According to ArcGIS 10.5 calculations, the areas covered by the five levels are 9.28 km², 550.95 km², 79,529.55 km², 208,586.99 km², and 36,353.35 km², respectively. The Very Unsuitable areas are mainly located in the waters northwest of Kyushu Island, Japan; the Unsuitable level is primarily found in the northwest region of Kyushu Island, parts of the Northern Bohai Sea, and parts of the East China Sea; and the Highly Suitable areas mainly appear in the southeast coastal waters of Vietnam, the waters northwest of Taiwan Island, and the northwest coastal waters of Malaysia. Figure 8 presents the site selection suitability classification map (Figure 8).

4. Discussion

4.1. Spatial Patterns of Site Selection Results and Underlying Driving Mechanisms

This study confirms that the macro-distribution of solar energy resources in the study area exhibits a pronounced latitudinal zonality with a “high in the south and low in the north” pattern. The Northern South China Sea, which is closer to the equator, has an annual theoretical power generation potential approaching 400 kWh/m², which is highly consistent with the global consensus that “equatorial and tropical waters are the golden zones for offshore photovoltaic development” [7,38]. However, the detailed assessment of this study further reveals that resource endowment is not the sole determining factor. For example, mid-latitude waters such as the East China Sea and the Yellow Sea, although with lower annual radiation, have higher resource stability and relatively mild marine environments (e.g., fewer typhoons and lower wave heights), and thus also possess significant development value [6,22]. More notably, this study finds, through multi-year data and the relevant literature analysis, that the solar radiation in some areas of the Bohai Sea and the Yellow Sea shows a slight increasing trend, while the Northern South China Sea shows a slight decreasing trend [5]. This highlights the need to incorporate the potential impacts of climate change into the assessment of resource reliability in the full life-cycle project evaluation [22]. Therefore, a scientific assessment of power generation potential should not only focus on resource intensity but also comprehensively consider its stability and long-term evolution trends.

The comprehensive suitability assessment results delineated a priority zoning map for the development of offshore photovoltaic power stations within the study area. The results clearly indicate that resource endowment and final suitability are not entirely aligned. For example, although the Bohai Sea has relatively low solar radiation resources, some of its sheltered waters (with better wind and wave conditions) and the advantage of being close to the Bohai Economic Rim load center in China still show certain development potential after a comprehensive trade-off. This profoundly reveals that the final site suitability is the result of the interplay and trade-offs among, multiple factors, including resource conditions, geographical constraints (water depth and distance to shore), and socio-economic factors (distance to ports and airports) [1]. This kind of regional priority ranking based on the suitability map provides investors and decision-makers with scientific grounds, ranging from “macro guidance” to “micro site selection”.

4.2. Comparison with Existing Studies and Facilities, and Framework Validation

This study, by introducing a three-stage framework of “theoretical potential–spatial exclusion–multi-criteria evaluation”, not only confirms the macro conclusion consistent with Wen and Lin [5] that the South China Sea has superior overall development potential compared to the Bohai Sea, but also further identifies micro hotspots, such as the southeast coast of Vietnam and the west side of the Taiwan Strait. Additionally, it quantifies the restrictive effects of factors like port accessibility and typhoon and wave risks on local areas of the Bohai Sea, providing refined spatial details and decision-making dimensions for macro judgments.

Although large-scale and commercial offshore photovoltaic projects in the region are still in the initial stage, the distribution of existing or under-construction demonstration facilities generally aligns with the suitability pattern revealed by this study [39]. For example, pilot projects along the southeast coast of Vietnam, near the northern coast of the South China Sea, and within the Bohai Bay are all located in “highly suitable” or “moderately suitable” zones, especially in the nearshore belt, with water depths of 10–50 m and distances to shore of 10–50 km. This offers preliminary empirical support for the validity of this study’s framework [4].

In response to the limitations of previous offshore energy site selection methods based on GIS and MCDM [28], such as weak systematicness and lack of specificity, this study proposes a three-stage GIS-FAHP framework. Through the serial process of “theoretical potential → rigid exclusion → flexible decision-making”, it systematically addresses the separation of “absolutely infeasible” and “relatively more optimal” decisions on a regional scale. The framework innovatively reconstructs the criterion system centered on “power generation potential” and refines constraints such as “distance to ports” and “typhoon risk zones” according to the characteristics of offshore photovoltaics. This achieves adaptive migration and localization innovation of the wind power site selection paradigm [1], significantly enhancing the systematicness, adaptability, and refinement of site selection.

4.3. Robustness, Uncertainty, and Limitations of the Methodology

This study heavily relies on expert judgment based on FAHP, which, while effectively dealing with the fuzziness of subjective judgment, also brings inherent uncertainty associated with the expert sample. To enhance robustness under the small sample of expert judgments, this study has taken several measures: (1) strictly conducting individual consistency checks (CR < 0.1) to ensure the internal logical consistency of each expert’s judgment [40]; (2) publishing the original expert scores as Supplementary Materials to enhance process transparency (Table S1); and (3) introducing sensitivity analysis (for example, conducting ±10% fluctuation tests on the core criterion weights), the results of which show that the spatial distribution pattern of highly suitable areas has strong robustness, with no significant changes in the main hotspot regions. These measures partially address the uncertainty challenges of small sample aggregation and draw on practices of using FAHP to handle similar situations in related field [31,32].

However, the limitations of the study must be clearly acknowledged. First, although FAHP and sensitivity analysis provide robustness assurance, the weight system essentially still reflects the knowledge and preferences of the specific group of experts invited in this study. When directly transferring the conclusions to other sea areas with different socio-economic development stages or policy environments (such as the North Sea in Europe), the criteria and weights need to be recalibrated. Second, the threshold settings in the spatial exclusion stage (such as water depth of 50 m, and 10 km from airports) are based on literature reviews and existing engineering practices, but with the rapid evolution of floating photovoltaic technology (such as new structures adapted to deeper waters and higher wave conditions [10]) and possible adjustments to future airspace management policies, these thresholds need to be dynamically updated. Third, some datasets (such as marine functional zoning and shipping route data) have limitations in terms of temporal representativeness and spatial resolution. In the highly dynamic marine environment, this may introduce spatiotemporal uncertainty into the evaluation results.