A method for constructing and evaluating environmental impact causal networks based on a SPER framework

Abstract

The application discloses a kind of environmental influence causal network construction and evaluation method based on SPER framework.Method includes: constructing SPER four-layer framework, including: source layer, transmission layer, exposure layer, response layer;With directional link characterizes the causal relationship between each node, forms four-layer causal network;Assign seven attributes to each directional link;From each source node, along directional link traversal, according to the direction of attribute, complete causal path from source node to terminal effect node through transmission node and exposure node is constructed;With path minimum risk value quantifies the overall risk level of each path;S4, based on the frequency of occurrence of each node in complete causal path, position distribution and the risk level of front and rear link, identify three kinds of key nodes.The application aims to build four-layer causal network structure, realize the semi-quantitative system evaluation of the cumulative effect of multiple stressors under the condition of data scarcity, and identify the nodes and links in the causal transmission chain, to provide a scientific basis for differentiated environmental management.

Description

Technical Field

[0001] This invention belongs to the field of environmental assessment technology, specifically relating to a method for constructing and assessing environmental impact causal networks based on the SPER (Source-Pathway-Exposure-Response) framework. Background Technology

[0002] Currently, the closest existing technical solutions for environmental impact assessment of deep-sea mining activities mainly include inventory method, matrix method, scenario analysis method, spatial analysis method, and life cycle assessment.

[0003] The inventory method creates an assessment list by listing the potential environmental impacts of mining activities. The matrix method builds upon the inventory method by introducing an impact severity rating, presenting the interaction between the activity and various environmental factors in a matrix format. Scenario analysis involves setting different mining activity scenarios (such as mining intensity, technical solutions, mitigation measures, etc.) and combining numerical simulations or empirical models to predict the scope and extent of environmental impact under each scenario. Spatial analysis utilizes Geographic Information Systems (GIS) to spatially overlay and buffer zone analyze the spatial distribution of mining activities with basic layers such as biological resource distribution maps and sediment type maps to assess the spatial impact range of mining activities. Life cycle assessment systematically evaluates the environmental impact of a product or activity by tracking its environmental impact throughout the entire process from raw material acquisition to final disposal. Mature methodologies exist for assessing the environmental footprint of industrial products, and some studies have attempted to apply this method to the environmental impact assessment of deep-sea mineral development.

[0004] However, the aforementioned existing technologies have the following significant shortcomings when applied to environmental impact assessments for deep-sea mining: First, there is a lack of ability to construct complete causal chains, resulting in fragmented assessments. Inventory and matrix methods are essentially static, two-dimensional relational descriptive tools, capable of reflecting only a one-to-one relationship between a single activity and a single environmental factor, failing to characterize the interactions between environmental factors and the cumulative effects of multiple pressure sources. While scenario analysis can predict the scope and extent of impact, it typically assumes that each pressure source acts independently, making it difficult to handle the interactive effects of multiple pressure sources. Spatial analysis is essentially a static snapshot-style assessment, unable to reflect the temporal evolution and nonlinear dynamics of impacts. Although life-cycle assessment can track the entire process of impact, the complexity of deep-sea ecosystems far exceeds that of industrial systems, making it difficult to construct complete ecological causal chains.

[0005] Secondly, there is a lack of ability to assess the cumulative effects of multiple pressure sources coupled together. Deep-sea mining generates multiple pressure sources simultaneously, including sediment plumes, hard matrix removal, underwater noise, artificial light, heavy metal release, and dissolved inorganic carbon release. These pressure sources have complex interaction effects. Existing methods typically assess each pressure source independently and then simply superimpose them, failing to address the synergistic and antagonistic effects of multiple pressure sources and severely underestimating the cumulative effect.

[0006] Third, there is a lack of effective systematic assessment methods under conditions of data scarcity. Spatial analysis methods and life-cycle assessments require high-resolution baseline data, but ecological baseline data in deep-sea areas are extremely scarce, making it difficult to meet the implementation conditions of these methods. While inventory methods and expert-based methods can be implemented under low-data conditions, the assessment results are highly subjective and have poor repeatability, failing to provide sufficient scientific rigor and systematicity.

[0007] Fourth, there is a lack of analytical capabilities to identify key transmission paths and bottlenecks. Even if some methods can identify certain impact paths, they cannot systematically assess the relative importance of different paths, let alone identify bottlenecks that limit the transmission of impacts, thus making it difficult to provide differentiated mitigation measures for management decisions. Summary of the Invention

[0008] To address the shortcomings of existing technologies, such as the lack of ability to construct complete causal chains, the lack of ability to assess the cumulative effects of multiple pressure sources, the difficulty in conducting systematic assessments under data-scarce conditions, and the lack of ability to identify critical paths and bottlenecks, this invention provides a method for constructing and assessing environmental impact causal networks based on the SPER framework. This method aims to achieve a semi-quantitative systematic assessment of the cumulative effects of multiple pressure sources under data-scarce conditions by constructing a four-layer causal network structure of pressure source-transmission process-exposure pathway-response, and to identify key nodes and bottlenecks in the causal transmission chain, thus providing a scientific basis for differentiated environmental management.

[0009] According to one aspect of the present invention, a method for constructing and evaluating environmental impact causal networks based on the SPER framework is provided, the method comprising: A1. Construct a four-layer SPER framework, including: a source layer, identifying multiple types of environmental stressors as source nodes; a transmission layer, identifying multiple transmission processes of the stressors in the environmental medium as transmission nodes; an exposure layer, identifying multiple exposure pathways of organisms to the stressors or their effects as exposure nodes; and a response layer, identifying multiple effect nodes exhibited by organisms from the individual to the ecosystem level after exposure as effect nodes; and use directed links to represent the causal relationships between the nodes, forming a four-layer causal network. A2. Assign seven attributes to each directed link: importance of the parent node. Characterizes the spatial and temporal importance of the influence of the cause on the effect; child node sensitivity. Characterizes the sensitivity of the outcome to changes in the cause; the probability of causal event occurrence P; and the overall strength of the effect. Risk level Level of evidence The level of uncertainty characterizes the sufficiency of scientific evidence. Characterizes the main types of sources of uncertainty; A3. Starting from each source node, traverse along the directed links, constructing a complete causal path from the source node through the transit and exposure nodes to the terminal effect node in the direction of source node → transit node → exposure node → effect node; select the path with the minimum risk value. Quantify the overall risk level of each path, among which, For the first on the path The node and the first Risk level of links between nodes; A4. Based on the frequency of occurrence, location distribution, and risk level of preceding and following links of each node in the complete causal path, three types of key nodes are identified: common skeleton nodes, which are nodes that appear repeatedly in multiple paths from different source nodes and are located in the cascade segment of effect nodes; source-specific front-end entry nodes, which are nodes that appear only as initial effect nodes in specific source node paths; and bridge / terminal amplifier nodes, which are nodes that receive low-level response inputs and output high-level responses with output risk higher than input risk.

[0010] The above technical solution provides an evaluation method based on the SPER framework, which forms a complete process for causal network construction and evaluation by constructing a four-layer causal network, quantifying links with multiple attributes, traversing paths and identifying key nodes.

[0011] Compared to static tools like checklists or matrix methods that only describe the relationship between a single activity and a single factor, this method differs in that it doesn't simply list and superimpose individual stressors. Instead, it uses directed links to connect the four layers of source, transmission, exposure, and response into a complete causal chain. Based on this framework, the complete transmission process of "stressor generation → environmental transmission → biological exposure → individual response → population / community consequences" can be traced. Existing methods struggle to answer questions like "which path does the impact take?" and "which segment is the most vulnerable?" because they lack path structure and a cross-level link quantification system. This method constructs an traversable path topology through S1 and introduces seven attributes—I (spatial-temporal importance), Sen (response sensitivity), P (probability of occurrence), C (overall strength), R (risk level), E (level of evidence), and U (uncertainty type)—through S2, enabling the risk level of each link to be distinguished and compared. Building upon this, S3 uses the lowest-risk link on the path as the risk value for the entire path. This judgment corresponds to the actual situation in causal transmission where "the weakest link determines the overall transmission efficiency," unlike the simple summation or averaging of the risks of each link. S4 further utilizes path data to identify three types of nodes: common skeleton nodes appear repeatedly in multiple paths, representing common vulnerable points where multiple pressure sources converge; source-specific front-end entry nodes appear only in specific source paths, serving as diagnostic criteria for identifying major pressure sources; and bridge / terminal amplifier nodes are located between low-level and high-level responses, with output risk higher than input risk, making them key locations affecting whether amplification occurs. These mechanisms collectively support the identification and comparison of cumulative effect transmission paths, remaining operational even under conditions of insufficient data.

[0012] Through the above four steps, this method advances environmental impact assessment from static correlation description to a level where causal paths are traceable, bottlenecks are identifiable, and the contributions of stressors are comparable, providing a basis for subsequent monitoring program design and priority determination of mitigation measures.

[0013] In some embodiments, the importance of the parent node and the sensitivity of the child nodes All use a 4-level rating system; the probability of occurrence P of the causal event is classified into 5 levels, including rare, unlikely, possible, very likely, and almost certain.

[0014] The above technical solution limits the importance of the parent node's role. Sensitivity of child nodes A four-level rating system is used to determine the probability of causal events. A classification system with five levels is adopted.

[0015] In the aforementioned method, and It is to calculate the overall strength The basic attributes, It is to determine the risk level One of the two inputs. If the hierarchical structure is not specified, different implementers may use the same link... The values ​​may vary significantly, affecting the repeatability of the evaluation results. A four-level rating scale is used to define... and This is equivalent to setting a unified measurement standard for judgments like "importance" and "sensitivity," which are inherently qualitative. The system employs five levels: "rare, unlikely, possible, very likely, and almost certain," with each level corresponding to a specific probability range (e.g., <1%, 1%-20%, 20%-50%, 50%-95%, ≥95% as described in the instruction manual). This transforms the "probability of occurrence" from a vague qualitative statement into an operational level judgment. In practice, environmental impact assessments for deep-sea mining often rely on expert judgment and limited observational data; different experts may have different judgments on the same causal relationship. With a fixed grading system, even if the information sources and judgment subjects differ, the assessment results remain comparable and verifiable. Compared to simple three-level (possible / uncertain / impossible) or two-level (will / will not), the five-level probability classification can accommodate more intermediate states. For example, the probability difference between "possible" and "very likely" may imply completely different response measures in management decisions.

[0016] In some embodiments, the combined effect strength Divided into four intervals based on numerical values: For negligible overall strength, For low overall strength, For medium overall strength, For high overall strength; the aforementioned risk level Divided into five levels: rare risk, low risk, medium risk, high risk, and extremely high risk, through... The level is row, The risk matrix is ​​determined by the interval of the column.

[0017] The above technical solution limits the intensity of the combined action. The four numerical ranges and risk levels pass interval and Methods for determining the risk matrix through joint risk assessment.

[0018] Depend on and Multiplying them together, the value falls between 1 and 16. If the interval is not divided, The value itself is a relatively abstract numerical value—C=9 and C=10 may not have any essential difference in practical meaning, but they are different numerically. Divided into After four equidistant intervals, consecutive numerical values ​​are mapped to four levels with clear semantic meanings: "negligible," "low," "medium," and "high." This reduces the cost of understanding and avoids over-interpreting minor differences at numerical boundaries. The method for determining the risk level R goes a step further: By means of The level is row, The risk matrix, with intervals as columns, is used to look up a table to determine the value. Essentially, this method allows... and In decision When it has asymmetric weights—for example, the same For high, Found when "possibly" and "almost certainly". The possible values ​​are "high" and "extremely high" respectively; similarly "Very likely" Found when "medium" and "high" It could also be different. The matrix form is essentially a semi-quantitative risk synthesis, merging two different dimensions of input (intensity and probability) into a five-level risk output. This approach does not rely on a large amount of measured data to fit a functional relationship, but rather on pre-setting lookup rules based on domain knowledge and expert judgment, which is precisely the approach that is feasible under conditions of data scarcity.

[0019] In some embodiments, the level of evidence A three-tier classification system of high, medium, and low uncertainty levels is adopted; use Parameter uncertainty, Model structure uncertainty, A three-level classification system for situational uncertainty.

[0020] The above technical solution defines the level of evidence. The three-level classification method (high, medium, low), and uncertainty levels The three-level division method ( Parameter uncertainty, Model structure uncertainty, Scenario uncertainty).

[0021] In the aforementioned evaluation system, This constitutes a characterization of the "strength" and "risk" of the causal link itself. However, environmental impact assessments for deep-sea mining face a practical problem: the risk levels of the same set of links may be the same, but the strength of the evidence supporting them may be completely different—for example, the link "sediment plume → advection transport" is supported by multiple field observations and numerical simulations (high level of evidence), while the link "heavy metal release → biomagnification" may only have indirect evidence from shallow-water analog studies (medium or low level of evidence). If the assessment results only give the risk level without indicating the sufficiency of the evidence behind it, decision-makers may give equal trust to high-risk links with low evidence support and high-risk links with high evidence support, which is dangerous from a management perspective.

[0022] The three-level division is essentially a confidence label: it does not change The value itself, but telling the user this How reliable are the values? High-level values ​​mean there is direct evidence to support them, medium-level values ​​mean there is indirect or analogical evidence, and low-level values ​​mainly rely on expert judgment. The three-level classification further illustrates the types of sources of uncertainty— This indicates that there is a consensus on the causal direction, but the parameters are not precise enough. The functional form representing causality itself has multiple possibilities. This indicates that the uncertainty primarily stems from the inherent uncertainty of future mining scenarios. The management implications of these three categories differ: The number can be reduced by supplementing observations. More mechanistic studies are needed to differentiate between the different hypotheses. This depends on the specific mining plan itself. In this plan, each link... and These attributes are evaluated and recorded together with other attributes, which gives each link in the causal network a "credibility label" and a "source of uncertainty description". This allows subsequent users to distinguish which conclusions are reliable and which need to be treated with caution.

[0023] In some embodiments, in A4, the common skeleton node is a node that appears in the path of no less than three different source nodes, is located in the effect node cascade segment, and has an average risk level of ingress links that is not lower than medium risk. The source-specific front-end entry node is a node that appears as an initial effect node only in the path of one or two source nodes, whose exposure path is directly related to the physical and / or chemical properties of the source node, and whose path from the node converges to the common skeleton node. The bridge / terminal amplifier node is a node that receives low-level response input and outputs high-level response, with the average risk level of outgoing links being higher than the average risk level of incoming links, and the betweenness centrality being higher than a preset threshold.

[0024] In the above technical solution, the quantitative judgment criteria for common skeleton nodes, source-specific front-end entry nodes and bridge / terminal amplifier nodes are further defined.

[0025] The aforementioned method S4 has already provided basic definitions for three types of critical nodes, but the expressions such as "multiple," "repeatedly appearing," and "specific" in the definitions leave room for interpretation. This scheme specifies the criteria for judgment: the criteria for common skeleton nodes are that they appear in at least three source node paths, are located in the cascade segment of effect nodes, and have an average risk of not less than medium risk in the ingress link—the first two criteria define the quantitative meaning of "common" and "skeleton," while the ingress link risk condition avoids including all high-frequency nodes in the skeleton, excluding those nodes that appear frequently but have low risk contribution. The three criteria for source-specific front-end entry nodes—source number limit (1-2), direct correlation between exposure path and source characteristics, and path convergence to common skeleton nodes—jointly ensure that the identified nodes are indeed unique to a specific source, the mechanism is interpretable, and they ultimately converge into a common transmission path. Bridge / terminal amplifier nodes, based on the aforementioned definitions, add the requirement that their betweenness centrality is higher than a preset threshold. This is an objective measure in network topology analysis used to quantify the criticality of a node as a "bridge."

[0026] The purpose of the aforementioned quantitative judgment is to ensure consistency in the results obtained by different evaluators when identifying nodes in the same set of path data, due to the clear node selection criteria, thus avoiding identification discrepancies that may arise from qualitative descriptions. The three types of nodes each correspond to different management implications: common skeleton nodes are vulnerable links shared by multiple pressure sources; focusing on monitoring these nodes can simultaneously block multiple transmission paths, which is more efficient than monitoring each path individually; source-specific front-end entry nodes can serve as pressure source identification markers; when anomalies are detected in these nodes, the main pressure source can be traced based on the node type; bridge / terminal amplifier nodes have a higher output risk than input risk, meaning the consequences are amplified here; intervention at this stage has a greater blocking effect than intervention at the amplifier's front end.

[0027] In some embodiments, the method further includes A5, calculating a comprehensive score for each source node based on the complete causal path set of each source node and sorting them from high to low scores; The overall score is calculated using the following formula:

[0028] in, This is the arithmetic mean of the minimum risk values ​​of all paths to this source node. This represents the percentage of high-risk paths within the source node's path. This represents the total number of paths to the source node.

[0029] In the above technical solution, based on the evaluation system constructed by the aforementioned method, a source node importance ranking step is added, and each pressure source is quantitatively scored and ranked through a comprehensive scoring formula.

[0030] The aforementioned methods have completed the construction of causal networks and the quantification of path risks, resulting in a set of path-level risk data. However, in management decision-making, it is more important to answer the question of "which of the multiple pressure sources should be addressed first." This solution deals with the aggregation from path data to the priority of source nodes.

[0031] The three terms in the formula each have their own function. It is the average of the minimum risk values ​​of all paths under each source node, reflecting the overall risk level of the path of that source node. It represents the proportion of high-risk paths, measuring risk concentration—between the two source nodes. They may be the same, but one has a much higher proportion of high-risk paths than the other, meaning that the former's problems are more concentrated and prominent. The number of paths is taken into account; more paths indicate a more diverse range of influence pathways for the source node. However, its weight is only 0.1, far lower than the previous two items, to prevent the number of paths from dominating the ranking result. A weight of 2 is assigned... This approach prioritizes the proportion of high-risk paths in the ranking, aligning with typical management logic: pressure sources with concentrated high-risk paths should be controlled first. Together, these three factors create a comprehensive score that considers overall performance, highlights extreme cases, and takes into account the diversity of impacts.

[0032] In some embodiments, the method further includes A6, identifying the SPER stage containing the link with the lowest risk level on each complete causal path as a bottleneck, and classifying the bottleneck into four categories: source-forming bottlenecks. Located in the stage from the source node to the transit node; exposing implementation bottlenecks. Located in the stage from the transmission node to the exposure node; bottleneck in effect realization. It is located in the stage from the exposure node to the effect node; the subsequent amplification bottleneck. Located in the cascade stage between effect nodes; the frequency and proportion of the four types of bottlenecks in the paths of each source node are statistically analyzed.

[0033] In the above technical solution, based on the causal network constructed by the aforementioned method, a bottleneck identification and classification statistics step is added. The stage where the link with the lowest risk level on each path is located is identified as the bottleneck, and it is aggregated and statistically analyzed according to four categories.

[0034] The four types of bottlenecks correspond to different stages of the SPER chain: Located in the source-to-transmission stage, it means that although the pressure source itself exists, its conversion efficiency to the environmental medium is not high; Being in the transmission to exposure stage means that although environmental disturbances have occurred, the conditions for effective biological exposure are insufficient; Being in the exposure-to-effect stage means that although the organism has been exposed, its response to the exposure is weak or uncertain. The cascade stage between effects means that the initial effect has occurred, but the transmission from individuals to the population or community is limited.

[0035] These four categories have completely different management implications. For example, the bottleneck in sediment plumes mainly focuses on... (The subsequent magnification shows it accounts for approximately 57.7%), indicating that it can effectively trigger initial effects, but the transmission from individual responses to population consequences is limited. In this situation, management should focus on monitoring population-level changes rather than investing excessive resources in reducing plume concentrations at the source. Bottlenecks in heavy metal release are distributed in... (approximately 29.2%) and (Approximately 28%), indicating that the main problem lies in the efficiency of metal transformation from nodules to dissolved states and the efficiency of dissolved metal transfer to organisms. In response, management should focus on blocking the metal transformation process and biological exposure pathways, rather than investing excessive resources in population monitoring. Statistical analysis of the frequency and proportion of the four types of bottlenecks in each source node pathway essentially creates a "conduction obstruction distribution map" for each pressure source, allowing identification of where the conduction of that pressure source is blocked, thus enabling the allocation of management resources to the areas truly restricting conduction.

[0036] According to another aspect of the present invention, a system for constructing and evaluating environmental impact causal networks based on the SPER framework is provided, comprising, based on the above method, the following: The node identification module is used to construct the SPER four-layer framework, including: a source layer, which identifies multiple types of environmental stressors as source nodes; a transmission layer, which identifies multiple transmission processes of the stressors in the environmental medium as transmission nodes; an exposure layer, which identifies multiple exposure pathways of organisms to the stressors or their effects as exposure nodes; and a response layer, which identifies multiple effect nodes exhibited by the organisms after exposure, from the individual to the ecosystem level as effect nodes; and directed links are used to represent the causal relationships between the nodes, forming a four-layer causal network. The attribute assignment module is used to assign seven attributes to each directed link: the importance of the parent node. Characterizes the spatial and temporal importance of the influence of the cause on the effect; child node sensitivity. Characterizes the sensitivity of the outcome to changes in the cause; the probability of causal event occurrence P; and the overall strength of the effect. Risk level Level of evidence The level of uncertainty characterizes the sufficiency of scientific evidence. Characterizes the main types of sources of uncertainty; The path analysis module is used to traverse directed links from each source node, constructing a complete causal path from the source node through the transit and exposure nodes to the terminal effect node in the direction of source node → transit node → exposure node → effect node; and minimizing the path risk value. Quantify the overall risk level of each path, among which, For the first on the path The node and the first Risk level of links between nodes; The critical node identification module is used to identify three types of critical nodes based on the frequency of occurrence, location distribution, and risk level of preceding and following links of each node in the complete causal path: common skeleton nodes, which are nodes that appear repeatedly in multiple paths from different source nodes and are located in the cascade segment of effect nodes; source-specific front-end entry nodes, which are nodes that appear only as initial effect nodes in specific source node paths; and bridge / terminal amplifier nodes, which are nodes that receive low-level response inputs and output high-level responses with output risk higher than input risk.

[0037] In order to better utilize the above methods, this application proposes an environmental impact causal network construction and evaluation system based on the SPER framework. Each module corresponds to a step of the above methods, and its specific principles have been described above and will not be repeated here.

[0038] According to another aspect of the present invention, an apparatus for constructing and evaluating environmental impact causal networks based on the SPER framework is provided, comprising: At least one processor and a memory communicatively connected to said at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method described above.

[0039] In the above technical solution, to better operate and process the method, the method is stored in memory, and the processor executes the stored method. It should be noted that the principle and effect of each step have been described above and will not be elaborated upon here.

[0040] According to another aspect of the present invention, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the above-described method.

[0041] In the above technical solution, to better operate and use the method, the method is stored in a computer-readable storage medium and implemented using a processor. It should be noted that the principle and effect of each step have been described above and will not be elaborated upon here. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating an embodiment of a method for constructing and evaluating environmental impact causal networks based on the SPER framework according to the present invention. Figure 2 This is a schematic diagram of a four-layer SPER framework structure of an embodiment of the method for constructing and evaluating environmental impact causal networks based on the SPER framework of the present invention. Figure 3 This invention relates to an embodiment of a method for constructing and evaluating causal networks of environmental impacts based on the SPER framework, specifically focusing on the comprehensive effect strength. With causal probability Joint risk assessment matrix diagram Figure 4 This is a schematic diagram of a local structure of a causal network centered on a source node (such as S1 sediment plume or S2 hard matrix removal), according to an embodiment of the SPER framework-based method for constructing and evaluating environmental impact causal networks. Figure 5 This is a schematic diagram illustrating the bottleneck analysis results of an embodiment of the environmental impact causal network construction and evaluation method based on the SPER framework of the present invention. Figure 6 This is a schematic diagram of an embodiment of an environmental impact causal network construction and evaluation system based on the SPER framework of the present invention. Detailed Implementation

[0044] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Example 1 This invention provides a method for constructing and assessing environmental impact causal networks based on the SPER framework, addressing the technical problems in existing technologies such as the lack of ability to construct complete causal chains from multiple pressure sources to multi-level ecological consequences, the lack of ability to assess the cumulative effects of multiple pressure sources, and the difficulty in conducting systematic assessments under conditions of scarce data. This embodiment uses polymetallic nodule mining activities in the Clarion-Clipperton Fault Zone (CCZ) as the assessment object to fully implement the method of this invention.

[0046] Please see Figure 1 A method for constructing and evaluating causal networks of environmental impacts based on the SPER framework, the method comprising: A1. Construct a four-layer SPER framework, including: a source layer, identifying multiple types of environmental stressors as source nodes; a transmission layer, identifying multiple transmission processes of the stressors in the environmental medium as transmission nodes; an exposure layer, identifying multiple exposure pathways of organisms to the stressors or their effects as exposure nodes; and a response layer, identifying multiple effect nodes exhibited by organisms from the individual to the ecosystem level after exposure as effect nodes; and use directed links to represent the causal relationships between the nodes, forming a four-layer causal network. In this embodiment, the input for this step is a description of the technical solution for deep-sea mining activities, including the type of mining equipment (mining machine, riser system, mining vehicle, surface support vessel), operating parameters (operating depth, operating speed, operating width, annual mining area), emission parameters (tailings discharge depth, emission concentration), and environmental baseline characteristics of the mining area.

[0047] This step assesses deep-sea mining activities, following the four-layer structure of the SPER framework. (Please refer to [link / reference]). Figure 2 This diagram illustrates the four-layer SPER framework structure adopted in this invention, from top to bottom: Source (6 types of pressure sources), Pathway (15 types of transport processes), Exposure (12 types of exposure pathways), and Response (23 types of response nodes, divided into three sub-layers). Arrows indicate the direction of causal links between and within layers, different colors or line types distinguish physical, chemical, and biological-ecological transport processes, and node size indicates the importance of each layer. This diagram visually demonstrates the system framework and component composition of the environmental impact assessment for deep-sea mining.

[0048] The system identifies the following four types of nodes: (1) Source Nodes – Identify 6 types of stress sources Source nodes represent environmental pressure sources directly generated by deep-sea mining activities. Through systematic analysis of the mining operation process, the following six types of pressure sources were identified: S1 Sediment Plume: Generated by mechanical disturbance of sediments by seabed collection equipment and tailings water discharge. Specifically, it includes two subcategories: one is the plume generated by sediment collection equipment, characterized by a high initial suspended solids concentration (…). ), low release altitude (distance from the seabed) Secondly, there are tailings water plumes, which are categorized into surface, mid-layer, and near-bottom layers based on their depth. Sediment plumes are the most direct manifestation of the environmental impact of deep-sea mining, and their diffusion range and duration are core parameters for assessment.

[0049] S2 Hard Matrix Removal: Removal of deep-sea bottom hard matrix habitat due to nodule collection processes. The operating width of the ore-collecting machine is typically... Movement speed is The annual direct disturbance area is The removal of hard matrix directly disrupts the biological community that uses polymetallic nodules as an attachment substrate, causing irreversible damage to sessile organisms (such as sponges, corals, and foraminifera).

[0050] S3 underwater noise: originates from four types of equipment: ore collectors (operating frequency) Sound source level re ), riser system (sound source level approximately Mining vehicle (sound source level approximately) ) and surface support vessels (sound source class approximately Underwater noise, through sound wave propagation, affects the acoustic communication, navigation, predation, and predator avoidance behaviors of deep-sea organisms.

[0051] S4 Artificial Light: The mining machine is equipped with LED or halogen lamps to assist in deep-sea visual operations, with a total power of [missing information]. Natural light at the bottom of the deep sea is almost zero (photosynthetically active radiation is close to zero), and the introduction of artificial light sources poses a significant light environmental pressure on deep-sea organisms that are adapted to the dark environment.

[0052] S5 Heavy Metal Release: Polymetallic nodules contain heavy metals such as manganese (Mn), nickel (Ni), cobalt (Co), and copper (Cu), with concentrations significantly higher than background values ​​in open ocean sediments. Mining disturbances can lead to nodule fragmentation, sediment resuspension, and pore water release, thereby triggering metal speciation (from stable oxide-bound states to dissolved and exchangeable states) and the release of bioavailability.

[0053] S6 DIC release: Deep-sea bottom waters are characterized by high DIC concentrations. The DIC concentration in sediment pore water is higher (average approximately...). Mining disturbances release DIC-rich pore water into the bottom waters, potentially altering the local carbonate chemical balance.

[0054] (2) Pathway Nodes – Identifying 15 types of transit processes Transmission nodes represent the physical, chemical, and biological transfer processes in the environmental medium from the generation of a pressure source to the formation of biological exposure. Identify the following 15 types of transmission processes: Physical transfer process: advective transport Turbulent diffusion Flocculation and settling Resuspension Acoustic propagation Light attenuation.

[0055] Chemical transport process: Redox transformation Organic complexation Adsorption-desorption Carbonate equilibrium.

[0056] Biological-ecological transfer processes: Organic matter mineralization Dissolved oxygen consumption Biomagnification Habitat fragmentation Microhabitat loss.

[0057] (3) Exposure Nodes – Identifying 12 types of exposure pathways Exposure points represent the pathways and ways in which an organism is actually exposed to a stressor or its environmental effects. Identify the following 12 types of exposure pathways: Physical exposure pathways: Gill particle contact. Physical burial Acoustic exposure Light exposure.

[0058] Ingestion exposure routes: Filter-feeding ingestion Deposit-feeding ingestion.

[0059] Chemical exposure pathways: Dissolved uptake pH exposure Hypoxia exposure Carbonate undersaturation exposure.

[0060] Ecological exposure pathways: Habitat loss Changes in food availability.

[0061] (4) Response Nodes – Identifying 23 types of responses Effector nodes represent the responses exhibited by organisms from the individual to the ecosystem level after exposure. The following 23 types of responses are identified and categorized into three classes based on biological organizational level: Individual-level response ( ): Decreased filter efficiency Decreased breathing efficiency Oxidative stress, Metal accumulation, Hearing impairment Visual impairment, Calcification inhibition, Growth inhibition, Reproductive damage, 0 behavior change Avoidance behavior Individual death.

[0062] Population-level response ( ): Decreased population density Decreased population replenishment rate Changes in population age structure Decreased genetic diversity Local extinction.

[0063] Community-ecosystem level response ( ): species richness decline Changes in community structure Decreased functional diversity Changes in food web structure Ecosystem function decline Loss of ecosystem services.

[0064] Output: A four-layer SPER node system is formed, consisting of 6 source nodes, 15 transit nodes, 12 exposure nodes, and 23 effect nodes, for a total of 56 nodes. Each node is accompanied by a type identifier, hierarchical attributes, and descriptive definitions.

[0065] For example, the steps are as follows: (1) Source node identification: Identify the following six types of pressure sources: S1 Sediment Plume: A collaborative operation scheme of ore collector, riser, and surface support vessel is adopted. During the operation of the ore collector, a high-concentration suspended solids plume is generated near the bottom (initial concentration 3000-10000 mg / L), and the tailings water is discharged at a depth of about 1000m to form a mid-level plume. Under typical CCZ flow field conditions (flow velocity 2-10 cm / s), the near-bottom plume can spread to a range of several kilometers within a few hours.

[0066] S2 Hard Substrate Removal: The mining machine operates with a width of 12m, a moving speed of 0.4m / s, operates for 300 days per year, and directly disturbs an area of ​​approximately 350km² per year. The polymetallic nodule coverage in the CCZ area is typically 20%–70%. As the main attachment substrate for sessile organisms, the removal of nodules directly leads to changes in habitat substrate conditions.

[0067] S3 underwater noise: The main noise source is the noise from mining machinery operation, with a source level of approximately 170 dB and a frequency range of 10–500 Hz. The noise propagates through seawater, and at a water depth of 5000 m, the propagation loss follows a spherical expansion law, with an impact range of up to tens of kilometers.

[0068] S4 Artificial Light: The mining machine is equipped with an LED lighting system with a total power of 5kW and a color temperature of approximately 5000K. Since the CCZ bottom layer has zero natural light, the introduction of artificial light sources creates an illuminated area with a diameter of approximately 10-20m.

[0069] S5 heavy metal release: CCZ nodules contain approximately 30% Mn, 1.3% Ni, 0.2% Co, and 1.0% Cu. Mining disturbances cause nodule fragmentation and sediment resuspension, releasing bound metals into the water, some of which are converted into dissolved and exchangeable states.

[0070] S6DIC release: DIC concentration in the bottom water of CCZ is approximately... The DIC concentration in sediment pore water is approximately Mining disturbances release pore water, increasing the DIC load on bottom water bodies.

[0071] (2) Transmission node identification: Based on the environmental characteristics of the CCZ area and the physicochemical properties of the mining pressure source, the following transmission processes are identified: The physical transfer process includes: Advection (horizontal movement of suspended particles with ocean currents) Turbulent diffusion (fluid mixing leading to plume dilution) Flocculation sedimentation (accelerated sedimentation after fine particles flocculate) Resuspension (bottom current disturbance causes sediments to be resuspended) Sound wave propagation (noise propagation in seawater), Light attenuation (absorption and scattering of light by suspended particles and seawater).

[0072] Chemical transport processes include: Redox transformation (changes in redox conditions driving metal morphology transformation) Organic coordination (organic ligands complexing with metal ions) Adsorption-desorption (equilibrium of adsorption and desorption on particulate matter surface) Carbonate equilibrium (dissolved inorganic carbon causes local pH and carbonate saturation changes).

[0073] Bio-ecological transfer processes include: Organic matter mineralization (decomposition of organic matter in sediments consumes oxygen) Dissolved oxygen consumption (respiration and oxidation processes consume dissolved oxygen in the atmosphere) Biomagnification (metal enrichment along the food chain). Habitat fragmentation (continuous habitat is broken into isolated patches) Microhabitat loss (reduction of the matrix on which tuberculosis attaches leads to the disappearance of the microhabitat).

[0074] (3) Exposure Node Identification: Based on the ecological characteristics of representative biological groups (including macrobenthic animals, microorganisms, swimming organisms, etc.) in the CCZ area, the following exposure pathways were identified: Gill particle contact (suspended particulate matter contacting the gills of benthic organisms) Filter feeding (filter-feeding organisms ingesting suspended particles) Sediment feeding (sediment-feeding organisms ingesting contaminated sediments) Absorption in dissolved state (absorption of dissolved metals through the body surface or gills) Physical burial (sedimentation particles covering the surface of organisms) Sound exposure (organisms in a noise field) Light exposure (organisms in areas with artificial light) Habitat loss (removal of hard matrix leads to habitat disappearance) Changes in food availability (changes in food supply and quality) pH exposure (pH changes in bottom water) Hypoxia exposure (decreased dissolved oxygen concentration) and E12 carbonate undersaturation exposure (decreased carbonate saturation affects calcified organisms).

[0075] (4) Effect node identification: Based on the ecological characteristics of known species in the CCZ region and the known responses of deep-sea organisms to various pressures, the following effect nodes were identified: Individual level: Decreased filter efficiency Decreased breathing efficiency Oxidative stress, Metal accumulation, Hearing impairment Visual impairment, Calcification inhibition, Growth inhibition, Reproductive damage, behavior change Avoidance behavior Individual death.

[0076] Population hierarchy: Decreased population density Decreased population replenishment rate Changes in population age structure Decreased genetic diversity Local extinction.

[0077] Community-ecosystem level: species richness decline Changes in community structure Decreased functional diversity Changes in food web structure Ecosystem function decline Loss of ecosystem services.

[0078] Directed links are used to represent the causal relationships between nodes, forming a four-layer causal network consisting of 6 source nodes, 15 transmission nodes, 12 exposure nodes, and 23 effect nodes.

[0079] A2. Assign seven attributes to each directed link: importance of the parent node. Characterizes the spatial and temporal importance of the influence of the cause on the effect; child node sensitivity. Characterizes the sensitivity of the outcome to changes in the cause; the probability of causal event occurrence P; and the overall strength of the effect. Risk level Level of evidence The level of uncertainty characterizes the sufficiency of scientific evidence. Characterizes the main types of sources of uncertainty; In this embodiment, the importance of the parent node is... and the sensitivity of the child nodes All use a 4-level rating system; the probability of occurrence P of the causal event is classified into 5 levels, including rare, unlikely, possible, very likely, and almost certain.

[0080] In this embodiment, the combined effect strength Divided into four intervals based on numerical values: For negligible overall strength, For low overall strength, For medium overall strength, For high overall strength; the aforementioned risk level Divided into five levels: rare risk, low risk, medium risk, high risk, and extremely high risk, through... The level is row, The risk matrix is ​​determined by the interval of the column.

[0081] In this embodiment, the level of evidence A three-tier classification system of high, medium, and low uncertainty levels is adopted; use Parameter uncertainty, Model structure uncertainty, A three-level classification system for situational uncertainty.

[0082] In this embodiment, the input to this step is the set of potential causal relationships between all nodes identified in step S1 (a set of candidate links constructed based on literature analysis and expert knowledge). This step performs a semi-quantitative evaluation of each candidate link (directed edge between nodes) in the causal network by defining the following seven attributes: (1) Importance of the role of the parent node

[0083] The importance of parent nodes characterizes the spatial extent and temporal duration of the influence of parent nodes (i.e., the initiators of causal relationships) on child nodes (i.e., the responders of causal relationships). A four-level rating system is used.

[0084] (Negligible): The impact persists only during mining activities and is spatially limited to a small test site (area). ); (Small): The impact lasts for weeks to months after mining activities cease, and the spatial extent is limited to a single mining cluster (approximately [area missing]). ); (China): The impact lasted for several years after mining activities ceased, and its spatial scope was limited to the mining contract area (approximately...). ); (Large): The impact is permanent or extends beyond the contract area, at a regional scale (area). It produces a significant effect.

[0085] Assessment Methodology: Based on observational evidence from literature review and numerical simulation results, the temporal duration and spatial scope of the impact are comprehensively assessed.

[0086] (2) Sensitivity of child nodes

[0087] Child node sensitivity characterizes how readily a child node responds to changes in its parent node; that is, the degree to which a unit change in the parent node causes a significant change in the child node. A four-level scoring system is used.

[0088] (Non-sensitive): Child nodes can be fully represented throughout the entire CCZ and have no obvious response to changes in the parent node; (Low sensitivity): Child nodes are likely to be fully represented across the entire mining contract area and have a weak response to changes in the parent node; (Medium Sensitivity): Child nodes are likely to be fully represented throughout the entire mining area and have a moderate degree of response to changes in the parent node; (Highly sensitive): Child nodes are only found in local test fields or their response to parent nodes is uncertain, but they respond strongly to changes in parent nodes.

[0089] Assessment method: Based on experimental observation data and field survey data from deep-sea ecology literature, and combined with species ecological characteristics (such as life history strategies, metabolic rate, tolerance range, etc.), a comprehensive judgment is made.

[0090] (3) Probability of causal events

[0091] The probability of a causal event represents the level of probability of that causal relationship actually occurring under typical mining operating conditions. A 5-level classification system is used:

[0092] The corresponding probability interval is: Rare: Probability of occurrence ; Unlikely: Probability of occurrence and ; Possible: Probability of occurrence and ; Very likely: probability of occurrence and ; Almost certain: probability of occurrence .

[0093] Assessment method: Based on existing deep-sea mining impact observation data (such as the DISCOL experiment, JPI Oceans mining experiment, etc.), analogy inferences from similar disturbance experiments, and expert judgment, the assessment method is determined comprehensively.

[0094] (4) Strength of the combined effect of the link

[0095] The combined effect strength of a link is calculated as the product of the importance of the parent node's effect and the sensitivity of the child node:

[0096] The range of values ​​is The larger the value, the stronger the potential impact of the causal link.

[0097] (5) Risk level

[0098] Risk level is based on the overall strength of the link's effect. and the probability of causal events The joint assessment is determined through a risk matrix. The rows of the risk matrix correspond to... The five levels are listed in the column. The four level ranges ( For negligible overall strength, For low overall strength, For medium overall strength, (For high overall intensity), matrix cell filling risk level:

[0099] The risk level is divided into 5 levels: R1 Rare Risk: Risks that require almost no attention; R2 Low risk: Risks that require monitoring but not active mitigation; R3 Medium risk: Risk requiring mitigation measures; R4 High: Risk requiring strict mitigation measures and continuous monitoring; R5 Very High: Risk that may cause irreversible ecological damage, requiring consideration of prohibiting or strictly restricting related activities.

[0100] Please see Figure 3 The figure shows the intensity of the combined effect in tabular form. With causal probability The joint risk assessment matrix. The rows of the table correspond to probabilities. The five levels (rare, unlikely, possible, very likely, almost certain) correspond to the overall strength. The four intervals ( , , , The cells are filled with the corresponding risk level (rare, low, medium, high, extremely high), and the increasing risk is visually represented by a color gradient (e.g., from gray to red). This diagram is used to illustrate the method for determining the risk level in step S3.

[0101] (6) Level of evidence

[0102] The level of evidence characterizes the sufficiency of scientific evidence supporting the determination of causality. A three-level classification system is used:

[0103] (High level of evidence): There must be at least three peer-reviewed papers providing direct evidence, or at least one controlled experimental study on a deep-sea species to support the causal relationship; (Medium level of evidence): Supported by 1-2 direct evidence articles, or multiple indirect / analogical evidence articles (such as shallow water species experiments, theoretical model inferences, etc.); (Low level of evidence): Based solely on expert judgment, theoretical inference, or a single preliminary study, lacking systematic scientific evidence.

[0104] (7) Uncertainty level

[0105] The uncertainty hierarchy characterizes the primary sources of uncertainty in this causal relationship assessment. A three-level classification system is adopted:

[0106] (Parameter uncertainty): There is a consensus on the direction (positive or negative) and basic form of causal relationship, but the accuracy of parameter values ​​is insufficient. The main uncertainty comes from parameter estimation error. (Model structure uncertainty): The mathematical function form of causal relationship itself has multiple possibilities, and different model assumptions may lead to qualitatively different prediction results; (Scenario uncertainty): The assessment results of causality are highly dependent on the implementation scenario of future mining activities (such as mining technology, mitigation measures, regulatory intensity, etc.), and the uncertainty of the future scenario is the main source of uncertainty in the assessment results.

[0107] This step outputs a complete evaluation result for each causal link, along with seven attributes, forming a directed graph of the causal network with attribute annotations. See also... Figure 4 This figure illustrates the local structure of a causal network centered on a source node (such as the S1 sediment plume or the S2 hard matrix removal), including typical path examples from the source node through transit nodes and exposure nodes to the effect nodes. Nodes in the figure are labeled with their names and types, and links are labeled with their combined effect strength. and risk level The numerical or color-coded values ​​are used to illustrate the specific structure of causal networks and the way link attributes are labeled.

[0108] For example, link " Sediment plume A quantitative assessment is conducted using "advection transport" as an example: Importance of the role of parent nodes (China): Sediment plumes can leave suspended particles in the water for hours to days after mining activities cease, covering the entire mining area.

[0109] child node sensitivity (Medium): Advection transport processes are moderately sensitive to changes in plume concentration and flow rate, and changes in flow field conditions can significantly affect transport efficiency.

[0110] Causal probability "Almost certain" (≥95%): Under typical CCZ flow field conditions, the advection transport of sediment plumes is an inevitable physical process.

[0111] Overall intensity .

[0112] Risk level "Extremely high" (Check risk matrix:) Corresponding to high overall strength, =“Almost certain” corresponds to an extremely high-risk cell.

[0113] Level of evidence High (H): Multiple deep-sea mining experiments have directly observed the advection transport process of sediment plumes, and numerical simulation studies provide support.

[0114] Uncertainty level (Parameter uncertainty): The physical mechanism of advection transport is well understood, and the main uncertainty comes from the precise values ​​of flow field parameters and plume release parameters.

[0115] Taking the link "E5 physical burial → decreased R2 breathing efficiency" as an example: (The effect of increased burial thickness on respiration efficiency is continuous and cumulative.) (Beneficial organisms are highly sensitive to burial.) "Very likely" (≥50%) , "Extremely high" =M (M) =U2 (Different species have different response functions to burial).

[0116] The above quantitative evaluation is performed on all candidate links in the network to form a complete causal network with seven attribute labels.

[0117] A3. Starting from each source node, traverse along directed links, constructing a complete causal path from the source node through the transit nodes and exposure nodes to the terminal effect node in the direction of source node → transit node → exposure node → effect node (wherein, after entering the response layer, the path first reaches the individual-level response node, and then reaches the population or community-level terminal effect node through the cascading links between effect nodes, so effect node → effect node represents the cascading transmission of the effect within the response layer); using the path with minimum risk value Quantify the overall risk level of each path, among which, For the first on the path The node and the first Risk level of links between nodes; In this embodiment, the input to this step is the causal network directed graph with attribute annotations constructed in step S2, which includes all directed links between nodes and the evaluation results of their seven attributes.

[0118] This step parses the following four types of linkages and constructs a complete causal path according to the chronological order of the SPER chain: (1) The link from the source node to the transit node ( ) Assess how each pressure source enters the environmental medium and initiates the transmission process. For each source-transmission node pair, determine whether a direct causal link exists (i.e., whether the pressure source can directly trigger the transmission process). If it exists, record the link and assign it the seven attributes defined in step S3.

[0119] For example: Sediment plume Advection transport (direct link, plumes achieve spatial diffusion through advection); Sediment plume Light attenuation (direct link, suspended particles cause light attenuation); Heavy metal release Redox transformation (direct link, changes in redox conditions drive metal morphology transformation).

[0120] (2) Passing the link from the node to the exposed node ( ) Assess how each transport process constitutes an exposure pathway for the organism. For each transport node-exposure node pair, determine whether the outcome of the transport process constitutes a condition for biological exposure.

[0121] For example: Horizontal transport Physical burial (advection transports suspended sediments to biological habitats, forming burial sites); Redox Conversion Dissolved absorption (redox transformation alters the dissolved concentration of metals, thus affecting biological absorption); Habitat fragmentation Habitat loss (habitat fragmentation directly leads to a reduction in habitat area).

[0122] (3) Links from exposed nodes to effect nodes ( ) Assess how each exposure pathway triggers a biological response. For each pair of exposure-effect node combinations, determine whether the exposure is likely to lead to a response at that level.

[0123] For example: Physical burial Decreased respiratory efficiency (physical burial affects the respiratory process of benthic organisms); Soluble absorption Metal accumulation (the accumulation of dissolved metals in the body after absorption by organisms). Habitat loss Decreased population density (reduction in habitat area leads to a decrease in environmental carrying capacity).

[0124] (4) Cascaded links from effect node to effect node ( ) Assess the cascading transmission relationships between different levels of effects, i.e., how individual-level responses are transmitted to the population level, and how population-level responses are transmitted to the community-ecosystem level.

[0125] For example: Decreased filter efficiency Growth inhibition (decreased filter feeding efficiency leads to insufficient energy intake, which in turn inhibits growth); Growth inhibition The population replenishment rate decreased (individual growth inhibition affects the quality and quantity of reproductive bodies, thus reducing population replenishment). Decreased population replenishment rate Species richness decline (the combined effect of declining population replenishment rates of multiple species leads to a decrease in community species richness).

[0126] Complete path construction: Starting from each source node, along... The network is traversed using a depth-first search to find all complete paths from the source node to the terminal effect node (an effect node with no outgoing degree). The path must start from the source node, pass through at least one transit node, one exposure node, and one effect node, and end at the terminal effect node.

[0127] Path risk quantification: For each complete path, the overall path risk is represented by the minimum path risk value.

[0128] in, For the first on the path The node and the first The risk level of the links between nodes (numerically represented as 1-5). This represents the overall risk level of the path. The reason for using the minimum value instead of the average is that, in a causal chain, risk transmission follows the "weakest link effect," meaning that the weakest link in the chain (the lowest-risk link) determines the efficiency with which the entire chain is triggered and transmitted.

[0129] A complete set of causal paths originating from each source node, with each path accompanied by a sequence of path nodes, a list of link attributes, and an overall risk level for the path.

[0130] For example, taking source node S1 (sediment plume) as an example, from Departure, according to Traverse the network in one direction. A typical complete path is: (Sediment plume) (Horizontal transport) (Flocculation and sedimentation) (Physical burial) (Decreased breathing efficiency) (Growth inhibition) (Population replenishment rate decline) (Decrease in species richness).

[0131] The risk levels of the links along this path are as follows: Extremely high, high, high, Extremely high, middle, middle, Medium. Minimum risk value of the path. Extremely high, high, high, extremely high, medium, medium, medium = Medium risk.

[0132] Path searches were performed on each of the six source nodes, ultimately forming a causal network containing key variables and complete paths. The node sequence, link attribute list, and overall risk level of each path were then analyzed.

[0133] A4. Based on the frequency of occurrence, location distribution, and risk level of preceding and following links of each node in the complete causal path, three types of key nodes are identified: common skeleton nodes, which are nodes that appear repeatedly in multiple paths from different source nodes and are located in the cascade segment of effect nodes; source-specific front-end entry nodes, which are nodes that appear only as initial effect nodes in specific source node paths; and bridge / terminal amplifier nodes, which are nodes that receive low-level response inputs and output high-level responses with output risk higher than input risk.

[0134] In this embodiment, the input to this step is the complete causal path set constructed in step S3, which includes the frequency of occurrence of each node in the path, its location distribution, and the risk level of the preceding and following links.

[0135] This step identifies three types of key nodes based on their location in the path, frequency of occurrence, and the risk level of their preceding and following connections: (1) Common Backbone Nodes Common skeletal nodes are effect nodes that repeatedly appear in multiple stressor pathways, forming a common transmission framework for the transformation of individual damage into population / community degradation. The identification criteria for such nodes are: Criteria for occurrence: It must appear in a path with at least 3 different source nodes; Location criterion: Located in the cascade segment of the effect node ( ), connecting individual-level responses and population / community-level responses; Risk standard: The average risk level of its inbound links (links pointing to this node) is not lower than medium risk.

[0136] Typical common skeleton nodes include: The population density decreases (almost all the paths from the source nodes pass through this node, which is the core hub for the transformation of individual effects into population consequences). The population replenishment rate decreases (the long-term effects of multiple source nodes are reflected through this node). Local extinction (an extreme manifestation of population consequences, located at the end of multiple pathways).

[0137] The significance of managing common backbone nodes lies in the fact that by focusing on monitoring and protecting the ecological processes represented by these nodes, the cascading amplification of multiple pressure source paths can be blocked simultaneously, achieving a "twice the result with half the effort" management effect.

[0138] (2) Source-Specific Front Entry Nodes Source-specific ingress nodes are initial effect nodes that determine how a specific pressure source enters the common framework. They are located at the front of the path of a specific source node and possess source identification characteristics. The identification criteria for such nodes are: Uniqueness criterion: It appears as an initial effect node only in the path of 1-2 source nodes; Source association criteria: The exposure pathway of this node is directly related to the physical / chemical characteristics of the source node; Pivotal criterion: Starting from this node, the path converges to the common skeleton node.

[0139] Typical source-specific front-end entry nodes include: Decreased filter efficiency (mainly due to) Sediment plume passes through Physical burial and Triggered by contact with gill particles, The iconic initial effect of the path); Visual impairment (mainly caused by) Artificial light passes through Light exposure triggers, is The iconic initial effect of the path); Metal accumulation (mainly by) Heavy metal release through Dissolution-initiated absorption is The iconic initial effect of the path.

[0140] The significance of managing source-specific front-end entry nodes lies in the fact that by monitoring the specific responses represented by these nodes, the current main stressor types can be traced and identified, providing a diagnostic basis for the formulation of targeted mitigation measures.

[0141] (3) Bridge / Terminal Amplifier Nodes Bridge / terminal amplifier nodes are critical nodes that determine whether an influence can be amplified or propagated to a higher level of consequences, acting as a "bridge" connecting upstream and downstream processes or as an "amplifier" for gain. The identification criteria for such nodes are: Bridging standard: The incoming links of this node come from lower-level responses, and the outgoing links point to higher-level responses; Amplification Criterion: The risk level of the outgoing link of this node is higher than the risk level of the incoming link (that is, the risk is "amplified" after passing through this node). Network location criteria: network topology analysis shows high betweenness centrality, such as ranking in the top 10% based on network size.

[0142] Typical bridge / terminal amplifier nodes include: The population replenishment rate decreased (from Growth inhibition and R9 reproductive damage reception effects, to Local extinction and The impact of declining species richness is an amplifier of individual-population consequences. Food web structure changes (receiving input from multiple population consequence nodes, and sending...) Ecosystem function decline and The output impact of ecosystem service loss is an amplifier of population-community consequences.

[0143] The significance of managing bridge / terminal amplifier nodes lies in the fact that these nodes are key hubs for the cascading amplification of risks. Intervening in the ecological processes represented by these nodes can effectively block the upward transmission of risks and prevent local impacts from escalating into systemic ecological consequences.

[0144] This step outputs a list of identification results for three types of key nodes, including node name, category, identification criteria (standards met), typical location in the path, frequency of occurrence statistics, and explanation of management significance.

[0145] For example, analysis based on the frequency and location of each node in the complete causal path: Common skeleton nodes: (Population density decrease) occurred in all 6 source node paths, with a frequency exceeding 800 times, located in The cascaded segments, with an average risk level of medium to high for incoming links, were identified as common skeleton nodes.

[0146] Source-specific front-end entry node: (Decreased filter efficiency) only It appears as an initial effect node in the path of the (sediment plume), and the exposure path is... (Gill particle contact) and (Physical burial) is directly related to the physical characteristics of sediment plumes, from The starting path converges to the common skeleton node. , was identified as The source-specific front-end entry node. (Metal accumulation) only It appears as an initial effect node in the (heavy metal release) pathway, and the exposure route is... (Dissolved absorption), directly related to the chemical characteristics of heavy metal release, has been identified as... The source-specific front-end entry node.

[0147] Bridge / Terminal Amplifier Node: (Decreased population replenishment rate) Received from (Growth inhibition) and The input of (reproductive damage) to (Local extinction) and (Decrease in species richness) Output, the average risk level of outgoing links is higher than that of incoming links, it has a high degree of intermediation centrality in the network topology, plays a role in risk amplification, and is identified as a bridge / terminal amplifier node.

[0148] In this embodiment, the method further includes: A5, calculating the comprehensive score of each source node based on the complete causal path set of each source node and sorting them from high to low scores; In this embodiment, the input to this step is the complete set of causal paths constructed in step S3, which includes the source node affiliation and path risk level of each path.

[0149] This step uses the following comprehensive scoring formula to rank the importance of each source node:

[0150] in, The minimum risk value of all paths under each source node level is the arithmetic mean. This value reflects the overall risk level of different source node paths and is the core basis for sorting. It has a weight of 1, which reflects the principle that risk level is the core evaluation dimension. This represents the proportion of high-risk paths at each source node level. This value reflects the concentration of high-risk paths at different source nodes and is a key indicator of risk severity. It has a weight of 2, reflecting the principle that high-risk paths are the key targets for control. This represents the sum of paths at each source node level.

[0151] For each source node, calculate the comprehensive score using the formula described above, and sort them from highest to lowest score to obtain the importance ranking of the source nodes. This step outputs the comprehensive score and importance ranking of each source node. A typical ranking result is: S1 Sediment plume > S2 Hard matrix removal > S5 Heavy metal release > S4 Artificial light > S6 DIC release > S3 Underwater noise.

[0152] For example, calculate the overall score for each source node: by (For example, a sediment plume): its path set contains 1426 paths, and the average minimum risk value of each path is... High-risk path proportion Number of paths The overall score is:

[0153] Similarly, the comprehensive scores of other source nodes were calculated, and the results are shown in Table 1 below: Table 1 Overall score calculation results

[0154] The table columns include source node number, number of paths, percentage of paths at each risk level, average minimum risk level, overall score, and ranking. (Sediment plume, score 2.91) (Hard matrix removal, score 2.88) (Heavy metal release, score 2.40) (Artificial light, score 2.39) (Dissolved inorganic carbon release, score 2.33) (Underwater noise, score 2.20).

[0155] The ordination results indicate that sediment plumes and hard matrix removal are the primary stressors of concern in this mining scenario, while underwater noise has the lowest relative importance.

[0156] In this embodiment, the method further includes: A6. For each complete causal path, identifying the SPER stage containing the link with the lowest risk level on the path as a bottleneck, and classifying the bottleneck into four categories: source-forming bottlenecks. Located in the stage from the source node to the transit node; exposing implementation bottlenecks. Located in the stage from the transmission node to the exposure node; bottleneck in effect realization. It is located in the stage from the exposure node to the effect node; the subsequent amplification bottleneck. Located in the cascade stage between effect nodes; the frequency and proportion of the four types of bottlenecks in the paths of each source node are statistically analyzed.

[0157] In this embodiment, the input to this step is the complete set of causal paths constructed in step S3, which includes the risk level of each link on each path.

[0158] This step identifies the bottleneck process that reduces the risk of each complete path, specifically the SPER stage containing the link with the lowest risk level on the path. Bottlenecks are categorized into the following four types: —Bottleneck at the source: The link with the lowest risk level on the path is located in the stage from the source node to the transmission node (S→P stage), indicating that the transmission and transformation process from the pressure source to the environmental medium is the bottleneck link that restricts the overall risk transmission. This type of bottleneck means that although the pressure source exists, its efficiency in entering the environment and forming a transmissible environmental disturbance is low.

[0159] —Bottleneck in Exposure Realization: The link with the lowest risk level on the path is located in the stage from the transmission node to the exposure node (P→E stage), indicating that the transformation process from environmental disturbance to biological exposure is the bottleneck limiting the overall risk transmission. This type of bottleneck means that although environmental disturbance has occurred, the conditions for effective biological exposure are insufficient.

[0160] —Bottleneck in effect realization: The link with the lowest risk level on the path is located in the stage from the exposure node to the initial effect node ( The phases (i.e., the transformation from exposure to an initial biological response) indicate that this is the bottleneck limiting overall risk transmission. Such bottlenecks mean that although organisms are exposed, their response to the exposure is weak or uncertain.

[0161] —Subsequent amplification bottleneck: The lowest-risk link in the path is located in the cascading stage between effect nodes (R→R stage), indicating that the amplification and transmission process from individual-level responses to population / community-level consequences is the bottleneck limiting the overall risk transmission. This type of bottleneck means that the initial effect has already occurred, but its cascading amplification to higher levels is limited.

[0162] Bottleneck Feature Analysis: For each source node, the frequency and proportion of the four types of bottlenecks in all its paths are statistically analyzed to identify differences in bottleneck characteristics among different pressure sources. If the bottleneck of a certain source node is mainly concentrated in (Subsequent amplification bottlenecks) indicate that this pressure source can stably trigger initial effects (high risk of upstream linkages), but the amplification and transmission of these effects to the population / community level is limited, and the overall risk is "cut off" at the individual level. Typical examples are the S1 sediment plume and the S2 hard matrix removal.

[0163] If the bottleneck of a certain source node is mainly concentrated in (The bottleneck is formed at the source) or (Exposure realization bottleneck) indicates that the environmental transformation process or exposure realization process of this pressure source limits the overall risk. Even if downstream effects are sensitive, the overall risk is still limited due to the low efficiency of upstream transmission. Typical examples are S5 heavy metal release and S6 DIC release.

[0164] If a source node experiences bottlenecks at multiple stages, it indicates that the overall transmission chain of that pressure source has multiple limitations, resulting in generally low risk transmission efficiency. Typical examples include S3 underwater noise and S4 artificial light.

[0165] This step outputs a bottleneck type distribution statistics table for each source node, including the frequency and percentage of each type of bottleneck and typical bottleneck link examples.

[0166] For example, please refer to Figure 5 This figure presents the bottleneck analysis results of step A6 in the form of a stacked bar chart or heatmap. The rows correspond to 6 source nodes, and the columns correspond to 4 types of bottlenecks (…). , , , The cell color depth or numerical value indicates the proportion of that type of bottleneck in the source node path. This graph is used to visually compare the differences in bottleneck characteristics among different pressure sources.

[0167] right Taking 1426 paths of sediment plumes as an example, bottleneck analysis was conducted, and the statistical results show that subsequent amplification of bottlenecks... It accounts for approximately 57.7%, representing a bottleneck in achieving its intended effect. Accounting for approximately 8.4%, this exposes the bottlenecks in achieving [the goal]. Accounting for approximately 14.1%, the source has become a bottleneck. It accounts for approximately 19.6%. Analysis conclusion: Sedimentary plumes can effectively trigger initial biological effects (with higher risks in upstream links), but there are significant limitations in the amplification and transmission of these effects to the population / community level. The main bottleneck lies in the cascading amplification process from individual responses to population consequences.

[0168] Pathway analysis of S5 (heavy metal release) shows that the source forms a bottleneck. Accounting for approximately 29.2%, this exposes bottlenecks in achieving [the goal]. It accounts for about 28%, and the bottleneck will be amplified later. It accounts for approximately 39.9%, representing a bottleneck in achieving its intended effect. It accounts for approximately 2.8%. Analysis conclusion: The overall risk of heavy metal release is limited by upstream processes—the efficiency of metal conversion from nodules / sediments to dissolved states and the efficiency of dissolved metal transfer to organisms are the main limiting factors.

[0169] Based on the above assessment results, the following management recommendations can be made: (1) Priority of key stress sources to be controlled: sediment plume > hard matrix removal > heavy metal release. Environmental monitoring and mitigation measures should prioritize the first three types of stress sources.

[0170] (2) Common skeleton nodes (such as Decreased population density The decline in population replenishment rate is a common vulnerable link in all stressor pathways. We should focus on monitoring the ecological processes represented by these nodes, which can simultaneously block the cascading amplification of multiple stressor pathways.

[0171] (3) Source-specific front-end entry node (e.g.) The decrease in filter feeding efficiency corresponds to sediment plume. Metal accumulation (corresponding to heavy metal release) can serve as a diagnostic indicator: when monitoring detects... In case of anomalies, the source can be traced back to the sediment plume pressure; when monitoring detects... In case of an anomaly, the source of the heavy metal release pressure can be traced back to the point where the anomaly occurred.

[0172] (4) The bottleneck analysis results show that for sediment plumes, the focus should be on controlling the cascading amplification process to the population / community level; for heavy metal release, the focus should be on controlling the environmental transformation (metal dissolution) and biological exposure process.

[0173] This invention, through the systematic construction of the SPER four-layer framework, realizes the construction of a complete causal chain from stressors to multi-level ecological consequences, overcoming the shortcomings of the "fragmented" assessment in existing technologies. Through a semi-quantitative assessment system of seven attributes, it fully utilizes existing scientific literature and theoretical inferences, enabling systematic causal relationship assessment even under data-scarce conditions. By quantifying the minimum risk value of pathways and bottleneck analysis, the system identifies the limiting links of different stressors in the causal transmission chain, providing a scientific basis for the design of targeted mitigation measures. Through the ranking of source node importance and the classification and identification of key nodes, it reveals the differentiated mechanisms of action of different stressors in the formation of cumulative effects, providing direct technical support for the formulation of differentiated environmental management strategies.

[0174] The table below compares the differences between this invention and existing deep-sea mining environmental impact assessment methods from multiple perspectives: Table 2 Technical Comparison Table

[0175] Example 2 This embodiment demonstrates the application of the method of the present invention in supporting the selection of mining technology solutions. Two alternative mining technology solutions are set up—Solution A (benchmark solution) and Solution B (low-disturbance solution). The method of the present invention is used to evaluate the difference in the cumulative effects of the two solutions, providing decision support for the selection of the optimal technology solution.

[0176] Scheme A (benchmark scheme) parameters: ore collecting machine operating speed 0.5 m / s, operating width 15 m, tailings discharge depth 1000 m, no sediment plume suppression measures.

[0177] Scheme B (low-disturbance scheme) parameters: ore collecting machine operating speed 0.3 m / s, operating width 10 m, tailings discharge depth 2000 m, equipped with a plume suppression hood (which can reduce the initial concentration of the near-bottom plume by about 50%).

[0178] Steps A1-A2 were implemented: The nodal systems of the two schemes were the same, but the parameter values ​​for S1 (sediment plume) and S2 (hard matrix removal) were different. Scheme B reduced the plume flow rate by approximately 40% and the hard matrix removal rate by approximately 33%.

[0179] Step A3 Implementation – Parameter Adjustment: For Scheme B, Importance of parent nodes The probability of causality was adjusted from 3 to 2 (reducing the spatial scope and temporal duration of the impact). Remain unchanged. Sensitivity of some downstream links. This may also be adjusted due to a decrease in exposure concentration. Based on the adjusted link attributes, the full path risk of Option B is recalculated. High-risk paths in Option B ( ≥4) The number is reduced by about 35% compared to Plan A, and the proportion of medium-risk and below paths increases accordingly.

[0180] Step A4 Implementation – Stability Analysis of Key Nodes: The common skeleton nodes and bridge / terminal amplifier nodes identified by the two schemes are basically the same ( (etc.), indicating that these nodes represent inherent vulnerabilities at the system level, which do not change with specific mining schemes. The source-specific front-end entry node changes: in Scheme B... (Decreased filter feeding efficiency) reduces source specificity (due to more pathways) The risk level has been reduced to below medium risk, indicating that low-perturbation schemes can reduce the source identification characteristics of certain paths.

[0181] Step A5 Implementation – Importance Ranking Comparison: The ranking of Option A is as follows: In the sorting of scheme B and The score gap narrowed ( The score decrease was more pronounced, but the overall ranking remained unchanged. This indicates that reducing plume intensity can simultaneously reduce... The relative importance and absolute risk contribution, but The primary source of stress remains the focus.

[0182] Step A6 Implementation – Bottleneck Feature Comparison: In Solution A The main bottleneck is (Subsequent amplification bottleneck, accounting for 57.7%); S1 in Scheme B The proportion decreased, while The proportion of bottlenecks in achieving the effect has increased. Analysis conclusion: Reducing the plume intensity makes "whether the initial effect can be triggered" a more prominent limiting factor, while the relative importance of "effect amplification" has decreased. This finding has guiding significance for the design of mitigation measures—under Option B, strengthening the monitoring of the initial effect is more critical than controlling cascade amplification.

[0183] Management Decision Support: Based on the above analysis, the following decision recommendations can be made: Scheme B is superior to Scheme A in terms of cumulative effect control and reduces high-risk pathways; the primary source of concern for both schemes is sediment plume, but the relative advantage of plume is weakened in Scheme B, and the relative importance of heavy metal release is increased; common framework nodes (such as population density decline and population replenishment rate decline) are common key monitoring indicators under both schemes.

[0184] Example 3 This embodiment demonstrates the application of the method of the present invention in optimizing the design of environmental monitoring schemes.

[0185] Monitoring Needs Background: Environmental monitoring in deep-sea mining is extremely costly, with a single CCZ (Central Control Zone) survey cruise costing millions of dollars, and is subject to limitations imposed by weather windows and equipment conditions. Therefore, it is necessary to design the optimal monitoring scheme under limited monitoring resources to maximize the amount of information obtained from monitoring efforts.

[0186] Monitoring optimization based on key node identification: Based on the three types of key nodes identified in step A4, a hierarchical monitoring scheme is designed: First layer – Common skeleton node monitoring (core monitoring layer): for (Population density decline) (Population replenishment rate decreased) Long-term, high-frequency monitoring of common skeletal nodes such as (local extinction) will be conducted. Monitoring methods include: repeated photographic surveys using autonomous underwater vehicles (AUVs) and remotely operated underwater vehicles (ROVs), combined with image analysis techniques to estimate population density and age structure changes; and collecting benthic organism samples at fixed stations for species identification and population parameter analysis. The recommended monitoring frequency is to establish a baseline 3 years before mining, monitor annually during mining, and then every 2 years after mining ceases, continuing for at least 20 years.

[0187] The second layer—source-specific front-end entry node monitoring (diagnostic monitoring layer): for... (Decreased filter efficiency) (Metal accumulation) Targeted monitoring is performed on source-specific nodes such as (visual impairment). When the first layer of monitoring detects an abnormality, the second layer of monitoring is activated to diagnose the main stressor type. Monitoring methods include: collecting representative species samples and analyzing tissue metal content (targeted). ), Deploy photosensitive sensors and camera systems at the edge of the illuminated area to observe biological behavioral responses (targeting ).

[0188] The third layer—bridge / terminal amplifier node monitoring (early warning monitoring layer): for... (Population replenishment rate decreased) Key monitoring nodes, such as those showing changes in food web structure, are crucial. Changes in the state of these nodes serve as early warning signals of risks escalating from local to systemic. Monitoring methods include: conducting reproductive biology surveys (gonadal development index, reproductive output, etc.) and stable isotope analysis (food web structure changes) on key species.

[0189] Expected results of the monitoring scheme: Compared with the conventional "full coverage and equalization" monitoring scheme, the hierarchical monitoring scheme based on key node identification can reduce monitoring costs by about 40% (reducing the monitoring frequency of non-critical areas) and increase the capture rate of key information by about 30% (concentrating resources on the nodes with the highest information content).

[0190] Based on the above three embodiments, compared with existing deep-sea mining environmental impact assessment technologies, the present invention has the following beneficial effects: Benefit 1: It realizes the construction of a complete causal chain from multiple pressure sources to multi-level ecological consequences.

[0191] Existing inventory and matrix methods can only describe the two-dimensional relationship between a single activity and a single environmental factor, failing to construct a complete causal chain from stressor generation to ecological consequences. This invention, through the SPER four-layer framework, systematically identifies 6 types of stressors, 15 types of transmission processes, 12 types of exposure pathways, and 23 types of response nodes. It constructs a complete causal network covering all stages from "source → transmission → exposure → individual response → population consequences → community / ecosystem consequences" through directed links, overcoming the fundamental deficiency of fragmented assessments in existing technologies. Managers can intuitively track all possible paths from any stressor to any ecological consequence through this network, achieving a systematic and comprehensive environmental impact assessment.

[0192] Benefit 2: It provides a semi-quantitative evaluation method that is feasible under conditions of data scarcity.

[0193] Existing spatial analysis methods and life-cycle assessment methods rely heavily on high-precision data, making them difficult to implement effectively in the data-scarce deep-sea environment. This invention utilizes a semi-quantitative assessment system based on seven attributes (Importance Level 4, Sensitivity Level 4, Probability Level 5, Comprehensive Strength Product, Risk Matrix, Evidence Level 3, and Uncertainty Level 3) to fully leverage existing scientific literature, theoretical inferences, and expert knowledge, enabling systematic causal relationship assessments even under low-data conditions. The explicit labeling of evidence levels and uncertainty levels clearly communicates the confidence level and limitations of the assessment results, avoiding pseudo-precision issues.

[0194] Benefit 3: It identified the critical paths and limiting factors that determine the long-term behavior of the system.

[0195] Existing technologies typically only provide a qualitative assessment that "a certain pressure source has an impact," failing to answer critical management questions such as "through which pathways is the impact transmitted" and "which links are limiting." This invention, through path minimum risk value quantification and four types of bottleneck analysis (source formation bottleneck, exposure realization bottleneck, effect realization bottleneck, and subsequent amplification bottleneck), systematically identifies the limiting links in the causal transmission chain of different pressure sources, providing a scientific basis for designing targeted mitigation measures. For example, for sediment plumes, the focus should be on controlling their cascading amplification to the population / community level, while for heavy metal releases, the focus should be on controlling their environmental transformation and biological exposure processes.

[0196] Benefit 4: It reveals the differentiated mechanisms of action of different stressors, supporting differentiated management and control.

[0197] This invention reveals the differentiated mechanisms by which different pressure sources contribute to the cumulative effect by ranking the importance of source nodes and classifying key nodes (common framework nodes, source-specific front-end entry nodes, and bridge / terminal amplifier nodes). The importance ranking (sediment plume > hard matrix removal > heavy metal release > artificial light > DIC release > underwater noise) clarifies the priority of control measures; the key node identification results reveal the intrinsic mechanism by which multiple pressure sources generate cumulative effects through a common framework. This information provides direct technical support for developing differentiated environmental management strategies (such as prioritizing key pressure sources, selecting key monitoring indicators, and designing mitigation measures).

[0198] Benefit 5: Supports the selection of mining technology solutions and the optimization of monitoring solutions.

[0199] As shown in Embodiments 2 and 3, the method of the present invention can assess the differences in environmental impact of different mining technology schemes by adjusting source node parameters, providing a quantitative comparative basis for the selection of technology schemes. Simultaneously, by identifying key nodes, the allocation of monitoring resources can be optimized, reducing monitoring costs while increasing the capture rate of key information. These functions significantly enhance the practical value of environmental impact assessment in management decision-making.

[0200] Example 4 Please see Figure 6 A system for constructing and evaluating causal networks of environmental impacts based on the SPER framework, and based on the method described in one of the embodiments, includes: The node identification module is used to construct the SPER four-layer framework, including: a source layer, which identifies multiple types of environmental stressors as source nodes; a transmission layer, which identifies multiple transmission processes of the stressors in the environmental medium as transmission nodes; an exposure layer, which identifies multiple exposure pathways of organisms to the stressors or their effects as exposure nodes; and a response layer, which identifies multiple effect nodes exhibited by the organisms after exposure, from the individual to the ecosystem level as effect nodes; and directed links are used to represent the causal relationships between the nodes, forming a four-layer causal network. The attribute assignment module is used to assign seven attributes to each directed link: the importance of the parent node. Characterizes the spatial and temporal importance of the influence of the cause on the effect; child node sensitivity. Characterizes the sensitivity of the outcome to changes in the cause; the probability of causal event occurrence P; and the overall strength of the effect. Risk level Level of evidence The level of uncertainty characterizes the sufficiency of scientific evidence. Characterizes the main types of sources of uncertainty; The path analysis module is used to traverse directed links from each source node, constructing a complete causal path from the source node through the transit and exposure nodes to the terminal effect node in the direction of source node → transit node → exposure node → effect node; and minimizing the path risk value. Quantify the overall risk level of each path, among which, For the first on the path The node and the first Risk level of links between nodes; The critical node identification module is used to identify three types of critical nodes based on the frequency of occurrence, location distribution, and risk level of preceding and following links of each node in the complete causal path: common skeleton nodes, which are nodes that appear repeatedly in multiple paths from different source nodes and are located in the cascade segment of effect nodes; source-specific front-end entry nodes, which are nodes that appear only as initial effect nodes in specific source node paths; and bridge / terminal amplifier nodes, which are nodes that receive low-level response inputs and output high-level responses with output risk higher than input risk.

[0201] In this embodiment, in order to better utilize the method described in one of the embodiments, this application proposes an environmental impact causal network construction and evaluation system based on the SPER framework. Each module corresponds to each step of the above method, and its specific principles have been described above and will not be repeated here.

[0202] Example 5 An environmental impact causal network construction and evaluation device based on the SPER framework includes: at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described in one of the embodiments.

[0203] In this embodiment, to better run and process the method described in one of the embodiments, the above method is stored in a memory, and the stored method is executed using a processor. It should be noted that the principle and effect of each step have been described above and will not be elaborated further here.

[0204] Example 6 A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in one of the embodiments.

[0205] In this embodiment, to better operate and use the method described in one of the embodiments, the above method is stored in a computer-readable storage medium, and the above method is implemented using a processor. It should be noted that the principle and effect of each step have been described above and will not be elaborated further here.

[0206] The above description is only a part of the embodiments of the present invention and does not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for constructing and evaluating causal networks of environmental impacts based on the SPER framework, characterized in that, The method includes: A1. Construct a four-layer SPER framework, including: a source layer, identifying multiple types of environmental stressors as source nodes; a transmission layer, identifying multiple transmission processes of the stressors in the environmental medium as transmission nodes; an exposure layer, identifying multiple exposure pathways of organisms to the stressors or their effects as exposure nodes; and a response layer, identifying multiple effect nodes exhibited by organisms from the individual to the ecosystem level after exposure as effect nodes; and use directed links to represent the causal relationships between the nodes, forming a four-layer causal network. A2. Assign seven attributes to each directed link: importance of the parent node. Characterizes the spatial and temporal importance of the influence of the cause on the effect; child node sensitivity. Characterizes the sensitivity of the outcome to changes in the cause; the probability of causal event P; and the overall strength of the effect. Risk level Level of evidence The level of uncertainty characterizes the sufficiency of scientific evidence. Characterizes the main types of sources of uncertainty; A3. Starting from each source node, traverse along the directed links, constructing a complete causal path from the source node through the transit and exposure nodes to the terminal effect node in the direction of source node → transit node → exposure node → effect node; select the path with the minimum risk value. Quantify the overall risk level of each path, among which, For the first on the path The node and the first Risk level of links between nodes; S4. Based on the frequency of occurrence, location distribution, and risk level of preceding and following links of each node in the complete causal path, three types of key nodes are identified: common skeleton nodes, which are nodes that appear repeatedly in multiple paths from different source nodes and are located in the cascade segment of effect nodes; source-specific front-end entry nodes, which are nodes that appear only as initial effect nodes in specific source node paths; and bridge / terminal amplifier nodes, which are nodes that receive low-level response inputs and output high-level responses with output risk higher than input risk.

2. The method for constructing and evaluating environmental impact causal networks based on the SPER framework as described in claim 1, characterized in that, The importance of the parent node and the sensitivity of the child nodes All use a 4-level rating system; the probability of occurrence P of the causal event is classified into 5 levels, including rare, unlikely, possible, very likely, and almost certain.

3. The method for constructing and evaluating environmental impact causal networks based on the SPER framework as described in claim 1, characterized in that, The combined effect intensity Divided into four intervals based on numerical values: For negligible overall strength, For low overall strength, For medium overall strength, For high overall strength; the aforementioned risk level Divided into five levels: rare risk, low risk, medium risk, high risk, and extremely high risk, through... The level is row, The risk matrix is ​​determined by the interval of the column.

4. The method for constructing and evaluating environmental impact causal networks based on the SPER framework as described in claim 1, characterized in that, The level of evidence A three-tier classification system of high, medium, and low is adopted; The uncertainty level use Parameter uncertainty, Model structure uncertainty, A three-level classification system for situational uncertainty.

5. The method for constructing and evaluating environmental impact causal networks based on the SPER framework as described in claim 1, characterized in that, In A4, the common skeleton node is a node that appears in a path of no less than three different source nodes, is located in the cascade segment of effect nodes, and has an average risk level of no less than medium risk for its incoming links. The source-specific front-end entry node is a node that appears as an initial effect node only in the path of one or two source nodes, whose exposure path is directly related to the physical and / or chemical properties of the source node, and whose path from the node converges to the common skeleton node. The bridge / terminal amplifier node is a node that receives low-level response input and outputs high-level response, with the average risk level of outgoing links being higher than the average risk level of incoming links, and the betweenness centrality being higher than a preset threshold.

6. The method for constructing and evaluating environmental impact causal networks based on the SPER framework as described in claim 1, characterized in that, The method also includes A5, which calculates the comprehensive score of each source node based on the complete causal path set of each source node and sorts them from high to low scores; The overall score is calculated using the following formula: in, This is the arithmetic mean of the minimum risk values ​​of all paths to this source node. This represents the percentage of high-risk paths within the source node's path. This represents the total number of paths to the source node.

7. The method for constructing and evaluating environmental impact causal networks based on the SPER framework as described in claim 1, characterized in that, The method also includes A6, identifying the SPER stage containing the link with the lowest risk level on each complete causal path as a bottleneck, and classifying bottlenecks into four categories: source-forming bottlenecks. Located in the stage from the source node to the transit node; exposing implementation bottlenecks. Located in the stage from the transmission node to the exposure node; bottleneck in effect realization. It is located in the stage from the exposure node to the effect node; the subsequent amplification bottleneck. Located in the cascade stage between effect nodes; the frequency and proportion of the four types of bottlenecks in the paths of each source node are statistically analyzed.

8. A system for constructing and evaluating causal networks of environmental impacts based on the SPER framework, characterized in that, The method according to any one of claims 1 to 7 includes: The node identification module is used to construct the SPER four-layer framework, including: a source layer, which identifies multiple types of environmental stressors as source nodes; a transmission layer, which identifies multiple transmission processes of the stressors in the environmental medium as transmission nodes; an exposure layer, which identifies multiple exposure pathways of organisms to the stressors or their effects as exposure nodes; and a response layer, which identifies multiple effect nodes exhibited by the organisms after exposure, from the individual to the ecosystem level as effect nodes; and directed links are used to represent the causal relationships between the nodes, forming a four-layer causal network. The attribute assignment module is used to assign seven attributes to each directed link: the importance of the parent node. Characterizes the spatial and temporal importance of the influence of the cause on the effect; child node sensitivity. Characterizes the sensitivity of the outcome to changes in the cause; the probability of causal event P; and the overall strength of the effect. Risk level Level of evidence The level of uncertainty characterizes the sufficiency of scientific evidence. Characterizes the main types of sources of uncertainty; The path analysis module is used to traverse directed links from each source node, constructing a complete causal path from the source node through the transit and exposure nodes to the terminal effect node in the direction of source node → transit node → exposure node → effect node; and minimizing the path risk value. Quantify the overall risk level of each path, among which, For the first on the path The node and the first Risk level of links between nodes; The critical node identification module is used to identify three types of critical nodes based on the frequency of occurrence, location distribution, and risk level of preceding and following links of each node in the complete causal path: common skeleton nodes, which are nodes that appear repeatedly in multiple paths from different source nodes and are located in the cascade segment of effect nodes; source-specific front-end entry nodes, which are nodes that appear only as initial effect nodes in specific source node paths; and bridge / terminal amplifier nodes, which are nodes that receive low-level response inputs and output high-level responses with output risk higher than input risk.

9. A device for constructing and evaluating environmental impact causal networks based on the SPER framework, characterized in that, include: At least one processor and a memory communicatively connected to said at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images