Data-driven oyster reef ecological restoration effect evaluation method and system

Abstract

The invention discloses a data-driven oyster reef ecological restoration effect evaluation method and system, and relates to the technical field of data processing. The method comprises the following steps: establishing a multi-modal data set; after feature extraction, a feature index set is established, oyster health comprehensive indexes are constructed, and system service indexes are constructed; perturbation observation of the contrast area and the restoration area is executed by using the perturbation event, ecological toughness evaluation is executed by using a perturbation observation result and the multi-modal data set, and an ecological toughness index is established; establishing a structured causal graph, performing attribution analysis of the restoration area and the contrast area, constructing an average processing effect of restoration measures on the oyster health comprehensive index and the system service index, and outputting an oyster reef ecological restoration effect evaluation result in combination with the ecological toughness index after executing correction of the oyster health comprehensive index and the system service index. The technical problem that the oyster reef ecological restoration effect evaluation is inaccurate in the prior art is solved, and the technical effect of improving the restoration effect evaluation accuracy is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a data-driven oyster reef ecological restoration effect evaluation method and system. BACKGROUND

[0002] As a typical coastal ecosystem, oyster reefs have important ecological functions and economic values. They not only provide habitats for a variety of aquatic organisms, but also improve water quality, sequester carbon, and resist coastal erosion. However, due to environmental changes, global oyster reef resources have declined significantly, and ecological restoration has become an important problem that needs to be solved. Existing oyster reef restoration effect evaluation methods rely on single biological or physicochemical indicators, lack comprehensive utilization of multi-source information, and are difficult to fully reflect the health status of oyster individuals, ecosystem services, and ecological resilience in the restoration area, resulting in inaccurate evaluation results and inability to provide reliable basis for optimization of restoration measures. SUMMARY

[0003] The present application provides a data-driven oyster reef ecological restoration effect evaluation method and system, which solves the technical problem of inaccurate oyster reef ecological restoration effect evaluation in the prior art.

[0004] In a first aspect, the present application provides a data-driven oyster reef ecological restoration effect evaluation method, which comprises: After configuring the control area and the restoration area, biological data, physicochemical data, acoustic fingerprint data, image data, and microbiome data of oyster individuals in the oyster reef are collected to establish a multi-modal data set. After feature extraction of the multi-modal data set, a feature index set is established, and an oyster health comprehensive index is constructed using the feature index set, and a system service index is constructed. A disturbance event is configured, and disturbance observation of the control area and the restoration area is performed using the disturbance event. Ecological resilience evaluation is performed using the disturbance observation results and the multi-modal data set, and an ecological resilience index is established. The oyster health comprehensive index and the system service index are used as result variables, and the multi-modal data set is used as causal input to establish a structured causal diagram. Based on the structured causal diagram, attribution analysis of the restoration area and the control area is performed, the average treatment effect of the restoration measures on the oyster health comprehensive index and the system service index is constructed, and after correction of the oyster health comprehensive index and the system service index based on the average treatment effect, the ecological resilience index is output to output the oyster reef ecological restoration effect evaluation result.

[0005] In a second aspect, the present application provides a data-driven oyster reef ecological restoration effect evaluation system, which comprises: The data acquisition module: after the configuration of the control area and the repair area, biological data, physicochemical data, acoustic fingerprint data, image data, and microbiome data of oyster individuals in the oyster reef are collected to establish a multi-modal data set; the feature extraction module: after feature extraction of the multi-modal data set, a feature index set is established, and the oyster health comprehensive index is constructed using the feature index set, and a system service index is constructed; the resilience evaluation module: a disturbance event is configured, and disturbance observation of the control area and the repair area is performed using the disturbance event, and ecological resilience evaluation is performed using the disturbance observation results and the multi-modal data set to establish an ecological resilience index; the attribution analysis module: the oyster health comprehensive index and the system service index are taken as result variables, and the multi-modal data set is taken as causal input to establish a structured causal diagram, and attribution analysis of the repair area and the control area is performed based on the structured causal diagram, and the average treatment effect of the repair measures on the oyster health comprehensive index and the system service index is constructed, and after correction of the oyster health comprehensive index and the system service index based on the average treatment effect, the ecological resilience index is combined to output the oyster reef ecological restoration effect evaluation result.

[0006] One or more technical solutions provided in the present application have at least the following technical effects or advantages: After the configuration of the control area and the repair area, biological data, physicochemical data, acoustic fingerprint data, image data, and microbiome data of oyster individuals in the oyster reef are collected to establish a multi-modal data set; after feature extraction of the multi-modal data set, a feature index set is established, and the oyster health comprehensive index is constructed using the feature index set, and a system service index is constructed. Then, a disturbance event is configured, and disturbance observation of the control area and the repair area is performed using the disturbance event, and ecological resilience evaluation is performed using the disturbance observation results and the multi-modal data set to establish an ecological resilience index. Finally, the oyster health comprehensive index and the system service index are taken as result variables, and the multi-modal data set is taken as causal input to establish a structured causal diagram, and attribution analysis of the repair area and the control area is performed based on the structured causal diagram, and the average treatment effect of the repair measures on the oyster health comprehensive index and the system service index is constructed, and after correction of the oyster health comprehensive index and the system service index based on the average treatment effect, the ecological resilience index is combined to output the oyster reef ecological restoration effect evaluation result. The technical problem of inaccurate oyster reef ecological restoration effect evaluation in the prior art is solved, and the technical effect of improving the accuracy of oyster reef ecological restoration effect evaluation by combining multi-modal data driving, causal attribution analysis, and ecological resilience evaluation is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0008] Figure 1 A schematic diagram of a data-driven method for evaluating the effectiveness of oyster reef ecological restoration, provided in an embodiment of this application; Figure 2 This is a schematic diagram of a data-driven oyster reef ecological restoration effect evaluation system provided in an embodiment of this application.

[0009] Figure labeling: Data acquisition module 11, feature extraction module 12, resilience assessment module 13, attribution analysis module 14. Detailed Implementation

[0010] This application provides a data-driven method and system for evaluating the effectiveness of oyster reef ecological restoration, which solves the technical problem of inaccurate evaluation of oyster reef ecological restoration effectiveness in the prior art.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0013] Example 1, as Figure 1 As shown, this application provides a data-driven method for evaluating the effectiveness of oyster reef ecological restoration, wherein the method includes: After configuring the control and restoration areas, biological, physicochemical, acoustic, image, and microbiome data of individual oysters in the oyster reef were collected to establish a multimodal dataset.

[0014] Further, the biological data includes oyster individual density, survival rate, shell size distribution, the physicochemical data includes water temperature, salinity, pH value, dissolved oxygen, nutrient salt concentration, the voiceprint data includes voiceprint data of multi-channel underwater acoustic signal, the image data includes data of reef coverage and crack information based on underwater image recognition, and the microbiome data includes data of microbial diversity and functional gene abundance obtained by high-throughput sequencing.

[0015] In the embodiment of the present application, the control area and the repair area are reasonably arranged in the sea area to be evaluated, wherein the control area is a natural oyster reef or a degraded reef area without artificial repair measures intervention, and the repair area is an oyster reef area where ecological restoration engineering measures have been implemented. Then, multi-source data collection work is carried out in the control area and the repair area simultaneously, including: collecting biological data of oyster individuals, specifically, using diving sampling and fixed-point monitoring to count oyster individual density, survival rate, shell size distribution, shell growth rate and the like; collecting physicochemical data, specifically, recording continuous or timed water physicochemical indexes such as water temperature, salinity, pH value, dissolved oxygen content, nutrient salt concentration by laying multi-parameter water quality monitors and automatic samplers; collecting voiceprint data, specifically, laying multi-channel underwater acoustic collection devices around the reef to record long-time voiceprint signals of oyster filter-feeding behavior and environmental background sound; collecting image data, specifically, obtaining image information such as reef surface coverage, crack information, and attached object distribution by underwater cameras, ROVs (remotely operated vehicles) or structured light imaging equipment; collecting microbiome data, specifically, collecting water samples, sediment samples and biological samples on oyster individuals and reef surfaces, and obtaining microbial diversity, community composition and functional gene abundance data by high-throughput sequencing technology. Through the acquisition of the above multi-source data, comprehensive data support covering oyster individuals, environmental conditions, community activity, structural characteristics and microbial functions is formed, thereby providing sufficient basis for subsequent feature extraction and repair effect evaluation.

[0016] After feature extraction is performed on the multi-modal data set, a feature index set is established, the oyster health comprehensive index is constructed by using the feature index set, and system service class indexes are constructed.

[0017] By feature extraction on the multi-modal dataset, a feature index set is established, such as extracting key physiological characteristics of oyster individual survival rate, density change rate and shell growth rate from biological data, extracting environmental characteristics of dissolved oxygen fluctuation rate and nutrient salt concentration change trend from physicochemical data, extracting dynamic behavior characteristics of frequency band energy distribution and filter-related acoustic signal ratio from acoustic data, extracting structural characteristics of reef coverage, crack rate and attached object distribution from image data, and extracting community function characteristics of community diversity index and functional gene abundance index from microbiome data. On this basis, the oyster health comprehensive index is constructed by weighted fusion of different dimensional features, which is used to quantitatively represent the overall health level of oyster individuals and communities; at the same time, the covariance relationship between the oyster health comprehensive index and the physicochemical environmental factors, reef structure characteristics and microbial function characteristics is combined to construct system service indexes, which are used to quantify the system ecological service effects of oyster reefs in water purification, sediment stability and carbon sink functions.

[0018] Further, after feature extraction on the multi-modal dataset, a feature index set is established, and the oyster health comprehensive index is constructed using the feature index set, and the system service indexes are constructed, including: After modality layering of the multi-modal dataset, independent normalization processing is performed respectively to establish a layered feature subset; principal component analysis of the layered feature subset is performed to determine the contribution factor of the layered feature factor, and the layered feature subset is weighted based on the contribution factor; the survival rate, shell growth rate and community density in the biological subset in the layered feature subset are extracted as the first health dimension; the biological diversity index and functional gene richness in the microbiome subset are extracted as the second health dimension; the frequency band complexity and typical filter sound signal ratio in the acoustic subset are extracted as the third health dimension; weighted fusion of the first health dimension, the second health dimension and the third health dimension is performed according to the weighted multi-modal coupling weight to establish the oyster health comprehensive index.

[0019] After feature extraction on the multi-modal dataset, the multi-modal dataset is processed in a modal hierarchical manner, and is divided into biological subsets, microbiome subsets, voiceprint subsets, and other different dimensional data groups; independent normalization processing is performed on each hierarchical subset to eliminate differences in different dimensions and measurement scales, and standardized hierarchical feature subsets are obtained. Subsequently, principal component analysis is performed on each hierarchical feature subset to determine the contribution rate of the key feature factor, and the hierarchical feature subsets are weighted based on the contribution rate to obtain a reasonable multi-modal weight distribution. On this basis, the survival rate of oysters, shell growth rate and community density indicators are extracted from the biological subsets as the first health dimension, the microbial diversity index and functional gene richness index are extracted from the microbiome subsets as the second health dimension, and the frequency band complexity and typical filter sound signal ratio indicators are extracted from the voiceprint subsets as the third health dimension. Finally, the first health dimension, the second health dimension and the third health dimension are weighted and fused according to the aforementioned multi-modal coupling weight to establish an oyster health comprehensive index that can comprehensively reflect the survival state of oyster individuals, community activity and microbial functional support.

[0020] Further, after establishing the oyster health comprehensive index, it includes: The nitrogen and phosphorus removal efficiency change rate index in the physicochemical subset is extracted, and the filter-feeding efficiency correlation factor in the oyster health comprehensive index is obtained, and the water quality purification efficiency index is established according to the nitrogen and phosphorus removal efficiency change rate index and the filter-feeding efficiency correlation factor; the matrix coverage and crack rate indexes in the image subset are extracted, and the community density factor in the oyster health comprehensive index is obtained, and the sediment stability index is established according to the matrix coverage and crack rate indexes and the community density factor; the carbon cycle functional gene abundance in the microbiome subset is extracted, and the community density factor in the oyster health comprehensive index is obtained, and the carbon sink rate index is established; the water quality purification efficiency index, the sediment stability index, the carbon sink rate index and the oyster health comprehensive index are subjected to covariance analysis to quantify the coupling degree of oyster individual health and system service function, and the system service index is established.

[0021] Based on the multi-modal dataset and the oyster health comprehensive index, the system service type index is constructed. Specifically, for the physicochemical subset, the nitrogen and phosphorus concentration data collected by the continuous water quality monitor are combined with the flow rate of the inlet and outlet sections of the repair area and the control area to calculate the nitrogen and phosphorus removal efficiency change rate. Based on the filter efficiency correlation factor (estimated by the voiceprint signal and individual density) in the oyster health comprehensive index, a linear regression model is established to generate the water purification efficiency index. Secondly, for the image subset, the matrix coverage rate and crack rate characteristic values of the reef surface are extracted by applying the underwater image segmentation algorithm (such as the convolutional neural network image segmentation model), and combined with the community density factor in the oyster health comprehensive index, the sediment stability index is established by using the multiple regression method to quantify the inhibition of sediment resuspension by the oyster reef structure. Thirdly, for the microbiome subset, the functional gene annotation results of high-throughput sequencing data are used to screen the functional gene abundance related to carbon cycle (such as carbon fixation, methane metabolism, and cellulose degradation related genes), and combined with the community density factor in the oyster health comprehensive index, the carbon sink rate index is calculated by using the weighted index method. Finally, the covariance analysis method is used to statistically couple the water purification efficiency index, the sediment stability index and the carbon sink rate index with the oyster health comprehensive index to obtain the coupling degree value of the oyster individual health and the system service function, and then the oyster reef system service type index is established to realize the comprehensive evaluation of the restoration effect.

[0022] The disturbance event is configured, the disturbance observation of the control area and the repair area is performed by using the disturbance event, the ecological resilience evaluation is performed by using the disturbance observation result and the multi-modal dataset, and the ecological resilience index is established.

[0023] Further, the disturbance event includes a biological invasion event, a pollutant instantaneous release time and an extreme temperature change event.

[0024] In the embodiments of the present application, controllable disturbance events are configured in the control area and the repair area to simulate extreme scenarios in real environments and perform ecological resilience evaluation. Specifically, the disturbance events include biological invasion events, pollutant instantaneous release events, and extreme temperature change events. The biological invasion event simulates the competitive exclusion effect by artificially releasing a certain density of invasive shellfish or crustacean individuals into the experimental water area; the pollutant instantaneous release event simulates the outbreak of red tide or land source pollution by adding a high concentration of nutrient salt solution or a short time high load of suspended particles to the target sea area; the extreme temperature change event simulates the scenario of extreme high temperature in summer or extreme low temperature in winter by arranging controllable heating or cooling devices in the local area to rapidly increase or decrease the water temperature in a short period of time. During the execution of the disturbance, a multi-chronological observation window is configured, and the physiological state of oyster individuals, environmental physicochemical parameters, community acoustic fingerprint characteristics, and microbial community changes before, during, and after the disturbance are dynamically recorded by using the arranged water quality online monitoring instrument, acoustic monitoring equipment, image acquisition device, and microbial sampling tool. Subsequently, the collected disturbance observation results are compared and analyzed with the baseline multi-modal data set to calculate the disturbance resistance index (the decline amplitude of the health index under the disturbance intensity), the recovery index (the recovery speed of the health index after the disturbance), and the super-compensation index (whether it exceeds the pre-disturbance level after the disturbance), and the ecological resilience index is established by a weighted comprehensive method for quantitatively representing the ecological stability and recovery ability of oyster reefs under different disturbance scenarios.

[0025] Further, the disturbance events are configured, the disturbance observation of the control area and the repair area is performed by using the disturbance events, the ecological resilience evaluation is performed by using the disturbance observation results and the multi-modal data set, and the ecological resilience index is established, including: Controllable disturbance events are set in the repair area and the control area, the controllable disturbance events include short-term nutrient salt pulse, sediment load disturbance, acoustic interference, and physical disturbance, and the disturbance intensity of the controllable disturbance events is set based on the statistical distribution of historical extreme environmental events; after the disturbance is performed, a multi-chronological observation window is configured to perform the disturbance observation of the control area and the repair area, and the disturbance observation results are established; the multi-modal data set is taken as the baseline data, the disturbance observation results are taken as the disturbance results, the disturbance resistance index, the recovery index, and the super-compensation index are extracted, and the ecological resilience index is established.

[0026] The controllable disturbance events are arranged in the repair area and the control area, including short-term nutrient salt pulse, sediment load disturbance, acoustic interference and physical disturbance. The disturbance intensity is determined based on the statistical distribution of the target sea area historical extreme environmental events, so as to ensure that the disturbance condition has environmental authenticity and representativeness. Among them, the short-term nutrient salt pulse simulates the risk of red tide outbreak by artificially adding high-concentration nitrogen and phosphorus nutrient salt solution; the sediment load disturbance simulates storm runoff or seabed disturbance by suspending sediment particles in the water body; the acoustic interference simulates ship or construction noise by playing high-intensity low-frequency noise through underwater speakers; and the physical disturbance simulates typhoon surge or trawl damage by mechanical dragging or local scouring. After the disturbance is implemented, in the multi-sequential observation window before, during and after the disturbance, the multi-parameter water quality probe, acoustic monitoring array, underwater camera system and microbial sampling tool arranged are used to continuously monitor the oyster individual state, environmental physicochemical condition, community acoustic fingerprint characteristics and microbial community structure, to form the disturbance observation results. Subsequently, the multi-modal data set before the disturbance is taken as the base state data, and the disturbance observation data is taken as the disturbance result, the health index change curve and the time sequence are constructed, the disturbance resistance index (index decline amplitude), the recovery index (recovery speed) and the super-compensation index (whether exceeding the base state level after recovery) of the oyster reef under the disturbance are calculated, and the ecological resilience index is established based on the weighted fusion method, so as to realize the quantitative evaluation of the ecological stability and recovery force of the repair area and the control area under different disturbance conditions.

[0027] The oyster health comprehensive index and the system service index are taken as the result variable, and the multi-modal data set is taken as the causal input, a structured causal diagram is established, and the attribution analysis of the repair area and the control area is carried out based on the structured causal diagram, the average treatment effect of the repair measures on the oyster health comprehensive index and the system service index is constructed, and the oyster health comprehensive index and the system service index are corrected based on the average treatment effect, and the ecological resilience index is output to output the oyster reef ecological restoration effect evaluation result.

[0028] The health comprehensive index of oysters and the system service class index are taken as result variables, and the multi-modal data set is taken as causal input variables. A structured causal graph is established by using a causal modeling method. Specifically, first, biological, physical and chemical, voiceprint, image, and microbiome characteristics in the multi-modal data set are combined with spatial information of the repair area and the control area to construct a node set. A structured learning algorithm such as a constraint-based PC algorithm or a score-based GES algorithm is used to identify the causal edge relationship between variables, and a causal network structure of oyster reef repair is formed. Then, parameter learning is performed on the causal graph to obtain the probability strength of each causal path. On this basis, a control analysis is performed on the repair area and the control area based on a causal inference framework such as a latent outcome model, and the average treatment effect (ATE) of the repair measures on the health comprehensive index of oysters and the system service class index is calculated, which is used to quantitatively identify the direct contribution of the repair measures. Further, the average treatment effect is corresponded to the health comprehensive index of oysters and the system service class index, and the corrected health index and service index values are obtained through error elimination and normalization correction under causal driving. Finally, the corrected health comprehensive index of oysters, the system service class index, and the ecological resilience index are jointly analyzed to comprehensively output the oyster reef ecological repair effect evaluation results, and the causal attribution, quantitative evaluation, and scientific judgment of the repair measures are realized.

[0029] The corrected health comprehensive index of oysters, the system service class index, and the ecological resilience index are jointly analyzed. Specifically, time series alignment and scale normalization processing are performed on the three types of indexes to ensure that the data from different sources are comparable under the same analysis framework. A multivariate statistical modeling method such as principal component analysis or a covariance structure model or a machine learning fusion algorithm such as random forest regression or Bayesian network fusion is used to jointly model the corrected health comprehensive index of oysters, the system service class index, and the ecological resilience index, identify the coupling relationship and comprehensive contribution degree among them, and generate a comprehensive evaluation score through weighted fusion. The weights can be allocated according to the contribution of the index to the explained variance or expert experience. The comprehensive evaluation score is taken as the output of the oyster reef ecological repair effect evaluation results, and the comprehensive quantitative judgment of the overall ecological health status, system service function, and disturbance resistance ability of the repair area is realized.

[0030] Further, based on the structured causal graph, an attribution analysis of the repair area and the control area is performed, and the average treatment effect of the repair measures on the health comprehensive index of oysters and the system service class index is constructed, including: Based on the average treatment effect channel, an average treatment effect control analysis of the repair area and the control area is performed, and the sensitivity disturbance identification of the average treatment effect control analysis result is performed to generate the average treatment effect.

[0031] Based on the established structured causal diagram, causal attribution analysis is performed on the data of the restoration area and the control area to construct the average treatment effect of the restoration measures on the oyster health comprehensive index and the system service index. Specifically, based on the average treatment effect channel, the multi-modal data of the restoration area and the control area are input into the causal inference model, and the latent outcome method or the propensity score matching method is used to control the analysis of the differences between the restoration area and the control area in the oyster health comprehensive index and the system service index, and the average treatment effect value of the restoration measures is obtained. Subsequently, the results of the average treatment effect control analysis are subjected to sensitivity perturbation identification, that is, by perturbing the key input variables (such as oyster density, nutrient salt concentration, acoustic fingerprint complexity, etc.), the sensitivity change range of the result index is detected to identify the possible confounding bias and uncertainty in the causal inference. Finally, the average treatment effect is corrected and verified according to the sensitivity perturbation identification results, and the average treatment effect value after robust correction is generated.

[0032] Further, based on the average treatment effect, the oyster health comprehensive index and the system service index are corrected, including: The average treatment effect is established in correspondence with the oyster health comprehensive index and the system service index; after the false judgment elimination under causal driving is performed using the correspondence, the index normalization processing is performed, and the oyster reef ecological restoration effect evaluation result is generated using the index normalization processing result.

[0033] After obtaining the average treatment effect of the restoration measures on the oyster health comprehensive index and the system service index, the original index is corrected based on the average effect result to improve the accuracy of the evaluation result. Specifically, the average treatment effect is established in correspondence with the oyster health comprehensive index and the system service index, and the offset direction and offset amplitude of the restoration measures on different indexes are determined by establishing a causal mapping function; the original index value is subjected to false judgment elimination under causal driving using the correspondence, that is, by correcting the false correlation caused by confounding variables, the non-real effect caused by environmental fluctuations or random errors is eliminated; after the false judgment elimination is completed, the corrected oyster health comprehensive index and system service index are subjected to normalization processing, so that they are quantitatively compared and fused in a unified scale; finally, the normalized index data are used for joint analysis with the ecological resilience index to generate the oyster reef ecological restoration effect evaluation result.

[0034] Further, after the oyster reef ecological restoration effect evaluation result is output in combination with the ecological resilience index, including: establish a mapping between the repair measures and the average treatment effect, establish a mapping data set; use the mapping data set as matching data to perform matching anomaly identification of the expert database, and establish a matching anomaly signal; perform attention verification on the control area and the repair area according to the matching anomaly signal, and update the attention verification result to the mapping data set to perform subsequent oyster reef ecological restoration management.

[0035] Specifically, a mapping relationship between repair measures and average treatment effects is established, different types of repair measures (such as artificial reef deployment, oyster seed supplementation, environmental regulation, etc.) are one-to-one corresponding to their corresponding average treatment effect results, and a mapping data set is formed; the mapping data set is input into the expert knowledge database as matching data, a matching anomaly identification mechanism based on a rule engine or a machine learning algorithm is called, a standard experience result or a historical evaluation sample is compared, possible abnormal deviations are identified, and a matching anomaly signal is output; after detecting the matching anomaly signal, attention verification is further performed on the repair area and the control area, that is, suspicious indicators (such as sudden drop of oyster density, abnormal fluctuation of nutrient salt, etc.) are verified by supplementing sampling or intensive monitoring; finally, the result of the attention verification is fed back and updated to the mapping data set, and the relationship between the repair measures and the effects is corrected and improved to support the dynamic management of subsequent oyster reef ecological restoration.

[0036] To sum up, the embodiments of the present application have at least the following technical effects: After the control area and the repair area are configured, biological data, physicochemical data, acoustic print data, image data, and microbiome data of oyster individuals in the oyster reef are collected, and a multi-modal data set is established; after feature extraction is performed on the multi-modal data set, a feature index set is established, an oyster health comprehensive index is constructed using the feature index set, and a system service index is constructed. Then, a disturbance event is configured, and disturbance observation of the control area and the repair area is performed using the disturbance event, and ecological resilience evaluation is performed using the disturbance observation result and the multi-modal data set, and an ecological resilience index is established. Finally, the oyster health comprehensive index and the system service index are used as result variables, the multi-modal data set is used as causal input, a structured causal diagram is established, attribution analysis of the repair area and the control area is performed based on the structured causal diagram, average treatment effects of the repair measures on the oyster health comprehensive index and the system service index are constructed, and after the oyster health comprehensive index and the system service index are corrected based on the average treatment effects, the ecological resilience index is output to output the oyster reef ecological restoration effect evaluation result. The technical problem of inaccurate oyster reef ecological restoration effect evaluation in the prior art is solved, and the technical effect of improving the accuracy of oyster reef ecological restoration effect evaluation by combining multi-modal data driving, causal attribution analysis, and ecological resilience evaluation is achieved.

[0037] Embodiment two, based on the same inventive concept as the data-driven oyster reef ecological restoration effect evaluation method in the preceding embodiments, such asFigure 2 As shown, the present application provides a data-driven oyster reef ecological restoration effect evaluation system, wherein the system comprises: The data acquisition module 11: after configuring the control area and the restoration area, biological data, physicochemical data, acoustic fingerprint data, image data, and microbiome data of oyster individuals in the oyster reef are collected to establish a multi-modal data set; the feature extraction module 12: after feature extraction is performed on the multi-modal data set, a feature index set is established, the oyster health comprehensive index is constructed using the feature index set, and system service class indicators are constructed; the resilience evaluation module 13: a disturbance event is configured, and disturbance observation of the control area and the restoration area is performed using the disturbance event, ecological resilience evaluation is performed using the disturbance observation results and the multi-modal data set, and an ecological resilience index is established; the attribution analysis module 14: the oyster health comprehensive index and the system service class indicators are taken as result variables, the multi-modal data set is taken as causal input, a structured causal diagram is established, attribution analysis of the restoration area and the control area is performed based on the structured causal diagram, the average treatment effect of the restoration measures on the oyster health comprehensive index and the system service class indicators is constructed, and after the oyster health comprehensive index and the system service class indicators are corrected based on the average treatment effect, the ecological resilience index is combined to output the oyster reef ecological restoration effect evaluation result.

[0038] Further, the data acquisition module 11 is used to perform the following method: The biological data includes oyster individual density, survival rate, and shell size distribution, the physicochemical data includes water temperature, salinity, pH value, dissolved oxygen, and nutrient salt concentration, the acoustic fingerprint data includes multi-channel underwater acoustic signal acoustic fingerprint data, the image data includes reef coverage and crack information data identified based on underwater image recognition, and the microbiome data includes data of microbial diversity and functional gene abundance obtained through high-throughput sequencing.

[0039] Further, the feature extraction module 12 is used to perform the following method: After the multi-modal data set is modally layered, independent normalization processing is respectively performed to establish a layered feature subset; principal component analysis of the layered feature subset is performed to determine a contribution factor of a layered feature factor, and the layered feature subset is weighted based on the contribution factor; survival rate, shell growth rate, and community density in a biological subset in the layered feature subset are extracted as a first health dimension; biological diversity index and functional gene richness in a microbiome subset are extracted as a second health dimension; frequency band complexity and a typical filter-feeding signal ratio in an acoustic fingerprint subset are extracted as a third health dimension; the first health dimension, the second health dimension, and the third health dimension are weighted and fused according to the weighted multi-modal coupling weight to establish the oyster health comprehensive index.

[0040] Further, the feature extraction module 12 is used to perform the following method: extract the nitrogen and phosphorus removal efficiency change rate index in the physicochemical subset, obtain the filter-feeding efficiency correlation factor in the oyster health comprehensive index, and establish a water quality purification efficiency index according to the nitrogen and phosphorus removal efficiency change rate index and the filter-feeding efficiency correlation factor; extract the matrix coverage rate and crack rate index in the image subset, and obtain the community density factor in the oyster health comprehensive index, and establish a sediment stability index according to the matrix coverage rate and crack rate index and the community density factor; after obtaining the community density factor in the oyster health comprehensive index, extract the carbon cycle functional gene abundance in the microbiome subset, and establish a carbon sink rate index; perform covariance analysis on the water quality purification efficiency index, the sediment stability index, the carbon sink rate index, and the oyster health comprehensive index, quantify the coupling degree of the oyster individual health and the system service function, and establish the system service index.

[0041] Further, the resilience evaluation module 13 is configured to perform the following method: A controllable disturbance event is arranged in the repair area and the control area, the controllable disturbance event including a short-term nutrient salt pulse, a sediment load disturbance, acoustic interference, and physical disturbance, and the disturbance intensity of the controllable disturbance event is set based on a statistical distribution of historical extreme environmental events; after the disturbance is performed, a multi-time sequence observation window is configured to perform disturbance observation on the control area and the repair area, and a disturbance observation result is established; the multi-modal data set is taken as base state data, and the disturbance observation result is taken as a disturbance result, and an anti-disturbance index, a recovery index, and an over-compensation index are extracted, and an ecological resilience index is established.

[0042] Further, the attribution analysis module 14 is configured to perform the following method: Based on the average treatment effect channel, average treatment effect control analysis of the repair area and the control area is performed; sensitivity disturbance identification of the average treatment effect control analysis result is performed, and an average treatment effect is generated.

[0043] Further, the attribution analysis module 14 is configured to perform the following method: A mapping between the repair measures and the average treatment effect is established, and a mapping data set is established; the mapping data set is taken as matching data, matching abnormality identification of the expert database is performed, a matching abnormality signal is established, and attention verification of the control area and the repair area is performed according to the matching abnormality signal, and the attention verification result is updated to the mapping data set, so as to perform subsequent oyster reef ecological restoration management.

[0044] Further, the resilience evaluation module 13 is configured to perform the following method: The disturbance event includes a biological invasion event, a pollutant instantaneous release time, and an extreme temperature change event.

[0045] Further, the attribution analysis module 14 is configured to perform the following method: The average treatment effect is correlated with the oyster health comprehensive index and the system service type index; after the misjudgment elimination under the causality driving is performed, the index normalization processing is executed, and the oyster reef ecological restoration effect evaluation result is generated by using the index normalization processing result.

[0046] It should be noted that the above sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes a specific embodiment of the present application. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0047] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0048] The present application and the drawings are only exemplary descriptions of the present application, and are considered to cover any and all modifications, changes, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalent technology, the present application intends to include these modifications and changes.

Claims

1. A data-driven oyster reef ecological restoration effect evaluation method, characterized in that, The method comprises: After configuring the control area and the repair area, biological data, physical and chemical data, acoustic fingerprint data, image data, and microbiome data of oyster individuals in the oyster reef are collected to establish a multi-modal data set; After feature extraction of the multi-modal data set, a feature index set is established, and an oyster health comprehensive index is constructed using the feature index set, and a system service class index is constructed; A disturbance event is configured, and disturbance observation of the control area and the repair area is performed using the disturbance event, ecological resilience evaluation is performed using the disturbance observation results and the multi-modal data set, and an ecological resilience index is established; The oyster health comprehensive index and the system service class index are taken as result variables, and the multi-modal data set is taken as causal input to establish a structured causal diagram, and attribution analysis of the repair area and the control area is performed based on the structured causal diagram, an average treatment effect of the repair measures on the oyster health comprehensive index and the system service class index is constructed, and after correction of the oyster health comprehensive index and the system service class index based on the average treatment effect, the ecological resilience index is output to output the oyster reef ecological restoration effect evaluation result.

2. The data-driven oyster reef ecological restoration effect evaluation method according to claim 1, wherein, The biological data includes oyster individual density, survival rate, and shell size distribution, the physical and chemical data includes water temperature, salinity, pH value, dissolved oxygen, and nutrient salt concentration, the acoustic fingerprint data includes multi-channel underwater acoustic signal acoustic fingerprint data, the image data includes reef coverage and crack information data identified based on underwater image recognition, and the microbiome data includes data of microbial diversity and functional gene abundance obtained through high-throughput sequencing.

3. The data-driven oyster reef ecological restoration effect evaluation method of claim 2, wherein, After feature extraction of the multi-modal data set, a feature index set is established, and an oyster health comprehensive index is constructed using the feature index set, and a system service class index is constructed, which comprises: After modal stratification of the multi-modal data set, independent normalization processing is performed respectively to establish stratified feature subsets; Principal component analysis of the stratified feature subsets is performed to determine contribution factors of stratified feature factors, and the stratified feature subsets are weighted based on the contribution factors; The survival rate, shell growth rate, and community density in the biological subset in the stratified feature subset are extracted as the first health dimension; The biological diversity index and functional gene richness in the microbiome subset are extracted as the second health dimension; The frequency band complexity and typical filter sound signal ratio in the acoustic fingerprint subset are extracted as the third health dimension; The first health dimension, the second health dimension, and the third health dimension are weighted and fused according to the weighted multi-modal coupling weight to establish the oyster health comprehensive index.

4. The data-driven oyster reef ecological restoration effect evaluation method of claim 3, wherein, After the oyster health comprehensive index is established, it comprises: The nitrogen and phosphorus removal efficiency change rate index in the physical and chemical subset is extracted, and the filter efficiency correlation factor in the oyster health comprehensive index is obtained, and a water quality purification efficiency index is established according to the nitrogen and phosphorus removal efficiency change rate index and the filter efficiency correlation factor; The matrix coverage and crack rate index in the image subset is extracted, and the community density factor in the oyster health comprehensive index is obtained, and a sediment stability index is established according to the matrix coverage and crack rate index and the community density factor; After extracting the carbon cycle function gene abundance in the microbiome subset and obtaining the community density factor in the oyster health comprehensive index, a carbon sink rate index is established; Covariance analysis is performed on the water quality purification efficiency index, sediment stability index, carbon sink rate index, and oyster health comprehensive index to quantify the coupling degree of oyster individual health and system service function, and a system service index is established.

5. The data-driven oyster reef ecological restoration effect evaluation method according to claim 1, characterized in that, The configuration disturbance event, the disturbance observation of the control area and the repair area are performed using the disturbance event, the ecological resilience evaluation is performed using the disturbance observation result and the multi-modal data set, and the ecological resilience index is established, including: Controllable disturbance events are set in the repair area and the control area, including short-term nutrient salt pulse, sediment load disturbance, acoustic interference, and physical disturbance, and the disturbance intensity of the controllable disturbance event is set based on the statistical distribution of historical extreme environmental events; After the disturbance is performed, a multi-time sequence observation window is configured to perform disturbance observation of the control area and the repair area, and disturbance observation results are established; The multi-modal data set is used as the base state data, and the disturbance observation result is used as the disturbance result to perform anti-disturbance index, recovery index, and over-compensation index extraction, and establish the ecological resilience index.

6. The data-driven oyster reef ecological restoration effect evaluation method of claim 1, wherein, The attribution analysis of the repair area and the control area is performed based on the structured causal diagram, and the average treatment effect of the repair measures on the oyster health comprehensive index and the system service index is constructed, including: Based on the average treatment effect channel, average treatment effect control analysis of the repair area and the control area is performed; Sensitivity disturbance identification of the average treatment effect control analysis result is performed to generate the average treatment effect.

7. The data-driven oyster reef ecological restoration effect evaluation method of claim 1, wherein, After the joint ecological resilience index outputs the oyster reef ecological restoration effect evaluation result, including: A mapping between the repair measures and the average treatment effect is established, and a mapping data set is established; The mapping data set is used as matching data to perform expert database matching anomaly identification, and a matching anomaly signal is established; According to the matching anomaly signal, attention verification of the control area and the repair area is performed, and the attention verification result is updated to the mapping data set to perform subsequent oyster reef ecological restoration management.

8. The data-driven oyster reef ecological restoration effect evaluation method of claim 1, wherein, The disturbance event includes biological invasion event, pollutant instantaneous release time, and extreme temperature change event.

9. The data-driven oyster reef ecological restoration effect evaluation method of claim 1, wherein, The average treatment effect is used to correct the oyster health comprehensive index and the system service index, including: The average treatment effect is used to establish an index correspondence relationship with the oyster health comprehensive index and the system service index; After false judgment elimination under causal driving is performed using the correspondence relationship, index normalization processing is performed, and the oyster reef ecological restoration effect evaluation result is generated using the index normalization processing result.

10. A data-driven oyster reef ecological restoration effectiveness evaluation system, characterized in that, A data-driven oyster reef ecological restoration effect evaluation method for implementing any one of claims 1-9, the system comprising: A data acquisition module: after configuring the control area and the repair area, biological data, physicochemical data, acoustic fingerprint data, image data, and microbiome data of oysters in the oyster reef are collected to establish a multi-modal data set; A feature extraction module: after feature extraction is performed on the multi-modal data set, a feature index set is established, the oyster health comprehensive index is constructed using the feature index set, and the system service index is constructed. The resilience evaluation module is configured to perform disturbance observation on the control area and the repair area by using the disturbance event, perform ecological resilience evaluation by using the disturbance observation result and the multi-modal data set, and establish an ecological resilience index; The attribution analysis module takes the oyster health comprehensive index and the system service index as a result variable, takes the multi-modal data set as causal input, establishes a structured causal graph, performs attribution analysis on the repair area and the control area based on the structured causal graph, constructs an average treatment effect of the repair measure on the oyster health comprehensive index and the system service index, and outputs an oyster reef ecological restoration effect evaluation result after correcting the oyster health comprehensive index and the system service index based on the average treatment effect and the ecological resilience index.

Images