Coastal wetland ecosystem condition evaluation method and device, medium and product

Abstract

The invention relates to the technical field of ecological assessment, and discloses a coastal wetland ecosystem condition assessment method and device, a medium and a product, different smooth regression functions are obtained through double generalized additive model processing, and the influence of an environmental single factor on vegetation and the cooperative influence of the environment and the vegetation on benthic organisms are accurately quantified. Furthermore, the index value is determined by taking historical data as a reference and combining a nonlinear rule of a smooth function, so that subjective weighting deviation is avoided. Furthermore, the weight is calculated based on the association between the real-time index value and the biological index, so that human subjective intervention is avoided, and the weight is ensured to reflect the actual contribution of the index to the ecological system. Meanwhile, the influence of single data deviation is reduced by using the multi-tree integration characteristic of the random forest model, so that the weight result is more stable. Furthermore, the complex ecosystem condition of the coastal wetland can be comprehensively evaluated in combination with the index value and the weight, the biodiversity is protected, and the stability of the ecosystem is maintained.

Description

Technical Field

[0001] This invention relates to the field of ecological assessment technology, specifically to a method, apparatus, medium, and product for assessing the status of coastal wetland ecosystems. Background Technology

[0002] Coastal wetlands are transitional zones between terrestrial and marine ecosystems. They provide ecological services such as flow regulation, mitigation of seawater intrusion, replenishment of groundwater, climate regulation, maintenance of biodiversity and ecological functions, protection of coastlines, tourism development, education and scientific research, and are a very important type of ecosystem.

[0003] In recent decades, with the increasing intensity of human activities, coastal wetlands have faced problems such as shrinking wetland area, eutrophication, invasive alien species, and reduced biodiversity. Therefore, a comprehensive evaluation and study of coastal wetlands is necessary. Assessment of the suitability of coastal wetland habitats can help implement more comprehensive conservation plans, including habitat restoration, protection of suitable habitats, and optimization of suitable habitat spatial layout, thereby protecting biodiversity and maintaining the species richness and stability of the ecosystem.

[0004] Currently, most habitat suitability assessment studies focus on a single species or community, with limited evaluation of multitrophic ecosystems. This makes it difficult to fully reflect the complex ecosystem conditions of coastal wetlands, resulting in an inability to protect biodiversity and maintain the species richness and stability of the ecosystem. Summary of the Invention

[0005] This invention provides a method, apparatus, medium, and product for assessing the status of coastal wetland ecosystems. This addresses the problem that most habitat suitability assessment studies focus on a single species or community, with limited evaluation of multi-trophic-level ecosystems. Consequently, these studies fail to comprehensively reflect the complex ecosystem conditions of coastal wetlands, leading to an inability to protect biodiversity and maintain the species richness and stability of the ecosystem.

[0006] In a first aspect, the present invention provides a method for assessing the status of coastal wetland ecosystems, the method comprising: The process involves acquiring the target real-time biological dataset and the target historical biological dataset of the coastal wetland to be evaluated. The target real-time biological dataset includes the target real-time environmental indicator dataset and the target real-time biological indicator dataset. Based on the target real-time biological dataset, a first generalized additive model and a second generalized additive model are used to obtain a first smooth regression function and a second smooth regression function. The first smooth regression function is used to characterize the impact of single-factor environmental indicators on vegetation biological indicators, and the second smooth regression function is used to characterize the impact of single-factor environmental indicators and vegetation biological indicators on benthic organisms. Multiple real-time evaluation indicator values ​​are determined based on the target historical biological dataset, the first smooth regression function, and the second smooth regression function. Based on the multiple real-time evaluation indicator values ​​and the target real-time biological indicator dataset, a random forest model is used to obtain multiple real-time evaluation indicator weight values. Finally, based on the multiple real-time evaluation indicator values ​​and their weight values, the ecosystem status of the coastal wetland to be evaluated is assessed, resulting in an assessment of the ecosystem status of the coastal wetland.

[0007] The coastal wetland ecosystem status assessment method provided by this invention, by acquiring a target real-time biological dataset containing environmental and biological indicators, can cover both the physical environment and biological community dimensions of coastal wetlands, filling the gap in existing assessments that neglect the overall ecosystem. Furthermore, by processing the data using a dual generalized additive model to obtain a first and second smooth regression function, the method accurately quantifies the impact of single environmental factors on vegetation and the synergistic effects of environment and vegetation on benthic organisms, overcoming the limitations of traditional linear models and thus accurately capturing the complex ecological relationships of coastal wetlands. Simultaneously, it ensures that the assessment covers key components of the ecosystem, solving the problem of the one-sidedness of existing assessments that only target single biological groups. Furthermore, by using historical data as a benchmark and combining the nonlinearity of the smoothing function to determine indicator values, it avoids subjective weighting bias and ensures that the indicator values ​​truly reflect the adaptability of the environment and organisms. Furthermore, by calculating weights based on the correlation between real-time indicator values ​​and biological indicators, it avoids subjective human intervention and ensures that the weights reflect the actual contribution of the indicators to the ecosystem. At the same time, by utilizing the multi-tree ensemble characteristics of the random forest model, it reduces the impact of single data bias, making the weighting results more robust. Furthermore, by combining indicator values ​​and weights to assess the ecosystem status of coastal wetlands, the assessment results cover multiple dimensions of vegetation and benthic organisms, providing precise guidance for the priority classification of habitat restoration and the determination of the scope of suitable habitat protection, which helps to protect biodiversity and maintain ecosystem stability.

[0008] In one alternative implementation, a target real-time biological dataset of the coastal wetland to be evaluated is obtained, including: Obtain the initial real-time biological dataset of the coastal wetland to be evaluated; remove outliers from the initial real-time biological dataset to obtain the first real-time biological dataset; perform multicollinearity processing on the environmental indicator data in the first real-time biological dataset to obtain the second real-time biological dataset; assign values ​​to the discrete data in the second real-time biological dataset to obtain the target real-time biological dataset.

[0009] The coastal wetland ecosystem status assessment method provided by this invention ensures that the data conforms to the actual ecological characteristics of the wetlands by removing outliers, thus improving data reliability. Furthermore, by performing multicollinearity processing on environmental indicator data, highly correlated environmental factors can be eliminated, avoiding evaluation bias caused by repeated contributions from multiple factors and reducing the computational load of subsequent models. At the same time, core environmental factors are retained, ensuring that the data is free of redundancy and that key information is complete, laying a high-quality data foundation for subsequent index selection in generalized additive models. Furthermore, by performing discrete data assignment processing, non-numerical discrete data can be converted into quantifiable values, which helps ensure the normal operation of subsequent generalized additive models and random forest models.

[0010] In one alternative implementation, the method further includes: We obtain historical environmental indicator datasets and historical biological indicator datasets from the target historical biological dataset. The historical biological indicator datasets include historical vegetation biological indicator datasets and historical benthic biological indicator datasets. Based on the Akaike information content criterion, we construct a first generalized additive model with the historical environmental indicator dataset as the explanatory variable and the historical vegetation biological indicator dataset as the response variable. Based on the Akaike information content criterion, we construct a second generalized additive model with the historical environmental indicator dataset and the historical vegetation biological indicator dataset as the explanatory variable and the historical benthic biological indicator dataset as the response variable.

[0011] The coastal wetland ecosystem status assessment method provided by this invention constructs a generalized additive model based on the Akaike information content criterion. Using historical environmental indicator datasets as explanatory variables and historical vegetation biological indicator datasets as response variables, a first generalized additive model is constructed. This approach balances model fit and complexity, enabling the selection of environmental factors with the most significant impact on vegetation while eliminating irrelevant factors. This ensures that the first smooth regression function output by the model accurately reflects the relationship between the environment and vegetation. Furthermore, a second generalized additive model is constructed based on the Akaike information content criterion, incorporating vegetation indicators as explanatory variables. This model captures the chain-like relationships between the environment, vegetation, and benthic organisms, overcoming the limitations of traditional single-factor evaluation. Simultaneously, the variable combinations selected based on the Akaike information content criterion ensure that the model can explain the major variations in benthic organisms, thus enabling the second smooth regression function to provide a scientific basis for benthic suitability assessment and filling the gap in multi-trophic level assessment models.

[0012] In one optional implementation, multiple real-time evaluation index values ​​are determined based on the target historical biological dataset, a first smooth regression function, and a second smooth regression function, including: Based on the first and second smooth regression functions, multiple coastal wetland habitat suitability evaluation indicators are determined; based on the target historical biological dataset, the first and second smooth regression functions, multiple evaluation indicator range thresholds are determined; based on the multiple evaluation indicator range thresholds, multiple coastal wetland habitat suitability evaluation indicators are assigned values ​​to obtain multiple real-time evaluation indicator values.

[0013] The coastal wetland ecosystem status assessment method provided by this invention, through the nonlinearity and significance analysis of a smoothing function, can screen out multiple evaluation indicators that have a significant impact on vegetation and benthic organisms. Simultaneously, it eliminates factors with no significant impact, ensuring that the evaluation indicators possess both statistical significance and ecological meaning, thus solving the problem of blind indicator selection. Furthermore, historical data ensures that the thresholds conform to the long-term ecological characteristics of wetlands, and the smoothing function ensures that the thresholds reflect the nonlinear relationship between the environment and organisms, avoiding a one-size-fits-all threshold setting. This makes the determined threshold ranges for multiple evaluation indicators both historically reasonable and ecologically scientific. Furthermore, assigning values ​​to evaluation indicators based on thresholds allows real-time evaluation indicator values ​​to be directly correlated with the current state of the real-time evaluation indicators, providing operable input for subsequent weight calculations.

[0014] In one alternative implementation, the method further includes: Obtain multiple historical evaluation index values; based on node purity increments, construct a random forest model with multiple historical evaluation index values ​​as dependent variables and historical biological index datasets as explanatory variables.

[0015] The coastal wetland ecosystem status assessment method provided by this invention, by acquiring multiple historical evaluation index values, can provide long-term training samples for a random forest model, ensuring that the model can learn the correlation between historical indicators and biological responses, thus improving the model's reliability. Furthermore, by using node purity increments, the improvement in the classification purity of biological indicators by each indicator can be accurately measured, thereby ensuring that the model can identify the actual ecological contribution of each indicator. Moreover, using historical biological indicator datasets as explanatory variables ensures that the constructed random forest model possesses historical statistical significance, providing a stable model framework for subsequent real-time weight calculations.

[0016] In one optional implementation, based on multiple real-time evaluation index values ​​and a target real-time biometric dataset, a random forest model is used to process multiple real-time evaluation index weight values, including: Multiple real-time evaluation index values ​​and target real-time biological index datasets are input into a random forest model to obtain multiple node purity increment values; the multiple node purity increment values ​​are normalized to obtain multiple real-time evaluation index weight values.

[0017] The coastal wetland ecosystem status assessment method provided by this invention calculates incremental values ​​based on real-time data of the wetland to be evaluated, ensuring that the weights reflect the actual impact of indicators under the current environment. This avoids evaluation lag caused by directly applying historical weights and improves the timeliness of the weights. Furthermore, normalization ensures that the weights of different indicators are comparable by summing the weights of the same dimension to one. Simultaneously, normalization eliminates the influence of differences in indicator dimensions, ensuring that the weights only reflect the contribution of the indicators and improving the scientific rigor of the evaluation index calculation.

[0018] In one optional implementation, the ecosystem status of the coastal wetland to be evaluated is assessed based on multiple real-time evaluation index values ​​and multiple real-time evaluation index weight values ​​to obtain the ecosystem status assessment result of the coastal wetland to be evaluated, including: Based on multiple real-time evaluation index values ​​and their weights, the habitat suitability evaluation value of the coastal wetland to be evaluated is determined. Using the habitat suitability evaluation value, the ecosystem status of the coastal wetland to be evaluated is assessed, and the ecosystem status assessment result of the coastal wetland to be evaluated is obtained.

[0019] The coastal wetland ecosystem status assessment method provided by this invention integrates multi-dimensional indicator values ​​into a single evaluation value through weighting, enabling direct comparison of wetland ecological status. Furthermore, based on the habitat suitability evaluation value, it is possible to clearly identify the advantageous areas and weak links of the wetland ecosystem, and the assessment results cover multiple trophic levels of vegetation and benthic organisms, providing guidance for evaluating the restoration effect after remediation and optimizing habitat spatial layout.

[0020] Secondly, the present invention provides a device for assessing the status of a coastal wetland ecosystem, the device comprising: The system comprises the following modules: an acquisition module for acquiring the target real-time biological dataset and the target historical biological dataset of the coastal wetland to be evaluated, including the target real-time environmental index dataset and the target real-time biological index dataset; a first processing module for processing the target real-time biological dataset using a first generalized additive model and a second generalized additive model to obtain a first smooth regression function and a second smooth regression function. The first smooth regression function characterizes the impact of single-factor environmental indicators on vegetation biological indicators, and the second smooth regression function characterizes the impact of single-factor environmental indicators and vegetation biological indicators on benthic organisms; a determination module for determining multiple real-time evaluation index values ​​based on the target historical biological dataset, the first smooth regression function, and the second smooth regression function; a second processing module for processing the multiple real-time evaluation index values ​​and the target real-time biological index dataset using a random forest model to obtain multiple real-time evaluation index weight values; and an evaluation module for evaluating the ecosystem status of the coastal wetland to be evaluated based on the multiple real-time evaluation index values ​​and the multiple real-time evaluation index weight values ​​to obtain the ecosystem status evaluation result of the coastal wetland to be evaluated.

[0021] Thirdly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the coastal wetland ecosystem status assessment method described in the first aspect or any corresponding embodiment thereof.

[0022] Fourthly, the present invention provides a computer program product, including computer instructions for causing a computer to execute the coastal wetland ecosystem status assessment method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the method for assessing the status of coastal wetland ecosystems according to an embodiment of the present invention; Figure 3 This is a structural block diagram of a coastal wetland ecosystem status assessment device according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0027] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0028] As an optional application scenario of this invention, the specific application environment architecture or specific hardware architecture on which the coastal wetland ecosystem status assessment method depends is described herein. For example... Figure 1 As shown, the architecture system may include at least one terminal device and at least one server. Figure 1 The system is illustrated in the example, which includes a computer 101, a mobile terminal 102, and a server 103, and the terminal devices such as the computer 101 and the mobile terminal 102 are connected to the server 103 through a network 110.

[0029] Specifically, the terminal device can be a smartphone, tablet, laptop, PDA, desktop computer, game console, smart TV, smart wearable device, in-vehicle terminal, VR (Virtual Reality) device, AR (Augmented Reality) device, etc. Server 103 can be a standalone physical server, a server cluster, a distributed system, or a cloud server providing cloud services. Network 110 can be a wired or wireless network, examples of which include, but are not limited to, the Internet, corporate intranet, local area network, wide area network, mobile communication network, and combinations thereof.

[0030] This invention provides a method for assessing the status of coastal wetland ecosystems. It uses a dual generalized additive model to determine evaluation indicators and combines a random forest model to determine the weights of the indicators, thereby achieving the effects of assessing multitrophic ecosystems, protecting biodiversity, and maintaining ecosystem stability.

[0031] According to an embodiment of the present invention, an embodiment of a method for assessing the status of a coastal wetland ecosystem is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0032] This embodiment provides a method for assessing the status of coastal wetland ecosystems, which can be used on the aforementioned mobile terminals, such as mobile phones and tablets. Figure 2 This is a flowchart of a method for assessing the status of coastal wetland ecosystems according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain the target real-time biological dataset and the target historical biological dataset of the coastal wetland to be evaluated.

[0033] The target real-time biological dataset includes the target real-time environmental index dataset and the target real-time biological index dataset.

[0034] In one optional embodiment, the target real-time environmental indicator dataset represents a quantitative dataset reflecting the physical, chemical, and substrate abiotic environmental characteristics of the wetland, obtained through on-site monitoring and sampling analysis during the current survey period of the coastal wetland to be evaluated. It may include sedimentation rate, elevation, water temperature, salinity, dissolved oxygen, redox potential, sediment organic carbon, sediment particle size, sediment total nitrogen content, sediment total phosphorus content, sediment potassium content, etc.

[0035] In one optional embodiment, the target real-time bioindicator dataset represents a quantitative dataset that reflects the growth, distribution, and diversity characteristics of wetland vegetation, benthic organisms, and other biological communities, obtained through field surveys and sample statistical analysis during the current survey period of the coastal wetland to be evaluated. It may include vegetation bioindicator data such as vegetation type and vegetation density, as well as benthic bioindicator data such as benthic animal habitat density, benthic animal biomass, and benthic animal diversity index.

[0036] In one optional embodiment, the target historical biological dataset refers to a long-term dataset containing historical environmental and biological indicators, collected prior to the current survey period of the coastal wetland to be evaluated through historical documents, past monitoring reports, long-term observation data, and other channels.

[0037] Step S202: Based on the target real-time biological dataset, the first smooth regression function and the second smooth regression function are obtained by processing the dataset through the first generalized additive model and the second generalized additive model.

[0038] In an alternative embodiment, the Generalized Additive Model (GAM) is a non-parametric extension of the generalized linear model. Its advantage is that it can directly handle the nonlinear relationship between the response variable and multiple explanatory variables, expressed as the following relationship (1): (1) In the formula: To represent the dependent variable, if abundance is the dependent variable, in order to deal with the case where the abundance is 0, the abundance can be increased by 1 and then logarithmicized. Indicates the intercept; Explanation variables The smooth regression function; Explanation variables The total number.

[0039] In one optional embodiment, the first generalized additive model represents a GAM model used to quantify the impact of single-factor environmental indicators on vegetation biological indicators; the second generalized additive model represents a GAM model used to quantify the impact of single-factor environmental indicators and vegetation biological indicators on benthic organisms.

[0040] In one optional embodiment, a first smooth regression function is used to characterize the influence of environmental single-factor indicators on vegetation biological indicators, and a second smooth regression function is used to characterize the influence of environmental single-factor indicators and vegetation biological indicators on benthic organisms.

[0041] In one optional embodiment, two types of targeted generalized additive models, namely the first GAM and the second GAM, are used to perform nonlinear correlation analysis on the target real-time biological dataset, which can ultimately output two types of smooth regression functions that reflect the core ecological relationships of wetlands, namely the first smooth regression function and the second smooth regression function.

[0042] For example, the target real-time biological dataset is input into the first GAM model, and then the environmental factor indicators in the target real-time biological dataset are used as the explanatory variables of the first GAM model, and the vegetation biological indicators in the target real-time biological dataset are used as the response variables of the first GAM model.

[0043] Furthermore, the first GAM model can fit the nonlinear relationship between variables through a nonparametric smoothing function. That is, it captures the change in the intensity of the influence of different environmental single factors on vegetation biological indicators in different value ranges by using the trend of the data itself. Then, when the fitting is completed, it outputs the first smoothing regression function that represents the influence of environmental single factor indicators on vegetation biological indicators.

[0044] Furthermore, the target real-time biological dataset is input into the second GAM model. Then, the environmental factor indicators and vegetation biological indicators in the target real-time biological dataset are used as explanatory variables of the second GAM model, and the benthic biological indicators in the target real-time biological dataset are used as response variables of the second GAM model.

[0045] Furthermore, the second GAM model also fits the complex relationships between variables through a nonparametric smoothing function. That is, it can not only capture the independent effects of individual environmental indicators / vegetation indicators on benthic organisms, but also capture the synergistic / antagonistic effects of the two. Then, when the fitting is completed, it outputs the corresponding second smoothing regression function representing the effects of single environmental indicators and vegetation biological indicators on benthic organisms.

[0046] In an optional embodiment, it is also possible to generate effect diagrams showing the significant impact of environmental single-factor indicators on vegetation biological indicators and effect diagrams showing the significant impact of environmental single-factor indicators and vegetation biological indicators on benthic organisms.

[0047] Step S203: Determine multiple real-time evaluation index values ​​based on the target historical biological dataset, the first smooth regression function, and the second smooth regression function.

[0048] In one alternative embodiment, by utilizing the reference value of historical data and constructing a first smooth regression function and a second smooth regression function, the influence relationship between the environment and organisms can be transformed into quantifiable and comparable real-time evaluation index values.

[0049] Step S204: Based on multiple real-time evaluation index values ​​and the target real-time biological index dataset, multiple real-time evaluation index weight values ​​are obtained through random forest model processing.

[0050] In one optional embodiment, the random forest model represents a machine learning model based on the idea of ​​ensemble learning. By constructing multiple independent decision trees, it performs parallel predictions on the input data and outputs the final result in the form of majority voting (classification task) or result averaging (regression task).

[0051] In one optional embodiment, multiple real-time evaluation index values ​​are used as input features of the random forest model, and the target real-time biological index dataset reflecting the current ecological status of the coastal wetland to be evaluated is used as the reference target for learning the random forest model. Then, the correlation between the two is learned and analyzed through the constructed random forest model, and the real-time evaluation index weight value corresponding to each real-time evaluation index can be finally output.

[0052] Step S205: Based on multiple real-time evaluation index values ​​and multiple real-time evaluation index weight values, assess the ecosystem status of the coastal wetland to be evaluated, and obtain the ecosystem status assessment results of the coastal wetland to be evaluated.

[0053] In one optional embodiment, the ecosystem status refers to the overall functional status and health level of the coastal wetland to be evaluated at the current stage, formed by the interaction between its internal biological community and abiotic environment. It is used to reflect whether the stability, self-regulation capacity, and service provision capacity of the wetland ecosystem are within a reasonable range, and whether there are problems such as degradation or imbalance.

[0054] In one optional embodiment, multiple real-time evaluation indicators are used as the basis for evaluation, and multiple real-time evaluation indicator weights are used as the evaluation basis. By scientifically integrating and calculating the indicator values ​​and corresponding weight values ​​of each indicator, an overall conclusion that can comprehensively reflect the health status and functional status of the coastal wetland ecosystem to be evaluated can be finally formed, namely, the ecosystem status assessment result.

[0055] The coastal wetland ecosystem status assessment method provided in this embodiment, by acquiring a target real-time biological dataset containing both environmental and biological indicators, can cover both the physical environment and biological community dimensions of coastal wetlands, filling the gap in existing assessments that neglect the overall ecosystem. Furthermore, by processing the data using a dual generalized additive model to obtain a first and second smooth regression function, the method accurately quantifies the impact of single environmental factors on vegetation and the synergistic effects of environment and vegetation on benthic organisms, overcoming the limitations of traditional linear models and thus accurately capturing the complex ecological relationships of coastal wetlands. Simultaneously, it ensures that the assessment covers key components of the ecosystem, solving the problem of the one-sidedness of existing assessments that only target single biological groups. Furthermore, by using historical data as a benchmark and combining the nonlinearity of the smoothing function to determine indicator values, subjective weighting bias is avoided, ensuring that the indicator values ​​truly reflect the adaptability of the environment and organisms. Furthermore, by calculating weights based on the correlation between real-time indicator values ​​and biological indicators, subjective human intervention is avoided, ensuring that the weights reflect the actual contribution of the indicators to the ecosystem. At the same time, by utilizing the multi-tree ensemble characteristics of the random forest model, the influence of single data bias is reduced, making the weighting results more robust. Furthermore, by combining indicator values ​​and weights to assess the ecosystem status of coastal wetlands, the assessment results cover multiple dimensions of vegetation and benthic organisms, providing precise guidance for the priority classification of habitat restoration and the determination of the scope of suitable habitat protection, which helps to protect biodiversity and maintain ecosystem stability.

[0056] In some optional implementations, the target real-time biological dataset in step S201 above is obtained through the following steps: Step a1: Obtain the initial real-time biological dataset of the coastal wetland to be evaluated.

[0057] In one optional embodiment, biological and related environmental data that can reflect the current (real-time) ecological status of the coastal wetland to be evaluated are collected through scientific observation and monitoring methods, and an unprocessed raw dataset, i.e., the initial real-time biological dataset, is formed.

[0058] Step a2: Remove outliers from the initial real-time biological dataset to obtain the first real-time biological dataset.

[0059] In one optional embodiment, outliers in the initial real-time biological dataset caused by acquisition errors, instrument malfunctions, or extreme accidental events (such as sudden pollution or human interference) are identified and removed to obtain the corresponding first real-time biological dataset, thus ensuring the reliability of the data.

[0060] In one alternative embodiment, outliers are identified by combining statistical methods with professional judgment.

[0061] For example, for normally distributed data, the standard deviation of the sample is first calculated. Then check the absolute value of the difference between each value and the mean. If the absolute value is greater than 3 times the standard deviation... If the value is not found, it is considered an outlier and is removed.

[0062] For non-normally distributed data, calculate the first quartile (Q). 25 ) and the third quartile (Q 75 Then calculate R=Q 75 -Q 25 It will be lower than Q. 25 -1.5R or higher than Q 75 Data with a value of +1.5R is considered outlier and is removed.

[0063] Step a3: Perform multicollinearity processing on the environmental indicator data in the first real-time biological dataset to obtain the second real-time biological dataset.

[0064] Because some environmental indicators may have multicollinearity, directly using them for subsequent analysis can lead to redundant model calculations and biased parameter estimation.

[0065] In one alternative embodiment, multicollinearity indicates a strong linear correlation between the indices.

[0066] In one alternative embodiment, by performing multicollinearity processing on the environmental indicator data in the first real-time biological dataset, a second real-time biological dataset with stronger independence among the indicators can be obtained.

[0067] In one optional embodiment, after the environmental factors pass the normality test, correlation analysis is performed on the screened environmental factors to test for multicollinearity. For environmental factors with high correlation (in this embodiment, |r|>0.75 is used as the criterion), it is considered that the factor has a collinearity problem, and one factor is retained, while other collinear factors should be discarded. Finally, the siltation data, elevation, organic carbon, total phosphorus, sorting coefficient, median particle size, pH, salinity, and dissolved oxygen (DO1) are retained as the main environmental factors, and a second real-time biological dataset with no collinearity of environmental indicators is obtained.

[0068] In an optional embodiment, multicollinearity can also be addressed using the following method: (1) Principal Component Analysis (PCA): When there are multiple highly correlated environmental indicators, PCA is used to transform them into a few uncorrelated principal component indicators, retaining the main information of the original data while eliminating collinearity; (2) Stepwise regression method: Incorporate environmental indicators one by one into the regression model, and remove indicators that contribute little to the model and cause collinearity by testing the significance of the indicators; (3) Ridge regression: If it is necessary to retain all environmental indicators (to avoid information loss), a penalty term can be added to reduce the impact of collinearity on the model. After processing, a second real-time biological dataset with no collinearity or collinearity within the allowable range is obtained.

[0069] Step a4: Assign values ​​to the discrete data in the second real-time biological dataset to obtain the target real-time biological dataset.

[0070] In one optional embodiment, the discrete data in the second real-time biological dataset are assigned values ​​to transform them into continuous numerical data, and finally a standardized target real-time biological dataset is obtained.

[0071] In one optional embodiment, the vegetation type is assigned a value, with the values ​​for bare beach, sea buckthorn area, and reed area being 3, 2, and 1, respectively; the substrate type is assigned a value, with the values ​​for silt, clayey silt, sandy silt, silty sand, and sand being 1, 2, 3, 4, and 5, respectively.

[0072] In one optional embodiment, the unordered discrete data and ordered discrete data in the second real-time biological dataset are first distinguished. Then, for the unordered discrete data, one-hot encoding can be used to assign values, which can avoid assigning false classification grades based on numerical values; for the ordered discrete data, ordered numerical mapping can be used to assign values.

[0073] In some optional implementations, the first generalized additive model and the second generalized additive model in step S202 above are obtained through the following steps: Step b1: Obtain the historical environmental index dataset and the historical biological index dataset from the target historical biological dataset.

[0074] In one alternative embodiment, the historical biometrics dataset includes a historical vegetation biometrics dataset and a historical benthic biometrics dataset.

[0075] For details, please refer to the content of the target real-time environmental indicator dataset and the target real-time biological indicator dataset in step S201 above, which will not be repeated here.

[0076] Step b2: Based on the Akaike information content criterion, the first generalized additive model is constructed using the historical environmental index dataset as the explanatory variable and the historical vegetation biological index dataset as the response variable.

[0077] In one optional embodiment, the Akaike information criterion (AIC) represents a standard for measuring the goodness of fit of a statistical model. It balances the goodness of fit and complexity among multiple possible statistical models, preventing overfitting due to excessive complexity. Furthermore, the higher the goodness of fit (smaller residuals), the lower the AIC value; the more parameters in the model (higher complexity), the higher the AIC value.

[0078] In one optional embodiment, the abiotic factors affecting the growth of coastal wetland vegetation, namely the historical environmental index dataset, are used as explanatory variables, and the survival and growth status data of coastal wetland vegetation, namely the historical vegetation biological index dataset, are used as response variables. By screening the optimal model structure through the Akaike Information Content Criterion (AIC), the first generalized additive model (GAM) describing the nonlinear relationship between environmental factors and vegetation indicators can be finally constructed.

[0079] Specifically, using the Akaike Information Content Criterion (AIC) as the model selection criterion, starting with a single variable, other environmental variables were added to the model with the smallest AIC value in turn. The changes in AIC values ​​under different combinations of environmental variables were observed. The combination of independent variables with lower AIC values ​​was used as the explanatory variables, and the model with the smallest AIC value was selected as the optimal model.

[0080] For example, by using historical environmental indicator datasets as explanatory variables and historical vegetation biological indicator datasets as response variables, and by substituting these explanatory and response variables into the generalized additive model shown in the above relation (1), a GAM model containing all candidate environmental indicators can be initially constructed.

[0081] Further, the AIC value of the initial all-variable GAM model is calculated. Then, the explanatory variables are adjusted by stepwise elimination or stepwise introduction, that is, each time an environmental indicator that contributes the least to the model fit is eliminated or an indicator that is not included is introduced, the GAM model is reconstructed and the AIC value is calculated.

[0082] Furthermore, repeat the above process until the model with the smallest AIC value is obtained, which is the first generalized additive model.

[0083] Step b3: Based on the Akaike information content criterion, a second generalized additive model is constructed, using historical environmental indicator datasets and historical vegetation biological indicator datasets as explanatory variables and historical benthic biological indicator datasets as response variables.

[0084] In one optional embodiment, since the survival of benthic organisms in coastal wetlands is not only directly affected by environmental factors but also indirectly affected by vegetation, a second generalized additive model describing the multi-factor association between environment, vegetation, and benthic organisms is constructed by using historical environmental index datasets and historical vegetation biological index datasets as explanatory variables, and the data on the survival and community status of benthic organisms, i.e., historical benthic organism index datasets, as response variables. The optimal structure is also selected through the AIC criterion.

[0085] For example, by substituting the extended explanatory variable (X+Z) and response variable W into the generalized additive model shown in the above relation (1), the corresponding initial GAM model can be initially constructed.

[0086] Further, the AIC value of the initial all-variable (X+Z) GAM model is calculated. Then, a stepwise optimization strategy can be used to adjust the explanatory variables, which may involve removing environmental indicators that have no significant impact on benthic organism indicators, or vegetation indicators that have no significant impact. After each adjustment, the model is rebuilt and the AIC value is calculated.

[0087] Furthermore, when the model's AIC value reaches its minimum, optimization stops, resulting in a second generalized additive model that simultaneously reflects the direct impact of environmental factors on benthic organisms, as well as the indirect impact of vegetation factors on benthic organisms, and achieves an optimal balance between goodness of fit and complexity.

[0088] In some optional implementations, step S203 above includes: Step S2031: Determine multiple coastal wetland habitat suitability evaluation indicators based on the first smooth regression function and the second smooth regression function.

[0089] In one optional embodiment, the habitat suitability of coastal wetlands depends on their ability to support the survival of vegetation, benthic organisms, and other organisms. The first and second smooth regression functions have quantified the correlation between environmental factors and the survival status of vegetation / benthic organisms. Therefore, the environmental and biological correlation factors that play a decisive role in habitat suitability can be extracted from the first and second smooth regression functions and defined as the coastal wetland habitat suitability evaluation index.

[0090] For example, firstly, the explanatory and response variables of the first and second smooth regression functions are determined. Then, environmental factors or organism-environment associations that are crucial to the survival of vegetation / benthic organisms can be screened based on the significance of the influence of the explanatory variables on the response variables in the functions, and these can be defined as indicators for evaluating the suitability of coastal wetland habitats. Simultaneously, a clear evaluation direction can be assigned to each selected indicator.

[0091] Step S2032: Determine the threshold ranges of multiple evaluation indicators based on the target historical biological dataset, the first smooth regression function, and the second smooth regression function.

[0092] In an optional embodiment, the evaluation index threshold is used to determine whether the habitat suitability evaluation index meets the standard.

[0093] In an optional embodiment, the suitable range threshold, critical range threshold, or unsuitable range threshold of each habitat suitability evaluation index can be derived in reverse using the target historical biological dataset and existing first and second smooth regression functions.

[0094] For example, samples with good biological survival status are selected from the target historical biological dataset, namely, high-quality habitat samples whose vegetation indicators and benthic biological indicators meet the standards for healthy habitats.

[0095] Furthermore, for each determined habitat suitability evaluation index, the actual values ​​of the corresponding environmental factors in the high-quality habitat samples are extracted, and their distribution range, i.e., the evaluation index range threshold, is statistically analyzed.

[0096] Furthermore, the initially determined range thresholds are substituted into the first and second smoothing regression functions to verify whether the response variables (biological indicators) are indeed in a healthy state under these range thresholds. If they are in a healthy state, the range thresholds are confirmed to be effective; if there are some values ​​where the response variable is not in a healthy state, the range thresholds are adjusted until the function validation results fully match the biological standards of high-quality samples, and finally, the appropriate range thresholds and critical range thresholds for each evaluation indicator are determined.

[0097] In an alternative embodiment, for continuously distributed data: (1) For smooth regression functions that are not monotonic, the "suitable" range threshold is the interval (x1, x2) of explanatory variables corresponding to more than 75% of the maximum value of the dependent variable, and the threshold is 25%-50% of the maximum value of the explanatory variable (x1, x2). 01 x1)(x2, x 02 The threshold value is set as the "suitable" range, used to explain the minimum values ​​of variables x0 and x. 01 Between and x 02 and the maximum value of the explanatory variable x m The threshold is within the "normal" range.

[0098] (2) For the monotonically changing smooth regression function, combined with many years of historical measured data and literature research, the 38th quantile (Q) is used. 38 ) and the 63rd quantile (Q 63 The "suitable" range threshold is defined as between 0 and 1, with the first quartile (Q) as the threshold. 25 ) and the 38th quantile (Q 38Between ) and the 63rd quantile (Q) 63 ) and the third quartile (Q 75 The threshold between ) is the "suitable" range, with the minimum value (Q1) and the first quartile (Q) as the thresholds. 25 Between and the third quartile (Q) 75 ) and maximum value (Q) 100 The range between ) is the "normal" threshold.

[0099] Furthermore, for discrete categorical data, based on the regression function, the explanatory variables corresponding to more than 75% of the maximum value of the dependent variable are considered "suitable," those corresponding to 50% to 75% of the maximum value of the dependent variable are considered "relatively suitable," and those corresponding to less than 50% of the maximum value of the dependent variable are considered "average."

[0100] Step S2033: Based on the threshold range of multiple evaluation indicators, assign values ​​to multiple coastal wetland habitat suitability evaluation indicators to obtain multiple real-time evaluation indicator values.

[0101] In one optional embodiment, the actual values ​​of the current habitat indicators of the coastal wetland to be evaluated are compared with the threshold standards, and then the qualitative suitability judgment is transformed into a quantitative value through quantitative assignment, and finally a real-time evaluation indicator value that can be directly used for subsequent weighted evaluation is obtained.

[0102] In one optional embodiment, the 1, 2, 3 assignment method is used to assign values ​​to the indicators: after determining the threshold, the indicators are assigned scores of 3, 2, and 1 according to the indicators "suitable", "relatively suitable", and "average".

[0103] In one optional embodiment, the current actual value corresponding to each habitat suitability evaluation index is extracted from the target real-time biological dataset. Then, a tiered assignment rule is set for each index according to the determined evaluation index range threshold, generally following the logic that the higher the suitability, the higher the score.

[0104] Furthermore, the actual value of each real-time indicator of the wetland to be evaluated is matched with the corresponding range threshold and assignment rules to calculate the quantitative score of each indicator, i.e., the real-time evaluation indicator value.

[0105] In some optional implementations, the random forest model in step S204 above is obtained through the following steps: Step c1: Obtain multiple historical evaluation index values.

[0106] The specific process can be referred to in step S203 above for obtaining multiple real-time evaluation index values, and will not be repeated here.

[0107] Step c2: Based on the node purity increment, a random forest model is constructed with multiple historical evaluation index values ​​as dependent variables and historical biological index datasets as explanatory variables.

[0108] In an optional embodiment, the node purity increment (IncNodePurity) represents the increase in the overall node purity compared to the original node after splitting the current node using a certain feature, i.e., the historical evaluation index value, during the decision tree construction process. It is used to measure the classification purity of nodes in the decision tree.

[0109] Furthermore, when a feature is added, this metric indicates the degree to which node purity increases compared to the case without that feature. A higher IncNodePurity value indicates that the feature contributes significantly to node classification, making the node purer.

[0110] In one optional embodiment, a historical biometric dataset is used as an explanatory variable, and multiple historical evaluation index values ​​are used as dependent variables. The optimal splitting feature and threshold are selected by incremental node purity, and multiple decision trees are generated and integrated. This ultimately yields a random forest model that can be used to predict the weights of real-time evaluation indicators.

[0111] For example, each decision tree is constructed according to the following process: (1) Random sampling of samples: n samples are randomly selected from the original sample set (X, Y) using the Bootstrap sampling method (n is the same as the original number of samples, and repeated sampling is allowed) as training samples for the current decision tree; at the same time, the remaining unselected samples (about 30%) are used as out-of-bag samples for subsequent model validation.

[0112] (2) Random feature selection: k features (k is the initialization parameter) are randomly selected from the total features, i.e., historical evaluation indicators, as candidate features that the current node can be used for splitting.

[0113] (3) Calculate the node purity increment and select the optimal splitting method: A. For each candidate feature, iterate through all its possible splitting thresholds; B. For each feature and threshold combination, split the current node sample into a left child node and a right child node; wherein, the sample feature value of the left child node is less than or equal to the threshold; and the sample feature value of the right child node is greater than the threshold. C. Calculate the purity of the original node before splitting (e.g., Gini coefficient G0), and the purity of the two child nodes after splitting (G1, G2), and calculate the total purity G_total after splitting by weighting the child node sample proportions; D. Calculate the node purity increment: Increment = G0 - G_total, that is, the larger the increment, the more obvious the improvement in node homogeneity after splitting; E. Select the feature and threshold combination that maximizes the increase in node purity as the optimal splitting method for the current node, and complete the node splitting.

[0114] Further, repeat steps (2)-(3) for the split child nodes until the nodes meet the stopping condition and finally generate a complete decision tree.

[0115] Further, repeat steps (1)-(3) above to generate a number of decision trees with initialization parameters set, and form a random forest model.

[0116] Furthermore, the out-of-bag samples are input into each decision tree that was not trained using these samples to obtain prediction results, which are then compared with the true Y values ​​of the out-of-bag samples to calculate the model accuracy or mean squared error. If the performance is not up to standard, the parameters are adjusted and the model is rebuilt until the model performance meets the standard, thus obtaining the final random forest model.

[0117] In some optional implementations, step S204 above includes: Step S2041: Input multiple real-time evaluation index values ​​and target real-time biological index datasets into the random forest model to obtain multiple node purity increment values.

[0118] In one optional embodiment, multiple real-time evaluation index values ​​and target real-time biological index datasets are input into a random forest model. The random forest model can then substitute the input real-time data into each decision tree to simulate splitting. That is, for each real-time evaluation index, the model will attempt to split a node of the decision tree into two child nodes using the index as the splitting variable, and calculate the change in node purity before and after the split, i.e., the node purity increment value.

[0119] In an optional embodiment, since random forests have an ensemble effect, the node purity increment of a single tree may be random. Therefore, the model can calculate the average node purity increment of the indicator across all decision trees as the final node purity increment value of the real-time evaluation indicator.

[0120] Step S2042: Normalize the purity increment values ​​of multiple nodes to obtain the weight values ​​of multiple real-time evaluation indicators.

[0121] In one optional embodiment, since the node purity increment values ​​of different real-time evaluation indicators may have different magnitudes, directly using the original increment value as the weight would lead to the large-magnitude indicator overdoing the evaluation result. Therefore, in this embodiment, normalization is used to map all increment values ​​to the 0-1 range, and the sum of all indicator weights is 1, thus finally obtaining the real-time evaluation indicator weight value.

[0122] The normalization method can be linear normalization, etc.

[0123] In some optional implementations, step S205 above includes: Step S2051: Determine the habitat suitability evaluation value of the coastal wetland to be evaluated based on multiple real-time evaluation index values ​​and multiple real-time evaluation index weight values.

[0124] In some optional implementations, the scores of each indicator are determined according to the different evaluation index values ​​corresponding to each level, and then the final evaluation result is obtained by weighting according to the evaluation index weight values, as shown in the following relationship (2): (2) In the formula: Indicates the habitat suitability assessment value; Indicators The score; Indicators The weight.

[0125] Step S2052: Using the habitat suitability assessment value, the ecosystem status of the coastal wetland to be evaluated is assessed to obtain the ecosystem status assessment results of the coastal wetland to be evaluated.

[0126] In some alternative implementations, the ecosystem status of the wetland to be evaluated can be determined based on the obtained habitat suitability assessment value and in conjunction with the designed grading criteria, as shown in Table 1 below: Table 1. Criteria for Habitat Suitability Index Evaluation

[0127] Furthermore, by determining the assessment results of the ecosystem status of the coastal wetlands to be evaluated, it is of guiding significance for assessing the restoration effect after treatment and optimizing the spatial layout of habitats.

[0128] In one example, to address the technical problem that existing coastal wetland habitat suitability assessments are too one-sided and unreliable, leading to unreasonable assessment results, a method for assessing coastal wetland habitat suitability is provided, which includes the following steps: Step 1: Data collection.

[0129] The required indicator parameters were determined through literature review.

[0130] Collect data on coastal wetland vegetation and benthic organisms through historical data or field surveys. These data mainly include environmental indicators (sedimentation rate, elevation, water temperature, salinity, dissolved oxygen, redox potential, sediment organic carbon, sediment particle size, sediment total nitrogen content, sediment total phosphorus content, sediment potassium content, etc.) and biological indicators (vegetation type, vegetation density, benthic animal habitat density, benthic animal biomass, benthic animal diversity index, etc.).

[0131] Step 2: Data quality control and preprocessing.

[0132] According to the research schedule requirements, mathematical statistics methods were used to remove outliers from the data.

[0133] After the environmental factors pass the normality test, correlation analysis is performed on the screened environmental factors to test for multicollinearity. For environmental factors with high correlation, it is considered that the factor has a multicollinearity problem, and one factor is retained, while the other multicollinear factors should be discarded.

[0134] Assign values ​​to discrete categorical data, such as substrate type and vegetation type.

[0135] Step 3: Determine the evaluation indicators based on the GAM model results.

[0136] The GAM model is a nonparametric extension of the generalized linear model, and its advantage is that it can directly handle the nonlinear relationship between the response variable and multiple explanatory variables. Furthermore, the expression of the GAM model is shown in the above relation (1).

[0137] Using the Akaike Information Content Criterion (AIC) as the model selection criterion, starting with a single variable, other environmental variables were added to the model with the smallest AIC value in turn. The changes in AIC values ​​under different combinations of environmental variables were observed. The combination of independent variables with lower AIC values ​​was used as the explanatory variables, and the model with the smallest AIC value was selected as the optimal model.

[0138] GAM models were established for different biological groups (vegetation, benthic organisms, etc.), and the model results were visualized.

[0139] Using the selected main environmental factors as explanatory variables and vegetation biological indicators as response variables, a GAM model was constructed to obtain smooth regression functions of the explanatory variables and to obtain effect diagrams showing the significant impact of single environmental factors on vegetation biological indicators.

[0140] Using the selected major environmental factors and vegetation biological indicators as explanatory variables and benthic biological indicators as response variables, a GAM model was constructed to obtain smooth regression functions for the explanatory variables and effect diagrams showing the significant impact of single-factor environmental indicators and vegetation biological indicators on benthic organisms.

[0141] The significance of the impact of each explanatory variable on the response variable and the goodness of fit of the model were analyzed. Based on the significance relationship between the single-factor indicators and biological indicators, the evaluation indicators for the suitability of coastal wetland habitats were determined.

[0142] Step 4: Determine the threshold values ​​for the evaluation indicators and assign them to the appropriate values.

[0143] Thresholds are determined based on the smoothed regression function of the explanatory variables, combined with historical measured data and literature research data over many years.

[0144] For continuously distributed data: (1) For smooth regression functions that are not monotonic, the "suitable" range threshold is the interval (x1, x2) of explanatory variables corresponding to more than 75% of the maximum value of the dependent variable, and the threshold is 25%-50% of the maximum value of the explanatory variable (x1, x2). 01 x1)(x2, x 02 The threshold value is set as the "suitable" range, used to explain the minimum values ​​of variables x0 and x. 01 Between and x 02 and the maximum value of the explanatory variable x m The threshold is within the "normal" range.

[0145] (2) For the monotonically changing smooth regression function, combined with many years of historical measured data and literature research, the 38th quantile (Q) is used. 38 ) and the 63rd quantile (Q 63 The "suitable" range threshold is defined as between 0 and 1, with the first quartile (Q) as the threshold. 25 ) and the 38th quantile (Q 38 Between ) and the 63rd quantile (Q) 63 ) and the third quartile (Q 75 The threshold between ) is the "suitable" range, with the minimum value (Q1) and the first quartile (Q) as the thresholds. 25 Between and the third quartile (Q) 75 ) and maximum value (Q) 100 The range between ) is the "normal" threshold.

[0146] Furthermore, for discrete categorical data, based on the regression function, the explanatory variables corresponding to more than 75% of the maximum value of the dependent variable are considered "suitable," those corresponding to 50% to 75% of the maximum value of the dependent variable are considered "relatively suitable," and those corresponding to less than 50% of the maximum value of the dependent variable are considered "average."

[0147] Furthermore, the 1, 2, 3 assignment method is used to assign values ​​to the indicators: after determining the threshold, the indicators are assigned scores of 3, 2, and 1 based on whether they are "suitable", "relatively suitable", or "average".

[0148] Step 5: Use the random forest model to determine the weights of the evaluation indicators.

[0149] Using the significance and R-squared of the regression equations of the influencing factors selected by the GAM model as the dependent variable and biological indicators as the explanatory variables, a random forest model was trained. 70% of the samples were randomly selected as the training set, and the remaining 30% as the prediction set, generating 1000 classic decision trees.

[0150] IncNodePurity measures the classification purity of nodes in a decision tree. When a feature is added, this metric indicates the increase in node purity compared to the state without that feature. A higher IncNodePurity value indicates that the feature contributes significantly to the node's classification, making the node purer.

[0151] The node purity increment (IncNodePurity) is used as a metric, and the IncNodePurity value is normalized and used as the evaluation weight for each metric.

[0152] Step Six: Calculate the evaluation index.

[0153] Based on the scores of different environmental factor indicators, each level is assigned a score, and the final evaluation result is obtained by weighting the indicators according to their weights, as shown in the above formula (2).

[0154] The coastal wetland habitat suitability assessment method provided in this example has the following beneficial effects: 1. This case study fills a gap in the evaluation of multi-system habitat suitability of coastal wetlands. By considering both vegetation adaptability and benthic animal adaptability, it provides a more comprehensive understanding of coastal mudflat ecosystems and has strong applicability.

[0155] 2. This example uses the GAM model to analyze and filter indicators, which can obtain reasonable and scientific evaluation indicators and ensure the objectivity of the evaluation indicators.

[0156] 3. This example uses the core indicator weights optimized by random forest to ensure the scientific nature and reliability of the evaluation indicator weights.

[0157] In an optional implementation, based on the coastal wetland habitat suitability assessment method provided in the above examples, a specific embodiment is provided, including: Step 1: Data collection.

[0158] The required indicator parameters were determined through relevant literature review, and relevant data were collected through field surveys and historical records. Three field surveys were conducted in June, October, and December 2024.

[0159] Benthic animal sampling was conducted using GPS. Benthic animal samples were collected at high, mid, and low tide zones, with one sample collected from each location. The sampling area was 0.25 cm x 0.25 cm. The collected benthic samples were vortexed and sorted using a 0.5 mm mesh sieve. All samples were fixed with 5% formalin solution. The species number, density, and biomass of benthic organisms were counted, and parameters such as diversity index H′, evenness J, and richness d were calculated.

[0160] A simultaneous survey was conducted on the wetland vegetation at each sampling section, including recording vegetation community type, plant height, density, and cover.

[0161] Water quality monitoring: Water quality was monitored simultaneously at the sampling sites for benthic animals in the high, middle and low tide zones using a portable multi-parameter water quality analyzer. The monitored indicators included water temperature, dissolved oxygen (optical), salinity, pH, oxidation-reduction potential (ORP), and conductivity (SPE).

[0162] Substrate sampling: Substrate samples were collected simultaneously at benthic animal sampling sites in the high, middle, and low tide zones, sealed in sample bags, and sent to the laboratory for analysis. Monitoring parameters included total organic carbon (TOC), total nitrogen, total phosphorus, potassium, petroleum hydrocarbons, and median particle size.

[0163] Step 2: Data quality control and preprocessing.

[0164] According to the research schedule requirements, mathematical statistics methods were used to remove outliers from the data.

[0165] In this embodiment, for normally distributed data, the standard deviation of the sample is first calculated. Then check the absolute value of the difference between each value and the mean. If the absolute value is greater than 3 times the standard deviation... If the value is not found, it is considered an outlier and is removed.

[0166] For non-normally distributed data, calculate the first quartile (Q). 25 ) and the third quartile (Q 75 Then calculate R=Q 75 -Q 25 It will be lower than Q. 25 -1.5R or higher than Q 75 Data with a value of +1.5R is considered outlier and is removed.

[0167] Furthermore, after the environmental factors passed the normality test, correlation analysis was performed on the screened environmental factors to test for multicollinearity. For environmental factors with high correlation (in this embodiment, |r|>0.75 is used as the criterion), it is considered that the factor has a multicollinearity problem, and one factor is retained, while other multicollinear factors should be discarded. Finally, the siltation and sedimentation data, elevation, organic carbon, total phosphorus, sorting coefficient, median particle size, pH, salinity, and dissolved oxygen (DO1) are retained as the main environmental factors.

[0168] Furthermore, the vegetation types were assigned values: the bare beach, the sea buckthorn area, and the reed area were assigned values ​​of 3, 2, and 1, respectively.

[0169] Furthermore, the substrate types are assigned values: silt, clayey silt, sandy silt, silty sand, and sand are assigned values ​​of 1, 2, 3, 4, and 5, respectively.

[0170] Step 3: Determine the evaluation indicators based on the GAM model results.

[0171] The GAM model is a nonparametric extension of the generalized linear model. Its advantage is that it can directly handle the nonlinear relationship between the response variable and multiple explanatory variables. The expression of the GAM model is shown in the above relation (1).

[0172] Furthermore, using the Akaike Information Content Criterion (AIC) as the model selection criterion, starting with a single variable, other environmental variables were added sequentially to the model with the smallest AIC value, and the model with the smallest AIC value was selected as the optimal model.

[0173] Furthermore, the GAM model was performed using the mgcv package in the R language, and the model results were visualized.

[0174] Furthermore, using major environmental factors such as siltation data, elevation, organic carbon, total phosphorus, sorting coefficient, median particle size, pH, salinity, dissolved oxygen (DO1), and substrate type as explanatory variables, and vegetation type as the response variable, a GAM model was constructed to obtain smooth regression functions of the explanatory variables, and an effect diagram showing the significant impact of single-factor environmental indicators on vegetation type was obtained.

[0175] Furthermore, using key environmental factors such as sedimentation data, elevation, organic carbon, total phosphorus, sorting coefficient, median particle size, pH, salinity, dissolved oxygen (DO1), and substrate type, along with vegetation type, as explanatory variables, and the benthic biodiversity index as the response variable, a GAM model was constructed. Smooth regression functions for the explanatory variables were obtained, resulting in an effect diagram showing the significant impact of single environmental factors and vegetation type on the benthic biodiversity index. In this embodiment, p < 0.05 was used as the significance criterion.

[0176] Furthermore, the GAM single-factor model was used to analyze the significance of the influence of each explanatory variable on the response variable and the goodness of fit of the model. Based on the significance relationship between the single-factor indicators and biological indicators (p<0.05), the evaluation indicators for the suitability of coastal wetland habitats were determined.

[0177] Furthermore, erosion and sedimentation changes, elevation, and salinity showed significant correlations with vegetation type (p<0.05), and vegetation type, substrate type, and pH showed significant correlations with benthic biodiversity index (p<0.05). Therefore, erosion and sedimentation changes, elevation, and salinity were used as indicators for evaluating vegetation habitat adaptability, while vegetation type, substrate type, and pH were used as indicators for evaluating benthic organism habitat adaptability.

[0178] Step 4: Determine the threshold values ​​for the evaluation indicators and assign them to the appropriate values.

[0179] Thresholds are determined based on the smoothed regression function of the explanatory variables, combined with historical measured data and literature research data over many years.

[0180] For continuously distributed data: (1) For smooth regression functions that are not monotonic, the "suitable" range threshold is the interval (x1, x2) of explanatory variables corresponding to more than 75% of the maximum value of the dependent variable, and the threshold is 25%-50% of the maximum value of the explanatory variable (x1, x2). 01 x1)(x2, x 02 The threshold value is set as the "suitable" range, used to explain the minimum values ​​of variables x0 and x. 01 Between and x 02 and the maximum value of the explanatory variable x m The threshold is within the "normal" range.

[0181] (2) For the monotonically changing smooth regression function, combined with many years of historical measured data and literature research, the 38th quantile (Q) is used. 38 ) and the 63rd quantile (Q 63 The "suitable" range threshold is defined as between 0 and 1, with the first quartile (Q) as the threshold. 25 ) and the 38th quantile (Q 38 Between ) and the 63rd quantile (Q) 63 ) and the third quartile (Q 75 The threshold between ) is the "suitable" range, with the minimum value (Q1) and the first quartile (Q) as the thresholds. 25 Between and the third quartile (Q) 75 ) and maximum value (Q) 100 The range between ) is the "normal" threshold.

[0182] Furthermore, for discrete categorical data, based on the regression function, the explanatory variables corresponding to more than 75% of the maximum value of the dependent variable are considered "suitable," those corresponding to 50% to 75% of the maximum value of the dependent variable are considered "relatively suitable," and those corresponding to less than 50% of the maximum value of the dependent variable are considered "average."

[0183] Furthermore, the 1, 2, 3 scoring method is used to assign values ​​to the indicators: After determining the threshold, the indicators are scored 3, 2, and 1 according to the criteria of "suitable", "relatively suitable", and "average", as shown in Table 2 below: Table 2

[0184] Step 5: Use the random forest model to determine the weights of the evaluation indicators.

[0185] Using the influencing factors selected by the GAM model as dependent variables and biological indicators as explanatory variables, a random forest model was trained. 70% of the samples were randomly selected as the training set, and the remaining 30% as the prediction set, generating 1000 classic decision trees.

[0186] IncNodePurity measures the classification purity of nodes in a decision tree. When a feature is added, this metric indicates the increase in node purity compared to the state without that feature. A higher IncNodePurity value indicates that the feature contributes significantly to the node's classification, making the node purer.

[0187] The node purity increment (IncNodePurity) is used as a metric, and the IncNodePurity value is normalized and used as the evaluation weight for each metric.

[0188] Furthermore, the random forest model was performed using the randomForest package in R, and the model calculation results are shown in Table 3 below.

[0189] Table 3. Model Calculation Results

[0190] Step Six: Calculate the evaluation index.

[0191] The scores of environmental factor indicators in different regions are assigned to each level to determine the score of each indicator. The final evaluation result is obtained by weighting the indicators according to their weights.

[0192] Furthermore, taking sample line 1 as an example for evaluation, as shown in Table 4 below.

[0193] Table 4. Evaluation Results of Sample Line 1

[0194] Furthermore, as shown in Table 4, the vegetation adaptability of transect 1 is (0.82+0.90+0.82) / 3=0.85, which is the "suitable" level.

[0195] Furthermore, the benthic animal suitability of transect 1 is (0.85+0.85+0.74) / 3=0.81, which is at the "suitable" level.

[0196] This embodiment also provides a coastal wetland ecosystem status assessment device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0197] This embodiment provides a device for assessing the status of a coastal wetland ecosystem, such as... Figure 3 As shown, the device includes: The acquisition module 301 is used to acquire the target real-time biological dataset and the target historical biological dataset of the coastal wetland to be evaluated. The target real-time biological dataset includes the target real-time environmental index dataset and the target real-time biological index dataset.

[0198] The first processing module 302 is used to process the target real-time biological dataset through a first generalized additive model and a second generalized additive model to obtain a first smooth regression function and a second smooth regression function. The first smooth regression function is used to characterize the influence of environmental single-factor indicators on vegetation biological indicators, and the second smooth regression function is used to characterize the influence of environmental single-factor indicators and vegetation biological indicators on benthic organisms.

[0199] The determination module 303 is used to determine multiple real-time evaluation index values ​​based on the target historical biological dataset, the first smooth regression function, and the second smooth regression function.

[0200] The second processing module 304 is used to obtain multiple real-time evaluation index weight values ​​based on the multiple real-time evaluation index values ​​and the target real-time biological index dataset through a random forest model.

[0201] The evaluation module 305 is used to evaluate the ecosystem status of the coastal wetland to be evaluated based on the multiple real-time evaluation index values ​​and the multiple real-time evaluation index weight values, and obtain the ecosystem status evaluation result of the coastal wetland to be evaluated.

[0202] The coastal wetland ecosystem status assessment device provided in this embodiment of the invention can execute the coastal wetland ecosystem status assessment method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules are the same as in the corresponding embodiments described above, and will not be repeated here.

[0203] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0204] The following is a detailed reference. Figure 4This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 401, which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) 402 or a program loaded from memory 408 into random access memory (RAM) 403. RAM 403 also stores various programs and data required for the operation of the electronic device. The processor 401, ROM 402, and RAM 403 are interconnected via bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0205] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0206] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 409, or installed from a memory 408, or installed from a ROM 402. When the computer program is executed by the processor 401, it performs the functions defined in the coastal wetland ecosystem status assessment method of the embodiments of the present invention.

[0207] Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0208] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the coastal wetland ecosystem status assessment method shown in the above embodiments is implemented.

[0209] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0210] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for assessing the condition of a coastal wetland ecosystem, characterized by, The method comprises: acquiring a target real-time biological data set of a coastal wetland to be evaluated, the target real-time biological data set comprising a target real-time environmental index data set and a target real-time biological index data set; based on the target real-time biological data set, processing through a first generalized additive model and a second generalized additive model to obtain a first smooth regression function and a second smooth regression function, the first smooth regression function being used to represent the influence of an environmental single-factor index on a vegetation biological index, and the second smooth regression function being used to represent the influence of an environmental single-factor index and a vegetation biological index on benthic organisms; determining a plurality of real-time evaluation index values according to the target historical biological data set, the first smooth regression function and the second smooth regression function; based on the plurality of real-time evaluation index values and the target real-time biological index data set, processing through a random forest model to obtain a plurality of real-time evaluation index weight values; evaluating the ecosystem condition of the coastal wetland to be evaluated according to the plurality of real-time evaluation index values and the plurality of real-time evaluation index weight values to obtain an ecosystem condition evaluation result of the coastal wetland to be evaluated.

2. The method of claim 1, wherein, The method further comprises: acquiring a target real-time biological data set of a coastal wetland to be evaluated, the target real-time biological data set comprising a target real-time environmental index data set and a target real-time biological index data set; acquiring an initial real-time biological data set of the coastal wetland to be evaluated; removing abnormal values in the initial real-time biological data set to obtain a first real-time biological data set; performing multicollinearity processing on environmental index data in the first real-time biological data set to obtain a second real-time biological data set; 3. The method of claim 1, wherein, performing assignment processing on discrete data in the second real-time biological data set to obtain the target real-time biological data set. The method further comprises: acquiring a historical environmental index data set and a historical biological index data set in the target historical biological data set, the historical biological index data set comprising a historical vegetation biological index data set and a historical benthic biological index data set; based on the Akaike information criterion, constructing the first generalized additive model by taking the historical environmental index data set as an explanatory variable and the historical vegetation biological index data set as a response variable; 4. The method of claim 1, wherein, based on the Akaike information criterion, constructing the second generalized additive model by taking the historical environmental index data set and the historical vegetation biological index data set as explanatory variables and the historical benthic biological index data set as a response variable. According to the target historical biological data set, the first smooth regression function and the second smooth regression function, determining a plurality of real-time evaluation index values, comprising: determining a plurality of coastal wetland habitat suitability evaluation indexes according to the first smooth regression function and the second smooth regression function; determining a plurality of evaluation index range thresholds according to the target historical biological data set, the first smooth regression function and the second smooth regression function; 5. The method of claim 3, wherein, performing assignment processing on the plurality of coastal wetland habitat suitability evaluation indexes according to the plurality of evaluation index range thresholds to obtain the plurality of real-time evaluation index values. The method further comprises: acquiring a plurality of historical evaluation index values; Based on the node purity increment, the plurality of historical evaluation index values are taken as the dependent variables, and the historical biological index data set is taken as the explanatory variables to construct the random forest model.

6. The method of claim 5, wherein, Based on the plurality of real-time evaluation index values and the target real-time biological index data set, a plurality of real-time evaluation index weight values are obtained through random forest model processing, including: The plurality of real-time evaluation index values and the target real-time biological index data set are input into the random forest model to obtain a plurality of node purity increment values. The plurality of node purity increment values are normalized to obtain the plurality of real-time evaluation index weight values.

7. The method of claim 1, wherein, According to the plurality of real-time evaluation index values and the plurality of real-time evaluation index weight values, the ecosystem condition of the coastal wetland to be evaluated is evaluated to obtain an ecosystem condition evaluation result of the coastal wetland to be evaluated, including: According to the plurality of real-time evaluation index values and the plurality of real-time evaluation index weight values, the habitat suitability evaluation value of the coastal wetland to be evaluated is determined. The habitat suitability evaluation value is used to evaluate the ecosystem condition of the coastal wetland to be evaluated to obtain an ecosystem condition evaluation result of the coastal wetland to be evaluated.

8. A coastal wetland ecosystem condition assessment device, characterized by, The device comprises: An acquisition module is configured to acquire a target real-time biological data set and a target historical biological data set of a coastal wetland to be evaluated, wherein the target real-time biological data set comprises a target real-time environmental index data set and a target real-time biological index data set; A first processing module is configured to process the target real-time biological data set through a first generalized additive model and a second generalized additive model to obtain a first smooth regression function and a second smooth regression function, wherein the first smooth regression function is used to represent the influence of an environmental single-factor index on a vegetation biological index, and the second smooth regression function is used to represent the influence of an environmental single-factor index and a vegetation biological index on benthic organisms; A determination module is configured to determine a plurality of real-time evaluation index values according to the target historical biological data set, the first smooth regression function, and the second smooth regression function; A second processing module is configured to process the plurality of real-time evaluation index values and the target real-time biological index data set through a random forest model to obtain a plurality of real-time evaluation index weight values; An evaluation module is configured to evaluate the ecosystem condition of the coastal wetland to be evaluated according to the plurality of real-time evaluation index values and the plurality of real-time evaluation index weight values to obtain an ecosystem condition evaluation result of the coastal wetland to be evaluated.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions, and the computer instructions are used to make the computer execute the coastal wetland ecosystem condition evaluation method in any one of claims 1 to 7.

10. A computer program product, characterised in that, The computer readable storage medium stores computer instructions, and the computer instructions are used to make the computer execute the coastal wetland ecosystem condition evaluation method in any one of claims 1 to 7.

Images