Method for monitoring and evaluating terrestrial ecosystem service functions

Abstract

The application provides a method for monitoring and evaluating the service function of a terrestrial ecosystem, which comprises: generating biological feature vectors and environmental feature vectors according to biological data and environmental data collected in an ecological protection zone; performing species identification on the biological feature vectors based on a biological large model and performing environmental analysis on the environmental feature vectors based on an environmental large model to obtain information about animals and plants and an environmental quality evaluation situation in the ecological protection zone; and performing comprehensive reasoning analysis on the biological, environmental and comprehensive feature vectors and the information about animals and plants and the environmental quality evaluation situation in the ecological protection zone based on an ecosystem evaluation model to evaluate the service function of the ecosystem of the ecological protection zone, wherein the comprehensive feature vectors are determined based on the fusion of the biological and environmental feature vectors and dynamically adjusted weights. The application can comprehensively understand the overall ecological condition of the ecological protection zone by using the large model technology and provide a more comprehensive, accurate, dynamic and scientific evaluation of the service function of the ecosystem.

Description

Technical Field

[0001] The present application relates to the field of ecological technology, and in particular to a method for monitoring and evaluating the service functions of terrestrial ecosystems. Background Art

[0002] With the rapid development of society and economy and the continuous growth of population, the demand for ecosystem services is increasing. However, the sustainability of ecosystem services faces many challenges. Therefore, accurately assessing ecosystem services is of great significance for formulating scientific and reasonable ecological protection policies.

[0003] At present, when conducting ecosystem service function assessments, the focus is mainly on the assessment of one or a few service functions. Although this can reflect the service capacity of a certain aspect of the ecosystem to a certain extent, it cannot fully reflect the service function value of the ecosystem as a whole.

[0004] Moreover, the evaluation of ecosystem service functions requires a large amount of data support, and data integration is difficult. This makes the evaluation of ecosystem service functions face many challenges in terms of data basis, affecting the accuracy and reliability of the evaluation results.

[0005] Since ecosystem service functions are changing dynamically, existing assessment methods often only focus on the status of ecosystem service functions at a specific point in time and lack monitoring and assessment of the dynamic changes in ecosystem service functions.

[0006] In summary, existing ecosystem service function assessment methods have many shortcomings in terms of comprehensiveness, accuracy, and dynamism, and are unable to meet the needs of sustainable ecosystem management and decision-making. Summary of the Invention

[0007] In view of the above problems, the embodiments of the present application provide a method for monitoring and evaluating terrestrial ecosystem service functions that overcomes the above problems or at least partially solves the above problems.

[0008] In a first aspect, an embodiment of the present application provides a method for monitoring and evaluating terrestrial ecosystem service functions, comprising:

[0009] Generate biological feature vectors and environmental feature vectors based on biological data and environmental data collected in the ecological protection area, wherein the biological data includes at least image data, audio data, and positioning data of plants and animals, and the environmental data includes at least meteorological data and physical environment data;

[0010] Performing species identification on the biological feature vector based on the biological macro model to obtain information on plants and animals within the ecological protection zone, and performing environmental analysis on the environmental feature vector based on the environmental macro model to obtain an environmental quality assessment within the ecological protection zone;

[0011] Based on the ecosystem assessment model, a comprehensive reasoning analysis is performed on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on plants and animals in the ecological protection zone, and the environmental quality assessment, to evaluate the ecosystem service function of the ecological protection zone. The comprehensive feature vector is determined based on the fusion of the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weights and environmental weights. The biological weights and environmental weights are dynamically adjusted based on the contribution of the biological data and the environmental data to the ecosystem service function assessment.

[0012] In a second aspect, an embodiment of the present application provides a system for monitoring and evaluating terrestrial ecosystem service functions, comprising:

[0013] a generation module for generating a biological feature vector and an environmental feature vector based on biological data and environmental data collected in the ecological protection area, wherein the biological data includes at least image data, audio data, and positioning data of plants and animals, and the environmental data includes at least meteorological data and physical environment data;

[0014] a processing and acquisition module, configured to perform species identification on the biological feature vector based on a biological macro model to obtain information on plants and animals within the ecological protection zone, and to perform environmental analysis on the environmental feature vector based on an environmental macro model to obtain an environmental quality assessment within the ecological protection zone;

[0015] An analysis and evaluation module is used to conduct comprehensive reasoning and analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on plants and animals in the ecological protection zone, and the environmental quality assessment based on the ecosystem assessment model, and evaluate the ecosystem service function of the ecological protection zone. The comprehensive feature vector is determined based on the fusion of the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weights and environmental weights. The biological weights and environmental weights are dynamically adjusted based on the contribution of the biological data and the environmental data to the ecosystem service function assessment.

[0016] The technical solution of the embodiment of the present application generates biological feature vectors and environmental feature vectors based on biological data and environmental data collected in the ecological protection zone, analyzes the biological feature vectors based on the biological big model to obtain information on animals and plants in the ecological protection zone, and analyzes the environmental feature vectors based on the environmental big model to obtain the environmental quality assessment situation in the ecological protection zone; after generating a comprehensive feature vector based on the dynamic fusion of the biological feature vector and the environmental feature vector, a comprehensive analysis is performed on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on animals and plants in the ecological protection zone, and the environmental quality assessment situation based on the ecosystem assessment model to evaluate the ecosystem service function of the ecological protection zone. The big model technology can be used to more comprehensively understand the overall ecological status of the ecological protection zone, provide a more comprehensive, accurate, dynamic and scientific ecosystem service function assessment, and provide strong support for ecological protection and sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram showing a method for monitoring and evaluating terrestrial ecosystem service functions provided by an embodiment of the present application;

[0018] Figure 2 A flowchart showing a specific implementation of the method for monitoring and evaluating terrestrial ecosystem service functions provided by an embodiment of the present application;

[0019] Figure 3 A schematic diagram showing a system for monitoring and evaluating terrestrial ecosystem service functions provided by an embodiment of the present application;

[0020] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] It should be understood that references throughout this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic associated with the embodiment is included in at least one embodiment of the present application. Therefore, the appearance of "in one embodiment" or "in an embodiment" throughout this specification does not necessarily refer to the same embodiment. Furthermore, these particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. The term "a plurality" in the embodiments of the present application may include two or more.

[0023] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0024] The present invention provides a method for monitoring and evaluating the service function of a terrestrial ecosystem. Figure 1 Shown, including:

[0025] Step 101: Generate biological feature vectors and environmental feature vectors based on biological data and environmental data collected in an ecological protection area. The biological data includes at least image data, audio data, and positioning data of animals and plants, and the environmental data includes at least meteorological data and physical environment data.

[0026] In this embodiment, a collection device is used to collect biological and environmental data within a selected ecological reserve during one or more specific time periods. The collection device, for example, is an unmanned device (e.g., drone or unmanned vessel) equipped with various sensors. By controlling the sensors on the unmanned device to collect data at a set frequency and adding attributes such as timestamps and location information, unified control of the sensors and synchronous data collection are achieved, thereby obtaining the required data.

[0027] The biological data collected by the collection equipment includes at least image data, audio data, and location data of the flora and fauna within the ecological reserve. The environmental data collected includes at least meteorological data and physical environment data. Specifically, environmental data includes, but is not limited to, temperature, humidity, soil conditions, water quality, light intensity, air pressure, wind speed, and season. By collecting environmental data, we can provide contextual information on the survival and activities of flora and fauna within the ecological reserve, which, combined with the biological data, can more comprehensively describe the regional characteristics of the ecological reserve.

[0028] Biological data collected in ecological reserves can be processed to generate biological feature vectors, and environmental feature vectors can be processed to generate environmental feature vectors. Biological feature vectors are used to indicate biological characteristics within an ecological reserve, while environmental feature vectors are used to indicate environmental characteristics within the reserve. By monitoring biological data within ecological reserves, information such as the species, abundance, and distribution of plants and animals within the reserve can be obtained. Monitoring environmental data within ecological reserves allows for environmental quality assessments, facilitating the timely identification of environmental issues within the reserve and providing a scientific basis for ecological protection and management.

[0029] Step 102: perform species identification on the biological feature vector based on the biological macro model to obtain information on plants and animals in the ecological protection zone; and perform environmental analysis on the environmental feature vector based on the environmental macro model to obtain an environmental quality assessment of the ecological protection zone.

[0030] After generating a biometric vector based on the collected biological data, species identification is performed on the biometric vector using a pre-built biometric model. The biometric model is a species identification model. By performing inference analysis on the biometric vector based on the biometric model, the model can obtain the species names and species name confidence scores corresponding to species within the ecological protection zone. The species name confidence score indicates the model's confidence in the predicted species name and is typically expressed as a probability value. For each species, the biometric model outputs at least one species name and at least one corresponding species name confidence score. After obtaining at least one species name and at least one species name confidence score corresponding to a particular species provided by the biometric model, the species name confidence score is modified based on prior information about the species. Prior information about a species refers to information about the characteristics, behavior, distribution, and other characteristics of a species that is known or assumed before specific observations or experiments are conducted. After modifying the species name confidence score using this prior information, the species name of the current species is determined based on the modified species name confidence score. For example, the species name with the highest confidence score is selected as the final species name to achieve species identification. By using the biological macromodel to analyze the biological feature vectors that indicate the biological characteristics in the ecological protection area, the species in the ecological protection area can be effectively identified, and then the information of animals and plants in the ecological protection area can be obtained.

[0031] After generating an environmental feature vector based on the collected environmental data, an environmental quality assessment is performed on the feature vector using a pre-built environmental macro model. The macro model is an environmental assessment model. By performing inference analysis on the environmental feature vector based on the macro model, an environmental quality assessment of the ecological protection zone can be obtained from the macro model. This environmental quality assessment output includes, for example, an environmental quality score, which is determined based on a comprehensive assessment of water quality, soil, and meteorological conditions within the ecological protection zone.

[0032] Furthermore, the environmental quality scores provided by the environmental big model include real-time scores and trend change scores, that is, the environmental quality assessment output by the environmental big model may include the assessed environmental quality change trends, for example, analyzing the changing trends of environmental characteristics over time, analyzing possible environmental anomalies (such as acid rain), and predicting environmental quality change trends in future periods.

[0033] By using the environmental big model to analyze the environmental characteristic vectors that indicate the environmental characteristics within the ecological protection zone, the environmental quality within the ecological protection zone can be effectively evaluated, and the environmental quality assessment situation within the ecological protection zone can be obtained.

[0034] Step 103: Based on the ecosystem assessment model, a comprehensive reasoning analysis is performed on the biological characteristic vector, the environmental characteristic vector, the comprehensive characteristic vector, the information on flora and fauna in the ecological protection zone, and the environmental quality assessment to evaluate the ecosystem service function of the ecological protection zone. The comprehensive characteristic vector is determined based on the fusion of the biological characteristic vector, the environmental characteristic vector, and the dynamically adjusted biological weights and environmental weights. The biological weights and environmental weights are dynamically adjusted based on the contribution of the biological data and the environmental data to the ecosystem service function assessment.

[0035] After obtaining the information of plants and animals in the ecological protection zone based on the biological big model and the environmental quality assessment situation in the ecological protection zone based on the environmental big model, the ecosystem assessment model is used to conduct comprehensive reasoning and analysis on the biological characteristic vectors, environmental characteristic vectors, comprehensive characteristic vectors, the information of plants and animals in the ecological protection zone and the environmental quality assessment situation to evaluate the ecosystem service function of the ecological protection zone.

[0036] The comprehensive feature vector is determined by integrating the biological feature vector, the environmental feature vector, and dynamically adjusted biological and environmental weights. The biological feature vector reflects information such as biodiversity and species richness, while the environmental feature vector encompasses environmental factors such as water quality, soil quality, and meteorological conditions. By integrating the biological and environmental feature vectors to form a comprehensive feature vector, the health of ecological reserves can be more comprehensively assessed. Furthermore, the comprehensive feature vector is not only based on the biological and environmental feature vectors but also takes into account the dynamically adjusted biological and environmental weights. Because the biological and environmental weights are dynamically adjusted based on the contribution of biological and environmental data to the assessment of ecosystem service functions, this dynamic adjustment mechanism allows for flexible adjustments based on the characteristics and conservation objectives of different ecological regions, making the assessment results more targeted.

[0037] The ecosystem assessment model in the embodiment of the present application is a large model for evaluating ecosystem service functions. The large model is used to intelligently and comprehensively analyze biological feature vectors, environmental feature vectors, comprehensive feature vectors, animal and plant information in ecological protection areas, and environmental quality assessments. It can provide a more comprehensive, accurate, dynamic and scientific ecosystem service function assessment, and provide strong support for ecological protection and sustainable development.

[0038] The biological feature vector and environmental feature vector in the embodiment of the present application belong to a collection of raw data, which contains detailed information about the organisms and environment in the ecological protection zone. The two vectors belong to basic data, and specific biological information (mainly animals and plants) and environmental quality assessments can be extracted from these data. The animal and plant information obtained based on the analysis of the biological feature vector belongs to a higher level of information, such as species diversity, the distribution of key species, etc.; the environmental quality assessments obtained based on the analysis of the environmental feature vector belong to a higher level of information, such as whether the water quality meets the standards, the degree of soil pollution, etc. Although the individual animal and plant information and environmental quality assessments provide important information about the ecological protection zone, they are the results of preliminary analysis and may lose some details and potential connections in the original data. While considering the animal and plant information and environmental quality assessments in the ecological protection zone, analyzing the biological feature vectors, environmental feature vectors, and comprehensive feature vectors can uncover more potential information and relationships.

[0039] By comprehensively analyzing biological characteristic vectors, environmental characteristic vectors, comprehensive characteristic vectors, information on plants and animals in ecological protection areas, and environmental quality assessments, we can reveal the complex interactions between organisms and the environment, and gain a more comprehensive understanding of the overall status of the ecosystem, thereby achieving a more comprehensive, accurate, and in-depth assessment of ecosystem service functions.

[0040] The purpose of ecosystem service function assessment is to evaluate the contribution of ecosystems to human society by analyzing biological data and environmental data. The ecosystem service functions of the ecological protection areas evaluated in the embodiments of this application include, for example, biodiversity, water purification, climate regulation, soil fertility maintenance, etc. By comprehensively and in-depth evaluating ecosystem service functions, more scientific decision-making support can be provided for ecological protection and resource management, thereby formulating effective protection strategies in a targeted manner.

[0041] The above implementation plan of the present application generates biological feature vectors and environmental feature vectors based on the biological data and environmental data collected in the ecological protection zone, analyzes the biological feature vectors based on the biological big model to obtain the information of animals and plants in the ecological protection zone, and analyzes the environmental feature vectors based on the environmental big model to obtain the environmental quality assessment situation in the ecological protection zone; after generating a comprehensive feature vector based on the dynamic fusion of the biological feature vector and the environmental feature vector, a comprehensive analysis is conducted on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information of animals and plants in the ecological protection zone and the environmental quality assessment situation based on the ecosystem assessment model to evaluate the ecosystem service function of the ecological protection zone. The big model technology can be used to more comprehensively understand the overall ecological status of the ecological protection zone, provide a more comprehensive, accurate, dynamic and scientific ecosystem service function assessment, and provide strong support for ecological protection and sustainable development.

[0042] The following describes the process of generating biological feature vectors and environmental feature vectors. When generating biological feature vectors and environmental feature vectors based on biological data and environmental data collected in ecological protection areas, the following steps are involved:

[0043] Preprocess the biological and environmental data collected in the ecological protection area to obtain the data to be analyzed;

[0044] A biometric feature vector is generated based on image features, audio features, and positioning features extracted from the data to be analyzed, and an environmental feature vector is generated based on water quality features, soil features, and meteorological features extracted from the data to be analyzed.

[0045] After acquiring biological and environmental data of the ecological protection area through data collection based on acquisition equipment, data preprocessing is performed on the collected data. This can provide relatively rich data while improving data quality and laying the foundation for subsequent data analysis.

[0046] The collected biological data includes image data, audio data, and positioning data of plants and animals within the ecological reserve. Image data preprocessing uses an image denoising algorithm to remove noise, and image enhancement and normalization are performed to improve image quality. Audio data preprocessing uses an audio denoising algorithm to remove background noise. Normalization and pre-emphasis can also be used to further process the audio. Pre-emphasis enhances high-frequency components in the audio signal, and normalization improves numerical stability. These audio processing techniques optimize audio data quality. Positioning data preprocessing involves converting longitude and latitude coordinates into two-dimensional coordinates and combining them with IMU (Inertial Measurement Unit) data through a weighted fusion algorithm to obtain more accurate positioning information. Environmental data preprocessing, for example, involves normalizing the environmental data to a uniform range.

[0047] The preprocessed image data can be stored in JPEG or PNG format, with timestamp and location information added as file name or metadata; the preprocessed audio data can be stored in WAV or MP3 format, with timestamp and location information added as file name or metadata; the preprocessed location data and environmental data can be stored in CSV or JSON format, including timestamp, longitude and latitude, environmental parameters and other information.

[0048] After preprocessing the collected data, the data to be analyzed is obtained, and a biometric feature vector is generated based on the image features, audio features, and positioning features extracted from the data to be analyzed. In addition, an environmental feature vector is generated based on the water quality features, soil features, and meteorological features extracted from the data to be analyzed.

[0049] Optionally, when generating a biometric feature vector based on image features, audio features, and positioning features extracted from the data to be analyzed, the method includes:

[0050] Extract image features, audio features, and positioning features with timestamps from the data to be analyzed, and align the extracted biometric features of each modality according to the timestamps;

[0051] assigning a first weight to each modality biometric feature based on the importance of each modality biometric feature;

[0052] The aligned biometric features of each modality are combined with the matched first weights to perform feature concatenation to obtain a first high-dimensional feature vector;

[0053] The weights of the biometric features of each modality in the first high-dimensional feature vector are dynamically adjusted to generate a biometric feature vector.

[0054] When extracting biometric features from the data to be analyzed, image features are extracted from image data, audio features are extracted from audio data, and positioning features are extracted from positioning data. For example, convolutional neural networks are used to extract features of plants and animals in images, such as texture, shape, and color; features such as the Mel-spectrum spectrum and Mel-frequency cepstral coefficients (MFCCs) of audio are extracted, which reflect the frequency and temporal characteristics of the audio; and the two-dimensional coordinates of the positioning data are extracted as features. After feature extraction, image features, audio features, and positioning features with precise timestamps are obtained. Based on the timestamps, feature alignment is performed on the biometric features of each modality, so that data at the same time point can be matched, ensuring data consistency across the time dimension.

[0055] After performing biometric feature extraction and aligning the biometric features of each modality, a corresponding first weight is assigned to each modality based on the importance of the biometric features of each modality, and the aligned biometric features of each modality are combined with the matching first weights to perform biometric feature splicing to obtain a first high-dimensional feature vector, which serves as the reference vector determined by multi-dimensional biometric feature fusion.

[0056] When assigning a corresponding primary weight to each modality of biometric features, the importance of each modality needs to be considered. Image features are typically used to identify the species, number, and distribution of organisms and are crucial for biodiversity and habitat assessments. Audio features are primarily used to identify the sounds of organisms, such as birdsong and insect calls, and are crucial for monitoring biological activity and behavioral patterns. Positioning features are used to determine the location and movement of organisms and are crucial for understanding their habitat ranges and migration paths. When assigning primary weights, weights can be manually set based on domain knowledge and relevant experience. For example, image features, considered the most critical in ecosystem assessments, can be assigned a higher weight, followed by audio features, and then positioning features. Weights can also be automatically learned through data-driven methods (such as machine learning). Cross-validation can also be used to evaluate model performance under different weight settings and select the optimal weight configuration.

[0057] After generating a first high-dimensional feature vector of associated biometric features based on feature concatenation, the weights of each modal biometric feature in the first high-dimensional feature vector are dynamically adjusted to generate the desired biometric feature vector. Because the importance of different modal biometric features may vary over time and with changing environmental conditions, the initially assigned first weights need to be dynamically adjusted to ensure that the biometric feature vector better reflects the current state of the ecosystem. Dynamic weight adjustment can be achieved through real-time monitoring data. For example, if audio features exhibit significant changes during a certain period, the weight of the audio features can be appropriately increased. Weights can also be adjusted based on historical data trends. For example, if the distribution range of a species fluctuates significantly within a certain season, the weight of the location features can be adjusted. Machine learning models (such as deep learning models) can also be used to automatically learn dynamic weight adjustments. For example, a neural network can be trained to automatically adjust weights based on dynamic changes in input features. During dynamic weight adjustment, the Transformer architecture can also be used to adaptively fuse biometric features from different modalities, dynamically adjusting the weights of each modality. The Transformer architecture, through its self-attention mechanism, dynamically focuses on the relationships between biometric features from different modalities, enabling adaptive feature fusion.

[0058] The following example illustrates the process of fusing multimodal biometric information and generating a biometric feature vector. When generating a biometric feature vector, the information entropy of each modal feature (image, audio, and location) is calculated in real time as a measure of feature importance. The information entropy of each modal feature is recalculated and its weights are updated every specified period (e.g., 30 minutes).

[0059] The calculation formula of information entropy is: H ( X )=-∑ p (x )log p ( x ); p(x) is the probability that the random variable X takes the value x, H (X) is a random variable that measures X An indicator of uncertainty. Weight Wi For example, the definition is: , H ( Xi ) is a random variable Xi The information entropy of It is the sum of the information entropy of all modal features.

[0060] After collecting biological data from a protected area for one hour, the information entropy of each modality in the first period (0-30 minutes) is: image feature entropy 2.3, audio feature entropy 1.8, positioning feature entropy 0.9. The sum of all modal feature information entropy is: 2.3+1.8+0.9=5.0. The weight in this period is: =2.3 / 5.0=0.46, =1.8 / 5.0=0.36, =0.9 / 5.0=0.18. In the second period (31-60 minutes), the audio feature entropy increases to 2.5. At this time, the sum of the information entropy of all modal features is: 2.3+2.5+0.9=5.7. The weight in this period is: =2.3 / 5.7 0.40, =2.5 / 5.7 0.44, =0.9 / 5.7 0.16.

[0061] By reasonably allocating the first weight and dynamically adjusting the weight, multimodal biological information can be better integrated, the quality of biological feature vectors can be improved, and the accuracy and robustness of ecosystem service function assessment can be improved.

[0062] Optionally, when generating an environmental feature vector based on water quality features, soil features, and meteorological features extracted from the data to be analyzed, the following steps are included:

[0063] Extract water quality features, soil features, and meteorological features with timestamps from the data to be analyzed, and align the extracted modal environmental features according to the timestamps;

[0064] assigning a second weight to each modal environment feature based on the importance of each modal environment feature;

[0065] The aligned features of each modal environment are combined with the matched second weights to perform feature splicing to obtain a second high-dimensional feature vector;

[0066] The weights of each modal environmental feature in the second high-dimensional feature vector are dynamically adjusted to generate an environmental feature vector.

[0067] When extracting environmental features from the data to be analyzed, water quality, soil, and meteorological characteristics are extracted. For example, for water resources, water quality (such as pH, dissolved oxygen, and turbidity) is extracted; for soil, fertility, moisture, and organic matter content are extracted; and for meteorological conditions, temperature, humidity, and wind speed are extracted.

[0068] After extracting environmental features and aligning the modal environmental features according to timestamps, a corresponding second weight is assigned to each modal environmental feature based on the importance of each modal environmental feature, and the aligned modal environmental features are combined with the matching second weights to perform environmental feature splicing to obtain a second high-dimensional feature vector, which serves as the reference vector determined by the fusion of multi-dimensional environmental features.

[0069] When assigning corresponding second weights to each modal environmental feature, the importance of each modal environmental feature needs to be considered. Water quality features are used to assess service functions such as water purification, water conservation, flood regulation, and biological habitats, and are crucial to maintaining the health and sustainable use of ecosystems. Soil features are used to assess soil fertility, soil erosion, and soil pollution, and are very important for service functions such as soil conservation and nutrient cycling. Meteorological features are used to assess climate conditions, such as temperature, humidity, and precipitation, and are very important for service functions such as climate regulation and biodiversity conservation. When assigning the initial second weight, based on the consideration of the importance of each environmental feature, an appropriate second weight can be assigned to each modal environmental feature according to the above-mentioned method of assigning the first weight.

[0070] After generating a second high-dimensional feature vector based on the concatenation of environmental features, the weights of each modal environmental feature in the second high-dimensional feature vector are dynamically adjusted to generate the desired environmental feature vector. Because the importance of different modal environmental features may vary over time and with changing environmental conditions, the second weights need to be dynamically adjusted. Dynamic adjustment of the weights of each modal environmental feature can be performed using the same approach as described above for adjusting the weights of each modal biometric feature. For example, weights can be dynamically adjusted based on real-time monitoring data. For example, if meteorological characteristics change significantly over a certain period of time, the weight of meteorological characteristics can be appropriately increased. Weights can also be dynamically adjusted based on trends in historical data. For example, if soil erosion in a certain area changes significantly over a certain season, the weight of soil characteristics can be increased. Weights can also be dynamically adjusted based on predicted trends in environmental characteristics. Machine learning models (such as deep learning models) can also be used to automatically learn dynamic weight adjustment. Weights can also be dynamically adjusted based on specific assessment task requirements. For example, when assessing water conservation, the weight of water quality characteristics can be increased, while when assessing soil conservation, the weight of soil characteristics can be increased. Similarly, when dynamically adjusting weights, the Transformer architecture can also be used to adaptively fuse different modal environment features and dynamically adjust the weights of each modal environment feature.

[0071] For example, the weights can be dynamically adjusted based on the future trends of various environmental characteristics (water quality, soil, and weather). The weight adjustment formula is, for example:

[0072]

[0073] in, is the learning rate (for example, set to 0.1); represents the weight at the current moment, Represents the weight of the previous moment; is the confidence interval at the current moment, indicating the uncertainty range of the prediction result at the current moment; is the confidence interval of the previous moment, indicating the uncertainty range of the prediction result at the previous moment. The smaller the confidence interval, the more certain the prediction result is; the larger the confidence interval, the higher the uncertainty of the prediction result; ( ) / Indicates the relative change of the confidence interval. This ratio reflects the degree of change of the confidence interval. If , indicating that the confidence interval becomes narrower, the prediction result becomes more certain, and the weight should be increased.

[0074] If the confidence interval of the water quality parameter prediction is narrowed (e.g. CI from 15% to 10%), or the soil parameter CI from 20% to 25%, the weight will be automatically adjusted: if the water quality weight is updated to , soil weight is updated to .

[0075] By reasonably allocating the second weight and dynamically adjusting the weight, multimodal environmental information can be better integrated, the quality of the environmental feature vector can be improved, and the accuracy and robustness of the ecosystem service function assessment can be improved.

[0076] The following describes the process of determining the biological macromodel and the environmental macromodel. When determining the biological macromodel: A first architecture model is pre-trained based on first biological sample data associated with plants and animals within the ecological reserve to determine the first pre-trained model; the first pre-trained model is adjusted based on labeled second biological sample data associated with plants and animals within the ecological reserve; and when the performance of the first pre-trained model is verified by biological validation data, the biological macromodel used for species identification is determined.

[0077] When determining a large biological model based on model training, the following stages are mainly involved: model pre-training, model fine-tuning, and model validation. The purpose of pre-training is to allow the model to learn universal feature representations on large-scale biological sample data. These feature representations can capture the basic patterns and structures in biological samples. The first biological sample data used in the model pre-training stage is unlabeled or weakly labeled feature data, which is large in quantity and covers a variety of plants and animals in ecological reserves. The first architecture model can be any deep learning model suitable for processing biological sample data. The feature representations learned through the pre-training model can serve as the basis for subsequent tasks, improving the model's adaptability to new tasks.

[0078] The purpose of model fine-tuning is to adapt the first pre-trained model to a specific species identification task. The second biological sample data used in the model fine-tuning phase is labeled feature data, including information such as species category. This data is typically relatively small in quantity but accurately labeled. By using this labeled second biological sample data for model fine-tuning, the model can learn more specific features related to species identification. Model fine-tuning typically involves freezing some layers of the first pre-trained model and training only the parameters of the last few layers or the entire model. This preserves the general features learned in the pre-training phase while learning specific features related to species identification. Model fine-tuning is used to optimize model parameters.

[0079] After fine-tuning the model parameters, the model validation phase begins. The purpose of validation is to evaluate the performance of the fine-tuned model on unseen data to ensure good generalization. The biological validation data used for model validation is annotated data independent of the training and fine-tuning data and is used to evaluate model performance. If the model's performance on the biological validation data meets the expected standards, the model is considered valid. In other words, a model that passes validation can be considered a biological model for species identification.

[0080] When determining the environmental big model: pre-train the second architecture model based on the first environmental sample data associated with the environment within the ecological protection zone to determine the second pre-trained model; adjust the second pre-trained model based on the labeled second environmental sample data associated with the environment within the ecological protection zone, the second environmental sample data at least includes labeled water quality data, soil data, and meteorological data; when the model performance of the second pre-trained model is verified by the environmental verification data, determine the environmental big model for environmental assessment.

[0081] The process of determining a large environmental model based on model training also involves model pre-training, model fine-tuning, and model validation. Data preparation is required before model pre-training. This data should include at least water quality data (such as pH, dissolved oxygen, and turbidity), soil data (such as soil fertility, moisture, and organic matter content), and meteorological data (such as temperature, humidity, and wind speed). Feature extraction is also required after data preparation, such as extracting numerical features of water quality parameters, soil parameters, and meteorological parameters. Following feature extraction, first environmental sample data associated with the environment within the ecological reserve is determined. Pre-training is then performed based on this first environmental sample data to determine the second pre-trained model. The goal of pre-training is to enable the model to learn universal feature representations on large-scale environmental sample data. These feature representations can capture the essential patterns and structures within the environmental samples. The second architecture model selected during the pre-training phase is a deep learning model suitable for processing environmental data, such as a Transformer architecture. The feature representations learned through pre-training can serve as a foundation for subsequent tasks, improving the model's adaptability to new tasks.

[0082] During the model fine-tuning phase, the second pre-trained model is adjusted using labeled second environmental sample data, optimizing model parameters to adapt it to environmental data monitoring and analysis tasks, such as water quality monitoring, soil monitoring, and meteorological condition monitoring. The goal of model fine-tuning is to adapt the second pre-trained model to specific environmental assessment tasks. By fine-tuning the model on labeled data, it can learn more specific features relevant to environmental assessment.

[0083] It's important to note that the environmental quality assessment output by the large-scale environmental model can include real-time environmental scores and environmental quality trends. To enable the large-scale environmental model to predict environmental trends, it's necessary to prepare data containing time series information. This data should cover environmental characteristics at multiple points in time, such as water quality, soil moisture, and meteorological conditions. Labeled data should include annotations such as environmental quality level and trend (increasing, decreasing, or stable). These annotations will be used for supervised learning, helping the model learn patterns of environmental change.

[0084] During the model validation phase, the validation set is used to evaluate model performance and ensure model accuracy and generalization. The goal of validation is to assess the performance of the fine-tuned model on unseen data and ensure good generalization. Environmental validation data is labeled data independent of the training and fine-tuning data and is used to evaluate model performance. Models that pass validation can be considered the larger environmental model for environmental assessment.

[0085] After determining the environmental macromodel, it can be used to assess water quality, soil conditions, and meteorological conditions. For example, in water quality assessment, water quality parameters are analyzed to assess water quality and its changing trends; in soil assessment, soil parameters are analyzed to assess soil quality and its changing trends; and in meteorological condition assessment, meteorological parameters are analyzed to assess meteorological conditions and their changing trends.

[0086] By determining the biological macro-model and using it to conduct species analysis, we can effectively identify species within the ecological protection zone and obtain information on animals and plants within the ecological protection zone; by using the environmental macro-model to monitor and analyze environmental data to assess environmental quality, it is helpful to discover environmental problems in a timely manner and provide a scientific basis for ecological protection and management.

[0087] The following describes the process of generating a comprehensive feature vector through feature fusion. To generate the comprehensive feature vector, initial weights are assigned to the biological and environmental feature vectors, based on the estimated importance of the biological and environmental data in ecosystem service function assessments. Based on these assigned initial weights, the biological and environmental feature vectors are concatenated to generate a high-dimensional baseline feature vector. The initial weights in the high-dimensional baseline feature vector are dynamically adjusted based on the real-time contribution of the biological and environmental data to ecosystem service function assessments to generate the comprehensive feature vector.

[0088] The comprehensive feature vector is a comprehensive feature vector that integrates biological features and environmental features. When generating the comprehensive feature vector, initial weights are first assigned to the biological feature vector and the environmental feature vector based on the estimated importance of biological data and environmental data in the assessment of ecosystem service functions.

[0089] When assigning initial weights, the following measures can be used: 1. Assign weights based on the experience of technical personnel in relevant fields and the contribution of biological and environmental factors to ecosystem service functions in similar ecosystems. 2. Assess the actual contribution of biological and environmental data to ecosystem service functions through statistical analysis and assign weights based on actual contributions. 3. Assign weights based on the category of ecosystem service functions; for example, if ecosystem service functions are primarily focused on provisioning services, the initial biological weight will be higher; if ecosystem service functions are primarily focused on regulating services, the initial environmental weight will be higher.

[0090] After assigning initial weights, the biological and environmental feature vectors are concatenated with their corresponding initial weights to generate a high-dimensional baseline feature vector. The initial weights in the high-dimensional baseline feature vector are then dynamically adjusted based on the real-time contribution of biological and environmental data to ecosystem service function assessments, generating a comprehensive feature vector.

[0091] When dynamically adjusting biotic and environmental weights, changes in ecosystem service functions can be considered. 1. Dynamically adjust biotic and environmental weights when changes in ecosystem service functions are detected. For example, if a region's water conservation function declines, a function primarily influenced by topography and vegetation cover, the environmental weight can be appropriately increased. 2. Dynamically adjust weights based on the objectives of the ecosystem service function assessment. For example, if the assessment goal is to prioritize the protection of endangered species, the biotic weight can be increased.

[0092] When dynamically adjusting biological and environmental weights, changes in biological and environmental data can also be considered. For example, newly collected biological and environmental data can be used to reassess the contribution of biological and environmental factors to ecosystem services and adjust weights accordingly. Alternatively, a dynamic model can be established to automatically adjust weights based on changes in biological and environmental data. Alternatively, biological and environmental weights can be dynamically adjusted based on assessment objectives. Furthermore, when dynamically adjusting weights, a Transformer architecture can be used to adaptively fuse biological and environmental feature vectors to dynamically adjust biological and environmental weights. For example, biological and environmental weights can be dynamically adjusted based on specific task objectives. In this scenario, baseline weights for different task objectives (assessment targets) are preset, and weights are adjusted based on real-time assessment needs. For example, if the assessment objective is species protection, the biological weight is 0.7 and the environmental weight is 0.3; if the assessment objective is environmental quality monitoring, the biological weight is 0.2 and the environmental weight is 0.8; and if the assessment objective is ecological function assessment, the biological weight is 0.5 and the environmental weight is 0.5.

[0093] Real-time weight calculation formula: W =β× W target+(1-β)× W current, β is the adjustment rate parameter (default 0.1), W Target is the weighted benchmark value preset according to the evaluation target. W Current is the weight value at the current moment before the real-time adjustment. The real-time weight calculation formula calculates the new weight by taking the weighted average of the target weight and the current weight. The adjustment rate parameter β controls the magnitude of the weight adjustment.

[0094] Initially, the biological weight is 0.5 and the environmental weight is 0.5. After switching the assessment target to species protection, the weights are gradually adjusted: in the first step, the biological weight is adjusted to 0.52 and the environmental weight is 0.48. In the second step, the biological weight is adjusted to 0.538 and the environmental weight is 0.462, and so on. After multiple iterations, the target weights are approached: the biological weight is 0.7 and the environmental weight is 0.3.

[0095] In another example, the Transformer's attention mechanism is used to automatically learn weights in the generation of a composite feature vector, with the weights determined by the attention score. For example, given the input features at a certain moment: biological features are [0.8, 0.2, 0.5] and environmental features are [0.6, 0.7, 0.3], the attention mechanism calculates: biological weight 0.6, environmental weight 0.4. The composite feature vector generated by integrating these features can be expressed as: 0.6 × [0.8, 0.2, 0.5] + 0.4 × [0.6, 0.7, 0.3] = [0.72, 0.4, 0.42].

[0096] When generating a comprehensive feature vector, initial weights are assigned to the biological feature vector and the environmental feature vector, and the weights of the two feature vectors are adjusted based on a dynamic adjustment mechanism. The weights can be flexibly adjusted according to the characteristics and protection goals of different ecological regions, making the evaluation results more targeted.

[0097] Existing technologies often struggle to integrate heterogeneous data from multiple sources (such as biological and environmental data), and their weights are fixed and cannot be dynamically adjusted. In an embodiment of the present invention, biological and environmental feature vectors are generated by integrating biological data (images, audio, and location) with environmental data (meteorological, soil, and water quality). A composite feature vector is then generated based on dynamically adjusted weights. This application addresses the data integration challenge and improves the flexibility and accuracy of assessments through a dynamic weight adjustment mechanism. This dynamic weight adjustment mechanism allows for flexible adjustments based on the data's real-time contribution to the ecosystem service function assessment, thereby more accurately reflecting the overall state of the ecosystem.

[0098] The following describes the process of evaluating the ecosystem service functions of ecological reserves based on a large model. Based on the ecosystem assessment model, a comprehensive reasoning analysis is conducted on biological feature vectors, environmental feature vectors, integrated feature vectors, information on flora and fauna within the ecological reserve, and environmental quality assessments to evaluate the ecosystem service functions of the ecological reserve. This includes:

[0099] Adjusting ecosystem assessment models based on ecosystem service function data to optimize them. Ecosystem service function data includes annotated biological and environmental data within ecological reserves;

[0100] Based on the optimized ecosystem assessment model, a comprehensive reasoning analysis is performed on the biological characteristic vectors, environmental characteristic vectors, comprehensive characteristic vectors, plant and animal information provided by the biological large model, and the environmental quality assessment provided by the environmental large model to generate ecosystem species diversity assessment results and ecosystem environment assessment results; the ecosystem environment assessment results include the assessment of the ecosystem's water purification function, the assessment of the ecosystem's climate regulation function, and the assessment of the ecosystem's soil fertility maintenance function.

[0101] Before conducting comprehensive reasoning and analysis of biological characteristic vectors, environmental characteristic vectors, comprehensive characteristic vectors, plant and animal information, and environmental quality assessment based on the ecosystem assessment model, it is necessary to use labeled ecosystem service function data to adjust the ecosystem assessment model for model optimization so that the model can adapt to the ecosystem service function assessment task.

[0102] After model optimization, a comprehensive reasoning analysis is conducted on the biological characteristic vectors, environmental characteristic vectors, comprehensive characteristic vectors, plant and animal information provided by the biological large model, and environmental quality assessment provided by the environmental large model based on the optimized ecosystem assessment model. This can reveal the complex interactions between organisms and the environment, as well as provide a more comprehensive understanding of the overall status of the ecosystem. The ecosystem assessment model outputs the ecosystem service function assessment results, which include the ecosystem species diversity assessment results and the ecosystem environment assessment results, to provide a more comprehensive, accurate and in-depth ecosystem service function assessment.

[0103] The species diversity assessment output by the ecosystem assessment model is determined based on information such as the identified plant and animal species, numbers, and distribution locations. The environmental assessment output by the ecosystem assessment model includes at least the assessment of the water purification function, the climate regulation function, and the soil fertility maintenance function. The water purification function is determined based on water quality parameters and plant and animal distribution; the climate regulation function is determined based on meteorological parameters and plant and animal distribution; and the soil fertility maintenance function is determined based on soil parameters and plant and animal distribution.

[0104] This solution achieves a comprehensive and dynamic assessment of ecosystem service functions through multi-dimensional data integration and dynamic weight adjustment, including species diversity, water purification, climate regulation, soil fertility maintenance and other functions. Compared with existing technologies that usually focus on static assessment of a single or a few service functions, this application achieves a comprehensive and dynamic assessment of ecosystem service functions through multimodal data fusion and dynamic analysis, which can more realistically reflect the changing trends of the ecosystem.

[0105] After evaluating the ecosystem service functions of the ecological protection zone based on the ecosystem assessment model, a visual assessment report is generated for display based on the ecosystem service function assessment results; the visual assessment report includes: a population distribution map showing the spatial distribution of species, a population density map showing the density distribution of species in different areas, a population dynamics map showing the changes in species numbers over time, an environmental quality score map, and a statistical map of the analysis results.

[0106] The species diversity assessment results provided by the ecosystem assessment model include, for example, population distribution data, population quantity data, plant and animal species, etc. Based on the above content, a population distribution map showing the spatial distribution of species and a population density map showing the density distribution of species in different regions can be generated. In combination with time information, a population dynamic change map showing the change in species quantity over time can be generated. The environmental assessment results provided by the ecosystem assessment model include, for example, the water purification function assessment, the climate regulation function assessment, and the soil fertility maintenance function assessment. Based on the above content, water quality scores, soil quality scores, and meteorological condition scores can be obtained to generate an environmental quality score map. Among them, the water quality score can include a real-time water quality score and a water quality change trend score. The soil quality score and meteorological condition score can also include a real-time score and a change trend score, so that the environmental quality score map can reflect the real-time environmental conditions and environmental trend changes.

[0107] Statistical graphs of analysis results can be generated based on population distribution maps, population density maps, population dynamic change maps, and environmental quality score maps. Population distribution maps, population density maps, population dynamic change maps, and environmental quality score maps are specific data visualization tools in ecosystem service function assessment, while statistical graphs of analysis results are a comprehensive summary of these specific data and analysis results.

[0108] Specifically, population distribution maps, population density maps, population dynamic change maps, and environmental quality score maps provide specific, itemized data and analysis results. The analysis result statistical maps integrate these specific data to provide more comprehensive statistical information. For example, population distribution maps and population density maps provide spatial distribution and density information of species. By summarizing this information, the overall characteristics of species distribution can be displayed.

[0109] Population distribution maps, population density maps, population dynamics maps, and environmental quality score maps show the biological and environmental conditions in the ecosystem from different perspectives. The analysis result statistical maps comprehensively analyze the analysis results from these multiple perspectives and can provide more comprehensive evaluation conclusions. For example, population dynamics maps and environmental quality score maps can respectively show the changing trends of species quantity and environmental quality. By combining these trends, the overall changes in ecosystem service functions can be evaluated.

[0110] Traditional assessment results are mostly presented in the form of reports or tables. This application uses visualization technology (such as charts and dynamic change graphs) to intuitively display the assessment results. By generating statistical graphs of analysis results, it provides comprehensive and intuitive ecosystem service function assessment results, which can provide a theoretical basis for more effective management and protection of ecosystems and enhance the practicality and operability of the results.

[0111] The following describes the method for monitoring and evaluating the terrestrial ecosystem service function according to the embodiment of the present application through a specific implementation process. Figure 2 As shown, the following steps are included:

[0112] Step 201: Preprocess the biological data and environmental data collected in the ecological protection area to obtain data to be analyzed.

[0113] Step 202: Extract image features, audio features, and positioning features with timestamps from the data to be analyzed, align the extracted biometric features of each modality according to the timestamps, and concatenate the aligned biometric features of each modality with the adjusted weights to generate a biometric feature vector.

[0114] Step 203: Extract water quality features, soil features, and meteorological features with timestamps from the data to be analyzed, align the extracted modal environmental features according to the timestamps, and concatenate the aligned modal environmental features with the adjusted weights to generate an environmental feature vector.

[0115] Step 204: Concatenate the biological feature vector and the environmental feature vector with the adjusted weights to generate a comprehensive feature vector.

[0116] Step 205: Perform species identification analysis on the biological feature vector based on the biological macro model to obtain information on animals and plants in the ecological protection area.

[0117] Step 206: Perform environmental analysis on the environmental feature vector based on the environmental macro model to obtain environmental quality assessment information within the ecological protection zone.

[0118] Step 207: Based on the ecosystem assessment model, a comprehensive reasoning analysis is performed on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the plant and animal information provided by the biological macro model, and the environmental quality assessment provided by the environmental macro model to generate an ecosystem service function assessment result.

[0119] Step 208: Generate a visual assessment report based on the ecosystem service function assessment results for display.

[0120] It should be noted that Figure 2 The illustrated process is a specific example, and the order of execution of the steps is not limited to this. The above implementation process uses large-scale modeling technology to comprehensively analyze biological feature vectors, environmental feature vectors, comprehensive feature vectors, information on flora and fauna within the ecological reserve, and environmental quality assessments to assess the ecosystem services of the ecological reserve. This allows for a more comprehensive understanding of the overall ecological status of the ecological reserve, providing a more comprehensive, accurate, dynamic, and scientific assessment of ecosystem services, and providing strong support for ecological protection and sustainable development.

[0121] The present application provides a system for monitoring and evaluating terrestrial ecosystem service functions, such as Figure 3 As shown, the system 300 for monitoring and evaluating terrestrial ecosystem service functions includes:

[0122] A generation module 301 is configured to generate a biological feature vector and an environmental feature vector based on biological data and environmental data collected in an ecological protection zone, wherein the biological data includes at least image data, audio data, and positioning data of plants and animals, and the environmental data includes at least meteorological data and physical environment data;

[0123] The processing and acquisition module 302 is configured to perform species identification on the biological feature vector based on the biological macro model to obtain information on plants and animals within the ecological protection zone, and to perform environmental analysis on the environmental feature vector based on the environmental macro model to obtain an environmental quality assessment within the ecological protection zone;

[0124] The analysis and evaluation module 303 is used to perform comprehensive reasoning and analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on plants and animals in the ecological protection zone, and the environmental quality assessment based on the ecosystem assessment model, and evaluate the ecosystem service function of the ecological protection zone. The comprehensive feature vector is determined based on the fusion of the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weights and environmental weights. The biological weights and the environmental weights are dynamically adjusted based on the contribution of the biological data and the environmental data to the ecosystem service function assessment.

[0125] Optionally, the generating module includes:

[0126] An acquisition submodule is used to pre-process the biological data and environmental data collected in the ecological protection area to obtain data to be analyzed;

[0127] A generating submodule is used to generate the biometric feature vector based on the image features, audio features and positioning features extracted from the data to be analyzed, and to generate the environmental feature vector based on the water quality features, soil features and meteorological features extracted from the data to be analyzed.

[0128] Optionally, the generating submodule includes:

[0129] a first extraction and alignment unit, configured to extract image features, audio features, and positioning features carrying timestamps from the data to be analyzed, and align the extracted biometric features of each modality according to the timestamps;

[0130] a first allocating unit, configured to allocate a first weight to each modal biometric feature based on the importance of each modal biometric feature;

[0131] A first splicing acquisition unit, configured to perform feature splicing on the aligned biometric features of each modality in combination with the matched first weights to obtain a first high-dimensional feature vector;

[0132] The first adjustment generation unit is configured to dynamically adjust the weights of the biometric features of each modality in the first high-dimensional feature vector to generate the biometric feature vector.

[0133] Optionally, the generating submodule includes:

[0134] a second extraction and alignment unit, configured to extract water quality features, soil features, and meteorological features with timestamps from the data to be analyzed, and align the extracted modal environmental features according to the timestamps;

[0135] a second allocating unit, configured to allocate a second weight to each modal environment feature based on the importance of each modal environment feature;

[0136] A second splicing acquisition unit is used to combine the aligned modal environment features with the matched second weights to perform feature splicing to obtain a second high-dimensional feature vector;

[0137] The second adjustment generation unit is used to dynamically adjust the weight of each modal environmental feature in the second high-dimensional feature vector to generate the environmental feature vector.

[0138] Optionally, the system further comprises:

[0139] A first determining module is configured to pre-train a first architecture model based on first biological sample data associated with animals and plants in the ecological protection zone to determine a first pre-trained model;

[0140] a first adjustment module, configured to adjust the first pre-trained model based on labeled second biological sample data associated with the animals and plants in the ecological protection area;

[0141] The second determination module is used to determine a biological macro model for species identification when the model performance of the first pre-trained model is verified by biological verification data.

[0142] Optionally, the system further comprises:

[0143] a third determining module, configured to pre-train the second architecture model based on first environmental sample data associated with the environment within the ecological protection zone to determine a second pre-trained model;

[0144] a second adjustment module, configured to adjust the second pre-trained model based on labeled second environmental sample data associated with the environment within the ecological protection zone, the second environmental sample data comprising at least labeled water quality data, soil data, and meteorological data;

[0145] The fourth determination module is used to determine the environmental macro model for environmental assessment when the model performance of the second pre-trained model is verified by environmental verification data.

[0146] Optionally, the system further comprises:

[0147] an allocation module, configured to allocate initial weights to the biological feature vector and the environmental feature vector, respectively, wherein the initial weights are allocated based on the estimated importance of the biological data and the environmental data in the ecosystem service function assessment;

[0148] a splicing generation module, configured to splice the biological feature vector and the environmental feature vector based on the assigned initial weights to generate a high-dimensional reference feature vector;

[0149] An adjustment generation module is used to dynamically adjust the initial weights in the high-dimensional benchmark feature vector based on the real-time contribution of the biological data and the environmental data to the ecosystem service function evaluation to generate the comprehensive feature vector.

[0150] Optionally, the analysis and evaluation module includes:

[0151] an adjustment and optimization submodule, configured to adjust the ecosystem assessment model based on ecosystem service function data to optimize the ecosystem assessment model, wherein the ecosystem service function data includes labeled biological data and environmental data within the ecological protection zone;

[0152] An analysis and generation submodule is used to perform comprehensive reasoning and analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the plant and animal information provided by the biological macro model, and the environmental quality assessment provided by the environmental macro model based on the optimized ecosystem assessment model to generate ecosystem species diversity assessment results and ecosystem environment assessment results;

[0153] The ecosystem environmental assessment results include at least the assessment of the ecosystem's water purification function, the assessment of the ecosystem's climate regulation function, and the assessment of the ecosystem's soil fertility maintenance function.

[0154] Optionally, the system further comprises:

[0155] A generation and display module is used to generate a visual evaluation report based on the evaluation results of the ecosystem service function of the ecological protection area for display;

[0156] The visual assessment report includes: a population distribution map showing the spatial distribution of species, a population density map showing the density distribution of species in different regions, a population dynamics map showing the change in species numbers over time, an environmental quality score map, and a statistical map of analysis results.

[0157] As for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0158] An embodiment of the present application also provides an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the various processes of the above-mentioned method embodiment for monitoring and evaluating terrestrial ecosystem service functions are implemented, and the same technical effects can be achieved. To avoid repetition, they will not be described here.

[0159] For example, Figure 4 FIG. 1 shows a schematic diagram of the physical structure of an electronic device. Figure 4 As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communications bus 440. The processor 410, communications interface 420, and memory 430 communicate with each other via the communications bus 440. The processor 410 may invoke logic instructions stored in the memory 430. The processor 410 is configured to execute the various steps of the method for monitoring and assessing terrestrial ecosystem service functions according to the embodiment of the present application, which will not be elaborated upon here.

[0160] In addition, the logic instructions in the aforementioned memory 430 can be implemented in the form of a software functional unit and, when sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application.

[0161] The present application also provides a computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements the various steps of the aforementioned method embodiment for monitoring and evaluating terrestrial ecosystem service functions, achieving the same technical effects. To avoid repetition, the details are omitted here. The computer-readable storage medium may be, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0162] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0163] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of this application.

[0164] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. A method for monitoring and evaluating terrestrial ecosystem service functions, characterized in that: include: Generate biological feature vectors and environmental feature vectors based on biological data and environmental data collected in the ecological protection area, wherein the biological data includes at least image data, audio data, and positioning data of plants and animals, and the environmental data includes at least meteorological data and physical environment data; Performing species identification on the biological feature vector based on the biological macro model to obtain information on plants and animals within the ecological protection zone, and performing environmental analysis on the environmental feature vector based on the environmental macro model to obtain an environmental quality assessment within the ecological protection zone; Based on the ecosystem assessment model, a comprehensive reasoning analysis is performed on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on flora and fauna within the ecological protection zone, and the environmental quality assessment to assess the ecosystem service function of the ecological protection zone; wherein the generation of the biological feature vector and the environmental feature vector based on the biological data and environmental data collected in the ecological protection zone includes: Preprocessing the biological data and environmental data collected in the ecological protection area to obtain data to be analyzed; generating the biometric feature vector based on image features, audio features, and positioning features extracted from the data to be analyzed, and generating the environmental feature vector based on water quality features, soil features, and meteorological features extracted from the data to be analyzed; The step of generating the biometric feature vector based on the image features, audio features, and positioning features extracted from the data to be analyzed includes: Extracting image features, audio features, and positioning features with timestamps from the data to be analyzed, and aligning the extracted biometric features of each modality according to the timestamps; assigning a first weight to each modality biometric feature based on the importance of each modality biometric feature; The aligned biometric features of each modality are combined with the matched first weights to perform feature concatenation to obtain a first high-dimensional feature vector; Dynamically adjusting the weights of the biometric features of each modality in the first high-dimensional feature vector to generate the biometric feature vector; The step of generating the environmental feature vector based on the water quality features, soil features, and meteorological features extracted from the data to be analyzed includes: Extracting water quality features, soil features, and meteorological features with timestamps from the data to be analyzed, and aligning the extracted modal environmental features according to the timestamps; assigning a second weight to each modal environment feature based on the importance of each modal environment feature; The aligned features of each modal environment are combined with the matched second weights to perform feature splicing to obtain a second high-dimensional feature vector; Dynamically adjusting the weight of each modal environmental feature in the second high-dimensional feature vector to generate the environmental feature vector; The comprehensive feature vector is determined based on the fusion of the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weight and environmental weight. The biological weight and the environmental weight are dynamically adjusted based on the contribution of the biological data and the environmental data to the ecosystem service function assessment.

2. The method for monitoring and evaluating terrestrial ecosystem service functions according to claim 1, characterized in that: The method further comprises: Pre-training a first architecture model based on first biological sample data associated with animals and plants in the ecological protection zone to determine a first pre-trained model; Adjusting the first pre-trained model based on labeled second biological sample data associated with animals and plants in the ecological protection area; When the model performance of the first pre-trained model passes the verification of the biological verification data, a biological macro model for species identification is determined.

3. The method for monitoring and evaluating terrestrial ecosystem service functions according to claim 1, characterized in that: The method further comprises: Pre-training a second architecture model based on first environmental sample data associated with the environment within the ecological protection zone to determine a second pre-trained model; adjusting the second pre-trained model based on labeled second environmental sample data associated with the environment within the ecological protection zone, the second environmental sample data including at least labeled water quality data, soil data, and meteorological data; When the model performance of the second pre-trained model is verified by the environmental verification data, an environmental macro model for performing environmental assessment is determined.

4. The method for monitoring and evaluating terrestrial ecosystem service functions according to claim 1, characterized in that: Also includes: Assigning initial weights to the biological feature vector and the environmental feature vector, respectively, wherein the initial weights are based on the estimated importance of the biological data and the environmental data in the ecosystem service function assessment; Based on the assigned initial weights, concatenating the biological feature vector and the environmental feature vector to generate a high-dimensional reference feature vector; The initial weights in the high-dimensional benchmark feature vector are dynamically adjusted based on the real-time contribution of the biological data and the environmental data to the evaluation of ecosystem service functions to generate the comprehensive feature vector.

5. The method for monitoring and evaluating terrestrial ecosystem service functions according to claim 1, characterized in that: The ecosystem assessment model is based on performing comprehensive reasoning and analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on flora and fauna within the ecological protection zone, and the environmental quality assessment to assess the ecosystem service function of the ecological protection zone, including: Adjusting the ecosystem assessment model based on ecosystem service function data to optimize the ecosystem assessment model, wherein the ecosystem service function data includes labeled biological data and environmental data within the ecological protection area; Based on the optimized ecosystem assessment model, a comprehensive reasoning analysis is performed on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the plant and animal information provided by the biological macro model, and the environmental quality assessment provided by the environmental macro model to generate an ecosystem species diversity assessment result and an ecosystem environment assessment result; The ecosystem environmental assessment results include at least the assessment of the ecosystem's water purification function, the assessment of the ecosystem's climate regulation function, and the assessment of the ecosystem's soil fertility maintenance function.

6. The method for monitoring and evaluating terrestrial ecosystem service functions according to claim 1 or 5, characterized in that: Also includes: Generate a visual assessment report based on the assessment results of the ecosystem service functions of the ecological protection area for presentation; The visual assessment report includes: a population distribution map showing the spatial distribution of species, a population density map showing the density distribution of species in different regions, a population dynamics map showing the change in species numbers over time, an environmental quality score map, and a statistical map of analysis results.

7. A system for monitoring and evaluating terrestrial ecosystem service functions, characterized in that: include: a generation module for generating a biological feature vector and an environmental feature vector based on biological data and environmental data collected in the ecological protection area, wherein the biological data includes at least image data, audio data, and positioning data of plants and animals, and the environmental data includes at least meteorological data and physical environment data; a processing and acquisition module, configured to perform species identification on the biological feature vector based on a biological macro model to obtain information on plants and animals within the ecological protection zone, and to perform environmental analysis on the environmental feature vector based on an environmental macro model to obtain an environmental quality assessment within the ecological protection zone; An analysis and evaluation module is configured to perform comprehensive reasoning and analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information on flora and fauna within the ecological protection zone, and the environmental quality assessment based on an ecosystem assessment model, to evaluate the ecosystem service function of the ecological protection zone; wherein the generation of the biological feature vector and the environmental feature vector based on the biological data and environmental data collected in the ecological protection zone includes: Preprocessing the biological data and environmental data collected in the ecological protection area to obtain data to be analyzed; generating the biometric feature vector based on image features, audio features, and positioning features extracted from the data to be analyzed, and generating the environmental feature vector based on water quality features, soil features, and meteorological features extracted from the data to be analyzed; The step of generating the biometric feature vector based on the image features, audio features, and positioning features extracted from the data to be analyzed includes: Extracting image features, audio features, and positioning features with timestamps from the data to be analyzed, and aligning the extracted biometric features of each modality according to the timestamps; assigning a first weight to each modality biometric feature based on the importance of each modality biometric feature; The aligned biometric features of each modality are combined with the matched first weights to perform feature concatenation to obtain a first high-dimensional feature vector; Dynamically adjusting the weights of the biometric features of each modality in the first high-dimensional feature vector to generate the biometric feature vector; The step of generating the environmental feature vector based on the water quality features, soil features, and meteorological features extracted from the data to be analyzed includes: Extracting water quality features, soil features, and meteorological features with timestamps from the data to be analyzed, and aligning the extracted modal environmental features according to the timestamps; assigning a second weight to each modal environment feature based on the importance of each modal environment feature; The aligned features of each modal environment are combined with the matched second weights to perform feature splicing to obtain a second high-dimensional feature vector; Dynamically adjusting the weight of each modal environmental feature in the second high-dimensional feature vector to generate the environmental feature vector; The comprehensive feature vector is determined based on the fusion of the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weight and environmental weight. The biological weight and the environmental weight are dynamically adjusted based on the contribution of the biological data and the environmental data to the ecosystem service function assessment.

Images