Cotton field carbon sink dynamic evaluation method and system based on multi-source data fusion

Abstract

The invention relates to a cotton field carbon sink dynamic evaluation method and system based on multi-source data fusion. The method comprises the following steps: firstly, comprehensively collecting multi-source data of a cotton field area environment, and processing the multi-source data by using a time window alignment technology to generate a multi-source heterogeneous data set; based on the data set, key features are extracted through a data mining means, and a key influence factor weight table is generated; and then, inputting the key influence factor weight table and the multi-source heterogeneous data set into the trained optimization evaluation model together to obtain cotton field carbon sink space-time distribution characteristics. And finally, carrying out correlation analysis on the spatial-temporal distribution characteristics and meteorological continuity data, and generating a curve capable of visually displaying the dynamic change of the cotton field carbon sink along with time. The method overcomes the limitation of a single data source, greatly improves the accuracy and comprehensiveness of cotton field carbon sink evaluation, can accurately reflect the real state of cotton field carbon sink under different time-space conditions, and provides a key basis for agricultural managers to formulate scientific and reasonable cotton field planting management strategies.

Description

Technical Field

[0001] The present invention belongs to the field of agricultural information technology, and in particular relates to a method and system for dynamically evaluating carbon sequestration in cotton fields by integrating multi-source data. Background Art

[0002] With the development of agricultural information technology, multi-source data fusion has emerged as a dynamic assessment technology for cotton field carbon sinks. As global climate change becomes increasingly severe, the role of agricultural ecosystems in the carbon cycle has garnered increasing attention. As a vital component of agricultural land, cotton fields' carbon sequestration capacity is crucial for mitigating greenhouse gas emissions and maintaining ecological balance. Accurately assessing the dynamics of cotton field carbon sinks not only helps optimize agricultural production management and improve resource utilization efficiency, but also provides a key basis for developing scientific agricultural strategies to address climate change. However, existing cotton field carbon sink assessment methods have numerous shortcomings. Previous assessments have often relied on a single data source. For example, relying solely on field soil sampling to analyze soil organic carbon content can provide localized soil carbon sink information but cannot fully reflect the dynamics of carbon sinks across the cotton field over different timescales. Alternatively, remote sensing imagery alone can be used to obtain vegetation information, but this fails to fully account for the continuous and complex impact of meteorological factors on the cotton field carbon cycle. This makes it difficult to achieve a comprehensive, accurate, and dynamic assessment of cotton field carbon sinks. Summary of the Invention

[0003] Based on this, it is necessary to address the above technical issues and provide a cotton field carbon sink dynamic assessment method and system that integrates multi-source data to improve the accuracy and comprehensiveness of cotton field carbon sink assessment.

[0004] First, this application provides a method for dynamic assessment of cotton field carbon sequestration using multi-source data fusion, including:

[0005] Acquire multi-source data of cotton field regional environment; multi-source data include remote sensing image data, meteorological continuity data and soil organic carbon sampling data.

[0006] The time window alignment technology is used to process multi-source data and fuse them to obtain multi-source heterogeneous data sets.

[0007] Based on multi-source heterogeneous data sets, features were extracted and principal component analysis was used to determine the contribution of each feature to carbon sink changes, and a weight table of key influencing factors was obtained.

[0008] The key influencing factor weight table and multi-source heterogeneous data sets were input into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks.

[0009] Correlation analysis is conducted on the spatiotemporal distribution characteristics and meteorological continuity data to generate a carbon sink dynamic change curve.

[0010] In one embodiment, multi-source data is processed using time window alignment technology to fuse multi-source heterogeneous data sets, including:

[0011] The time window alignment technology is used to extract the observation sequence of the corresponding time interval of the meteorological continuity data in the multi-source data to generate the associated meteorological data.

[0012] The spatial coordinate information of soil organic carbon sampling data in multi-source data is obtained, and the grid division parameters of the target area are generated according to the spatial resolution.

[0013] The spatial coordinates of the associated meteorological data are transformed according to the grid division parameters to obtain the spatially aligned meteorological raster data.

[0014] The corresponding grid cells in the meteorological raster data were extracted according to the spatial coordinate information to obtain the meteorological element values matching the soil organic carbon sampling points.

[0015] The spectral reflectance data of remote sensing image data and the numerical values of meteorological elements are normalized to generate a multidimensional feature matrix.

[0016] The spatial index matching of the multidimensional feature matrix was performed according to the spatial coordinates of the soil organic carbon sampling points, and a multi-source heterogeneous data set was obtained using the random forest algorithm; the multi-source heterogeneous data set included the mapping relationship between remote sensing spectral features, meteorological element features and soil organic carbon content.

[0017] In one embodiment, features are extracted based on multi-source heterogeneous data sets and the principal component analysis method is used to determine the contribution of each feature to carbon sink changes, thereby obtaining a weight table of key influencing factors, including:

[0018] The multi-source heterogeneous dataset is preprocessed to obtain a standardized dataset; the multi-source heterogeneous dataset includes photosynthesis efficiency characteristics and agronomic practice parameters.

[0019] The standardized data set was fused, and the dimensions of photosynthesis efficiency features were reduced using principal component analysis to generate a feature contribution matrix.

[0020] The principal component loading values were extracted from the characteristic contribution matrix and combined with the standardized weights of the agronomic practice parameters to construct a weight model of the factors affecting carbon sink changes.

[0021] The weight model is combined with real-time environmental monitoring data, and dynamic thresholds are used to screen key influencing factors to obtain an updated key influencing factor weight table.

[0022] In one embodiment, the standardized dataset is subjected to feature fusion, and the photosynthesis efficiency feature is reduced in dimension using principal component analysis to generate a feature contribution matrix, including:

[0023] The light intensity response parameters and chlorophyll dynamic indicators in the standardized data set were fused to generate a multidimensional feature vector.

[0024] The multidimensional feature vector is input into the principal component analysis model to obtain the principal component vector and variance ratio parameter.

[0025] The characteristic contribution weight is calculated according to the variance ratio parameter; the characteristic contribution weight includes the light intensity contribution factor and the chlorophyll contribution factor.

[0026] A feature contribution matrix is constructed based on the principal component vector and the feature contribution weight; the feature contribution matrix includes principal component load parameters and weight distribution parameters.

[0027] The principal component load parameters and weight distribution parameters are normalized to generate a standardized feature contribution matrix.

[0028] In one embodiment, the key influencing factor weight table and the multi-source heterogeneous data set are input into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks, including:

[0029] The data fusion algorithm is used to align the data in multi-source heterogeneous data sets in time and space to obtain a fused multidimensional feature data set.

[0030] After preprocessing the multidimensional feature data set, the feature extraction algorithm is used to perform weighted aggregation in combination with the key influencing factor weight table to generate a spatiotemporal feature set.

[0031] The spatiotemporal feature set is input into the trained optimization evaluation model for dynamic convolution calculation to obtain the dynamic change trend of carbon sinks.

[0032] The spatial distribution heat map of carbon sinks in cotton fields was generated according to the dynamic change trend, and the spatiotemporal distribution characteristics were obtained by feature extraction from the spatial distribution heat map of carbon sinks.

[0033] In one embodiment, the spatiotemporal feature set is calculated using the following formula:

[0034]

[0035] Among them, F(x,t) represents the spatiotemporal feature set, ω i Represents the weight coefficient of the i-th feature, f i (x, t) represents the value of the i-th feature in the spatiotemporal dimension, α i represents the feature importance coefficient, and n represents the total number of features.

[0036] In one embodiment, correlation analysis is performed on the spatiotemporal distribution characteristics and meteorological continuity data to generate a carbon sink dynamic change curve, including:

[0037] The dynamic response matrix of light energy utilization was obtained by performing correlation analysis on meteorological continuity data and characteristic parameters of photosynthesis efficiency.

[0038] The irrigation frequency and fertilization amount in the spatiotemporal distribution characteristics are extracted and combined with the dynamic response matrix to calculate the gradient distribution of the dynamic adjustment coefficient.

[0039] The gradient distribution is input into the carbon sink accumulation model and the weighted interval of the absorption rate is obtained according to the vegetation type.

[0040] The dynamic adjustment coefficient is spatially interpolated based on the weight interval to generate a carbon sink dynamic change curve covering the target area.

[0041] Secondly, this application also provides a cotton field carbon sink dynamic assessment system based on multi-source data fusion, which includes:

[0042] The data fusion module is used to obtain multi-source data on the cotton field area environment; the multi-source data includes remote sensing image data, meteorological continuity data and soil organic carbon sampling data; it is also used to process the multi-source data using time window alignment technology to fuse multi-source heterogeneous data sets.

[0043] The feature analysis module is used to extract features based on multi-source heterogeneous data sets and use principal component analysis to determine the contribution of each feature to carbon sink changes, thereby obtaining a weight table of key influencing factors. It is also used to input the weight table of key influencing factors and multi-source heterogeneous data sets into a trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks.

[0044] The result generation module is used to perform correlation analysis on spatiotemporal distribution characteristics and meteorological continuity data to generate a carbon sink dynamic change curve.

[0045] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above method when executing the computer program.

[0046] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the above method when executed by a processor.

[0047] The multi-source data fusion method, system, computer equipment, and storage medium for dynamic assessment of cotton field carbon sinks first comprehensively collects multi-source data on the cotton field regional environment. Then, time window alignment technology is used to refine the multi-source data and fuse it to generate a multi-source heterogeneous dataset. Based on this multi-source heterogeneous dataset, data mining is used to extract key features. Principal component analysis is then used to quantitatively determine the contribution of each feature to carbon sink changes, generating a weighted table of key influencing factors. Next, the weighted table of key influencing factors and the multi-source heterogeneous dataset are input into a pre-trained optimization assessment model to determine the spatiotemporal distribution characteristics of cotton field carbon sinks. Finally, correlation analysis is performed on the spatiotemporal distribution characteristics with continuous meteorological data to generate a curve that visually demonstrates the dynamic changes of cotton field carbon sinks over time. By fusing multi-source data, the limitations of a single data source are overcome, significantly improving the accuracy and comprehensiveness of cotton field carbon sink assessments and accurately reflecting the true state of cotton field carbon sinks under different spatiotemporal conditions. It provides a key basis for agricultural managers to formulate scientific and reasonable cotton field planting management strategies, helps to optimize agronomic measures, improve the carbon sequestration capacity of cotton fields, achieve the coordinated development of agricultural production and ecological environment protection, and play an important role in promoting the sustainable development of agriculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 A flowchart of a method for dynamic assessment of cotton field carbon sequestration using multi-source data fusion provided by an embodiment of the present invention;

[0050] Figure 2 This is a structural block diagram of the cotton field carbon sink dynamic assessment system based on multi-source data fusion provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0052] In one embodiment, Figure 1 As shown, this application provides a method for dynamic assessment of cotton field carbon sequestration based on multi-source data fusion, which can include the following steps:

[0053] Step S101, obtaining multi-source data of the cotton field regional environment; the multi-source data includes remote sensing image data, meteorological continuity data and soil organic carbon sampling data.

[0054] Specifically, during the cotton field regional environmental data collection phase, advanced satellite remote sensing technology was used to obtain remote sensing image data of different bands and high resolution. These data can clearly present key information such as the coverage, growth status, and vegetation index of cotton field vegetation. At the same time, by building a dense meteorological monitoring network and using high-precision meteorological sensors, continuous and stable meteorological data covering factors such as temperature, humidity, light intensity, precipitation frequency and magnitude are collected to accurately record the dynamic changes in the cotton field environment. In addition, in the cotton field, in strict accordance with scientific sampling specifications, professional soil sampling tools are used to conduct multi-point sampling in different areas and at different depths to obtain widely representative soil organic carbon sampling data to ensure that the spatial distribution characteristics of soil organic carbon in the cotton field can be accurately reflected.

[0055] Step S102: Process the multi-source data using time window alignment technology to fuse them into a multi-source heterogeneous data set.

[0056] Using time window alignment technology to address the differences in temporal scales between remote sensing imagery, meteorological continuity data, and soil organic carbon sampling data, a complex time series matching algorithm is employed to precisely align the various data types using a pre-set time window. Furthermore, advanced data fusion algorithms, such as a Bayesian-based fusion model, comprehensively consider the reliability, relevance, and uncertainty of data from different sources, enabling in-depth integration of multi-source data from diverse fields and formats. This fusion process fully exploits the potential connections between these data sources, generating a multi-source, heterogeneous dataset rich in information.

[0057] Step S103 , extracting features based on the multi-source heterogeneous data set and using principal component analysis to determine the contribution of each feature to the change in carbon sinks, and obtaining a weight table of key influencing factors.

[0058] Based on the generated multi-source heterogeneous data sets, professional data mining algorithms such as decision tree algorithms and association rule mining algorithms are used to deeply extract key features closely related to cotton field carbon sequestration from massive data, including but not limited to vegetation index, light and effective radiation absorption ratio, etc., which reflect photosynthesis efficiency, as well as weight parameters such as irrigation amount, fertilization type and frequency, which reflect the implementation of agronomic measures. Subsequently, the principal component analysis (PCA) method is introduced. This method converts many original features with complex correlations into a few independent comprehensive principal components through linear transformation. During the conversion process, by calculating the load of each original feature in the principal component, the contribution of each feature to the principal component is accurately quantified, and then the degree of influence of each feature on the change of cotton field carbon sequestration is clearly defined. Based on this analysis result, a key influencing factor weight table is constructed, which intuitively shows the relative importance of different features in influencing the change of cotton field carbon sequestration.

[0059] Step S104: input the key influencing factor weight table and the multi-source heterogeneous data set into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks.

[0060] The generated key influencing factor weight table is input together with the multi-source heterogeneous data set into an optimization evaluation model that has been pre-trained with a large amount of data and has been parameter optimized. This model is usually built based on machine learning algorithms, such as random forest algorithms and neural network algorithms, which have strong nonlinear fitting and data processing capabilities. During the model operation, the various features in the multi-source heterogeneous data set are weighted according to the input key influencing factor weights. Through complex model calculations, the status of cotton field carbon sinks at different times and spatial locations is simulated and predicted. After precise calculations of the model, the final output can accurately reflect the distribution characteristics of cotton field carbon sinks in the temporal and spatial dimensions. These results are presented in the form of intuitive data or visual graphics.

[0061] Step S105 , performing correlation analysis on the spatiotemporal distribution characteristics and the meteorological continuity data to generate a carbon sink dynamic change curve.

[0062] Specifically, an in-depth correlation analysis was conducted between the spatiotemporal distribution characteristics of cotton field carbon sinks, as output by the model, and previously collected continuous meteorological data. Using statistical methods such as the Pearson correlation coefficient and grey correlation analysis, the potential links between the spatiotemporal distribution of cotton field carbon sinks and meteorological factors were explored, and the degree to which meteorological conditions (such as temperature, sunlight, and precipitation) affect changes in cotton field carbon sinks was quantitatively analyzed. Based on the correlation analysis results, characteristics closely related to photosynthesis efficiency and dynamic adjustment coefficients for agronomic practice weight parameters were determined. These coefficients reflect the mechanism by which changes in meteorological factors affect key factors affecting cotton field carbon sinks. Finally, using time series analysis methods such as the ARIMA model and seasonal decomposition method, the dynamic adjustment coefficients were incorporated into the carbon sink change model to generate a curve that visually demonstrates the continuous dynamic changes in cotton field carbon sinks over time. This curve provides agricultural managers, researchers, and others with intuitive and accurate information on cotton field carbon sink dynamics, helping to make scientific and reasonable agricultural production decisions and promote the sustainable development of cotton field ecosystems.

[0063] The multi-source data fusion method for assessing the dynamic carbon sink in cotton fields first comprehensively collects multi-source data on the cotton field environment. Then, it uses time window alignment technology to refine the multi-source data and fuse it to generate a multi-source heterogeneous dataset. Based on this multi-source heterogeneous dataset, data mining is used to extract key features. Principal component analysis is then used to quantitatively determine the contribution of each feature to carbon sink changes, generating a weighted table of key influencing factors. Next, the weighted table of key influencing factors, along with the multi-source heterogeneous dataset, is input into a pre-trained optimization assessment model to determine the spatiotemporal distribution of the cotton field carbon sink. Finally, correlation analysis is performed on the spatiotemporal distribution features with continuous meteorological data to generate a curve that visually demonstrates the dynamic changes of the cotton field carbon sink over time. By fusing multi-source data, the limitations of a single data source are overcome, significantly improving the accuracy and comprehensiveness of the cotton field carbon sink assessment and accurately reflecting the true state of the cotton field carbon sink under different spatiotemporal conditions. It provides a key basis for agricultural managers to formulate scientific and reasonable cotton field planting management strategies, helps to optimize agronomic measures, improve the carbon sequestration capacity of cotton fields, achieve the coordinated development of agricultural production and ecological environment protection, and play an important role in promoting the sustainable development of agriculture.

[0064] In one embodiment, processing multi-source data using time window alignment technology to fuse a multi-source heterogeneous data set may include the following steps:

[0065] Step S201 , using time window alignment technology to extract observation sequences corresponding to time intervals of meteorological continuity data in multi-source data, and generate associated meteorological data.

[0066] Step S202: obtaining spatial coordinate information of soil organic carbon sampling data from multi-source data, and generating grid division parameters for the target area according to the spatial resolution.

[0067] Step S203 , performing spatial coordinate conversion on the associated meteorological data according to the grid division parameters to obtain spatially aligned meteorological grid data.

[0068] Step S204: extracting corresponding grid cells in the meteorological raster data according to the spatial coordinate information to obtain meteorological element values that match the soil organic carbon sampling points.

[0069] Step S205 , normalizing the spectral reflectance data of the remote sensing image data and the meteorological element values to generate a multi-dimensional feature matrix.

[0070] Step S206, performing spatial index matching on the multidimensional feature matrix according to the spatial coordinates of the soil organic carbon sampling points, and using a random forest algorithm to obtain a multi-source heterogeneous data set; the multi-source heterogeneous data set includes a mapping relationship between remote sensing spectral features, meteorological element features and soil organic carbon content.

[0071] First, using time window alignment technology, we precisely screened and extracted observation sequences corresponding to the desired timeframe from the multi-source meteorological data. Through a rigorous data screening process, we generated meteorological data closely aligned with the study period. Furthermore, we obtained precise spatial coordinate information for the soil organic carbon (SOC) sampling data from the multi-source data. Using specialized spatial analysis algorithms, we scientifically calculated and generated detailed gridding parameters for the target area, consistent with the established spatial resolution. Based on these parameters, we converted the generated associated meteorological data from the original meteorological data format to spatially aligned meteorological grid data using a complex and precise spatial coordinate conversion model, ensuring that the meteorological data and the SOC sampling data were spatially aligned. Subsequently, based on the spatial coordinate information of the SOC sampling data, we precisely located and extracted the corresponding grid cells in the meteorological grid data. Through meticulous data comparison and screening, we obtained meteorological element values that closely matched each SOC sampling point, establishing a deep temporal and spatial correlation between the meteorological and soil sampling data. Next, the spectral reflectance data from the remote sensing imagery and the acquired meteorological element values were normalized using a standardized method to eliminate dimensional differences between different data types, thereby generating a multidimensional feature matrix containing multi-source information. Finally, a spatial index matching operation was performed on the multidimensional feature matrix based on the spatial coordinates of the soil organic carbon sampling points. Leveraging the powerful classification and regression capabilities of the random forest algorithm, the various data types were organically integrated to successfully construct a multi-source heterogeneous dataset. This dataset clearly demonstrates the mapping relationship between remote sensing spectral characteristics, meteorological element characteristics, and soil organic carbon content.

[0072] From the perspective of data fusion, this embodiment achieves efficient fusion of data from different sources, formats, and dimensions through multi-link, refined spatiotemporal processing and data integration operations, breaking the data island phenomenon and greatly improving the integrity and availability of the data. In terms of analysis accuracy, the constructed multi-source heterogeneous data set covers a wealth of cotton field environmental information, and each data has an accurate spatiotemporal correspondence, which provides a solid foundation for in-depth exploration of the complex relationship between cotton field carbon sinks and multiple factors, and can significantly improve the accuracy and reliability of subsequent analysis results. From the perspective of application value, this multi-source heterogeneous data set can provide comprehensive and accurate data support for the construction of cotton field carbon sink assessment models, the optimization and formulation of agricultural production management strategies, and ecological environment monitoring and research, effectively promoting scientific development and practical application in the field of agricultural ecology, and playing a key role in helping the sustainable development of agriculture.

[0073] In one embodiment, extracting features based on multi-source heterogeneous data sets and using principal component analysis to determine the contribution of each feature to carbon sink changes to obtain a weight table of key influencing factors may include the following steps:

[0074] Step S301 , preprocessing the multi-source heterogeneous dataset to obtain a standardized dataset; the multi-source heterogeneous dataset includes photosynthesis efficiency characteristics and agronomic practice parameters.

[0075] Step S302 : performing feature fusion on the standardized data set, reducing the dimension of photosynthesis efficiency features using principal component analysis, and generating a feature contribution matrix.

[0076] Step S303 : extracting the principal component load values according to the characteristic contribution matrix and combining them with the standardized weights of the agronomic practice parameters to construct a weight model of the factors affecting carbon sink changes.

[0077] Step S304 : combining the weight model with the real-time environmental monitoring data, using dynamic thresholds to determine and screen key influencing factors, and obtaining an updated key influencing factor weight table.

[0078] Specifically, a comprehensive preprocessing was first performed on a multi-source, heterogeneous dataset containing photosynthetic efficiency characteristics and agronomic practice parameters. Using a normalization algorithm, the data were normalized based on their statistical characteristics, eliminating dimensional differences between different data dimensions and transforming the raw data into a standardized dataset with a mean of 0 and a variance of 1. Feature fusion was then performed on the standardized dataset to organically combine the various features to form a more representative comprehensive feature set. Furthermore, principal component analysis (PCA) was applied to focus on photosynthetic efficiency characteristics. Through linear transformation, the numerous complex and correlated raw photosynthetic efficiency characteristics were converted into a small number of independent principal components, achieving data dimensionality reduction and effectively reducing data complexity. A feature contribution matrix was also generated, clearly demonstrating the contribution of each original feature to the principal component construction. Next, the principal component loadings were accurately extracted from the feature contribution matrix. Combined with the standardized weights of the agronomic practice parameters, a mathematical modeling approach was used to construct a weighting model that quantifies the factors influencing carbon sequestration changes. This model comprehensively considers the effects of photosynthetic efficiency and agronomic practices on carbon sequestration changes. Finally, the constructed weight model is closely integrated with real-time environmental monitoring data. Using the dynamic threshold judgment algorithm, factors that have a key impact on carbon sink changes are flexibly screened out based on the ever-changing characteristics of environmental data, thereby obtaining an updated key influencing factor weight table, which can reflect the core driving factors of carbon sink changes in the current environment in real time.

[0079] Through standardized preprocessing and feature fusion, the standardization and comprehensiveness of the data have been improved. The application of principal component analysis not only simplifies the data structure, but also retains key information, greatly improving the efficiency of data analysis. The constructed carbon sink change influencing factor weight model and the updated key influencing factor weight table can accurately quantify the impact of various factors on carbon sink changes, providing a powerful tool for in-depth research on the carbon sink mechanism of cotton fields. In practical applications, it can help agricultural researchers and managers accurately grasp the key factors affecting cotton field carbon sinks based on real-time environmental data, and then make scientific and reasonable agricultural production decisions, optimize agronomic measures, enhance the carbon sink capacity of cotton fields, and promote the sustainable development of agricultural ecosystems.

[0080] In one embodiment, the standardized dataset is subjected to feature fusion, and the photosynthesis efficiency feature is reduced in dimension using principal component analysis to generate a feature contribution matrix, which may include the following steps:

[0081] Step S401 : performing feature fusion on the light intensity response parameters and chlorophyll dynamic indicators in the standardized data set to generate a multi-dimensional feature vector.

[0082] Step S402: Input the multidimensional feature vector into the principal component analysis model to obtain the principal component vector and variance ratio parameter.

[0083] Step S403 , calculating the feature contribution weight according to the variance ratio parameter; the feature contribution weight includes the light intensity contribution factor and the chlorophyll contribution factor.

[0084] Step S404: constructing a feature contribution matrix based on the principal component vector and the feature contribution weight; the feature contribution matrix includes principal component load parameters and weight distribution parameters.

[0085] Step S405 , normalizing the principal component load parameters and the weight distribution parameters to generate a standardized feature contribution matrix.

[0086] For the standardized dataset, further processing focused on light intensity response parameters and chlorophyll dynamics. First, a specialized data fusion algorithm was used to organically integrate the light intensity response parameters and chlorophyll dynamics. Through complex mathematical operations and logical associations, a multidimensional feature vector rich in information was generated. This vector comprehensively integrates key data dimensions related to light intensity and chlorophyll, providing a comprehensive data foundation for subsequent in-depth analysis. The resulting multidimensional feature vector was then input into a carefully constructed principal component analysis (PCA) model. During the model run, linear transformations were used to deconstruct and reconstruct the complex data within the multidimensional feature vector, outputting a principal component vector and a variance ratio parameter reflecting the degree of data dispersion. Based on the variance ratio parameter, a specific mathematical formula was used to calculate feature contribution weights, which clearly define the relative importance of the light intensity and chlorophyll contributions within the overall data structure. Next, a feature contribution matrix was constructed based on the PC vectors and the calculated feature contribution weights. This matrix includes the principal component loading parameters, which reflect the correlation between the original features within the principal components, as well as the weight distribution parameters, which visually display the weighting of different features. Finally, to ensure the consistency and comparability of the data, the principal component load parameters and weight distribution parameters are normalized. Through standardized mathematical transformation, the data are unified into a specific numerical range to generate a standardized feature contribution matrix.

[0087] This embodiment generates a multi-dimensional feature vector through feature fusion, fully explores the potential connection between light intensity and chlorophyll data, enriches the data content, and provides a more comprehensive data perspective for in-depth exploration of the carbon sequestration-related mechanism in cotton fields. The application of the principal component analysis model effectively realizes data dimensionality reduction and key information extraction, greatly improves the efficiency of data analysis, and can quickly extract core features from complex data. The generated standardized feature contribution matrix provides an accurate and standardized data basis for the subsequent construction of a weight model of factors affecting carbon sequestration changes, etc., and helps to more accurately quantify the impact of various factors on cotton field carbon sequestration changes. In scientific research in the field of agricultural ecology, it provides a powerful data processing and analysis method for revealing the inherent mechanism of cotton field carbon sequestration. In actual agricultural production management, it can help managers to scientifically formulate strategies to enhance the carbon sequestration capacity of cotton fields based on accurate data conclusions and promote the sustainable development of agricultural ecosystems.

[0088] In one embodiment, inputting the key influencing factor weight table and the multi-source heterogeneous data set into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks may include the following steps:

[0089] Step S501: Use a data fusion algorithm to perform spatiotemporal alignment on the data in the multi-source heterogeneous data set to obtain a fused multidimensional feature data set.

[0090] Step S502 : After pre-processing the multidimensional feature data set, the data set is combined with the key influencing factor weight table and weightedly aggregated using a feature extraction algorithm to generate a spatiotemporal feature set.

[0091] Step S503: Input the spatiotemporal feature set into the trained optimization evaluation model to perform dynamic convolution calculation to obtain the dynamic change trend of carbon sinks.

[0092] Step S504 : generating a heat map of the spatial distribution of carbon sinks in cotton fields according to the dynamic change trend, and performing feature extraction on the heat map of the spatial distribution of carbon sinks to obtain spatiotemporal distribution features.

[0093] Specifically, advanced data fusion algorithms were first used to perform spatiotemporal alignment on heterogeneous multi-source datasets with diverse sources and spatiotemporal characteristics. Through time alignment and spatial matching, remote sensing imagery, meteorological continuity data, and soil organic carbon sampling data were seamlessly integrated in the spatiotemporal dimensions, resulting in a fused multidimensional feature dataset. This dataset leverages the strengths of multiple data sources to comprehensively reflect the comprehensive characteristics of cotton fields under varying spatiotemporal conditions. Subsequently, rigorous preprocessing was performed on the fused multidimensional feature dataset to ensure data quality, including noise removal and missing value filling. Based on this, an efficient feature extraction algorithm was employed, combining a constructed weight table of key influencing factors, to perform weighted aggregation of the data according to the weights of each factor. This approach highlights the impact of key factors on cotton field carbon sequestration and generates a spatiotemporal feature set containing rich spatiotemporal information. This spatiotemporal feature set was then input into an optimization evaluation model pre-trained with extensive data and optimized for parameters. The model, using advanced computational methods such as dynamic convolution, deeply mines and analyzes the data in the spatiotemporal feature set, simulating the temporal evolution of cotton field carbon sequestration and accurately capturing its dynamic trends. Finally, based on the dynamic trends of carbon sinks, a professional visualization tool was used to generate a heat map of the spatial distribution of carbon sinks in cotton fields. This heat map uses intuitive color distribution to display the relative strength of carbon sinks in different regions. By extracting features from the heat map, the spatiotemporal distribution characteristics of carbon sinks in cotton fields were accurately determined.

[0094] From a data processing perspective, operations such as spatiotemporal alignment and weighted aggregation effectively integrate multi-source data, improving data integrity and usability and laying a solid foundation for subsequent analysis. In terms of model application, the dynamic convolution calculations used in the optimized assessment model accurately capture carbon sink trends, significantly improving assessment accuracy. The generated spatiotemporal distribution characteristics and related visualizations provide agricultural researchers and managers with clear and intuitive information on cotton field carbon sinks. In the scientific research field, this will facilitate in-depth study of cotton field carbon sink mechanisms and promote the development of agricultural ecological science.

[0095] In one embodiment, the spatiotemporal feature set can be calculated using the following formula:

[0096]

[0097] Among them, F(x,y) represents the spatiotemporal feature set, ω i Represents the weight coefficient of the i-th feature, f i (x, t) represents the value of the i-th feature in the spatiotemporal dimension, α i represents the feature importance coefficient, and n represents the total number of features.

[0098] This example uses a specific formula to calculate a set of spatiotemporal features, scientifically and accurately quantifying the impact of each feature on cotton field carbon sequestration, significantly improving data integrity and usability. This approach facilitates in-depth research on cotton field carbon sequestration mechanisms and promotes the scientific development of agricultural ecology. It also assists managers in developing scientifically sound agronomic practices, such as precision irrigation and fertilization, to enhance cotton field carbon sequestration capacity, achieving the coordinated development of agricultural production and ecological protection, and is of great significance for promoting sustainable agricultural development.

[0099] In one embodiment, performing correlation analysis on the spatiotemporal distribution characteristics and meteorological continuity data to generate a carbon sink dynamic change curve may include the following steps:

[0100] Step S601 : performing correlation analysis on the meteorological continuity data and the characteristic parameters of photosynthesis efficiency to obtain a dynamic response matrix of light energy utilization rate.

[0101] Step S602: extracting the irrigation frequency and fertilization amount from the spatiotemporal distribution characteristics and combining them with the dynamic response matrix to calculate the gradient distribution of the dynamic adjustment coefficients.

[0102] Step S603: Input the gradient distribution into the carbon sink accumulation model and divide it into different vegetation types to obtain a weighted interval of the absorption rate.

[0103] Step S604 : performing spatial interpolation on the dynamic adjustment coefficient based on the weight interval to generate a carbon sink dynamic change curve covering the target area.

[0104] Applying specialized statistical analysis methods to continuous meteorological data and characteristic parameters of photosynthetic efficiency, the authors examined the intrinsic relationships between meteorological factors such as light intensity, temperature, and humidity, and characteristic parameters of photosynthetic efficiency, such as vegetation photosynthetic rate and chlorophyll content, to accurately derive a dynamic response matrix for light energy utilization efficiency. This matrix quantitatively illustrates the changing patterns of light energy utilization efficiency under different meteorological conditions. Next, from previously acquired spatial and temporal distribution patterns of cotton field carbon sequestration, they precisely extracted irrigation frequency and fertilization rate, two agronomic parameters with significant impacts on the carbon cycle in cotton fields. Combining this with the dynamic response matrix for light energy utilization efficiency, a complex mathematical model was used to deeply analyze the impact of the interaction between meteorological factors and agronomic practices on cotton field carbon sequestration. The resulting gradient distribution of the dynamic adjustment coefficient was then calculated. This gradient distribution intuitively reflects the changing trends of the dynamic adjustment coefficient across different regions and conditions. This gradient distribution was then used as key input into a pre-built carbon sink accumulation model. The model meticulously divides the cotton fields into different vegetation types, such as cotton varieties. By simulating the carbon sink accumulation process, it scientifically determines the weighted ranges for the absorption rates corresponding to different vegetation types. Finally, based on the obtained weighted ranges, a spatial interpolation algorithm is used to spatially expand and estimate the dynamic adjustment coefficients based on known discrete data points, thereby generating a dynamic carbon sink curve that comprehensively covers the target cotton field area. This curve, with time as the horizontal axis, visually illustrates the continuous change of carbon sinks in the cotton field over time at different spatial locations.

[0105] By systematically analyzing the relationship between meteorological factors and photosynthetic efficiency, and integrating them with agronomic parameters for comprehensive calculations, this study deeply reveals the underlying mechanisms of carbon sequestration in cotton fields. This provides rich data support and theoretical foundations for academic research in the field of agricultural ecology, and promotes the understanding of the complex processes of the farmland carbon cycle. In practical agricultural production management, the generated carbon sequestration dynamics curve can provide managers with intuitive and accurate information on cotton field carbon sequestration. Based on the carbon sequestration trends reflected in the curve, managers can rationally adjust irrigation and fertilization strategies and optimize agronomic practices under different temporal and spatial conditions to enhance the carbon sequestration capacity of cotton fields, achieving the dual goals of energy conservation and emission reduction in agricultural production while protecting the ecological environment. This has critical practical guidance significance for promoting the effective implementation of agricultural sustainable development strategies.

[0106] In one embodiment, Figure 2 As shown, this application also provides a cotton field carbon sink dynamic assessment system based on multi-source data fusion, which may include:

[0107] The data fusion module 701 is used to obtain multi-source data of the cotton field regional environment; the multi-source data includes remote sensing image data, meteorological continuity data and soil organic carbon sampling data; it is also used to process the multi-source data using time window alignment technology to fuse the multi-source data to obtain a multi-source heterogeneous data set.

[0108] The feature analysis module 702 is used to extract features based on multi-source heterogeneous data sets and use the principal component analysis method to determine the contribution of each feature to carbon sink changes, and obtain a key influencing factor weight table; it is also used to input the key influencing factor weight table and the multi-source heterogeneous data sets into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks.

[0109] The result generation module 703 is used to perform correlation analysis on the spatiotemporal distribution characteristics and the meteorological continuity data to generate a carbon sink dynamic change curve.

[0110] The multi-source data fusion cotton field carbon sink dynamic assessment system utilizes advanced remote sensing technology, a meteorological monitoring network, and specialized soil sampling methods to comprehensively acquire multi-source data on the cotton field environment. This includes remote sensing imagery reflecting the status of cotton field vegetation, continuous meteorological data recording environmental dynamics, and soil organic carbon sampling data representing soil carbon content. Subsequently, a time window alignment technique is used to time-align the multi-source data using a precise time matching algorithm to account for differences in the temporal scales of different data sources. Furthermore, a sophisticated data fusion algorithm is employed to deeply fuse the various data types, comprehensively considering factors such as data reliability and relevance, to generate a multi-source heterogeneous dataset. The feature analysis module, using specialized data mining algorithms from the multi-source heterogeneous dataset generated by the data fusion module, extracts features closely related to cotton field carbon sinks, such as photosynthetic efficiency and agronomic parameters. Principal component analysis is used to reduce the dimensionality of these numerous complex and correlated features, quantitatively determining the contribution of each feature to carbon sink changes and constructing a weighted table of key influencing factors. This weight table, along with a multi-source heterogeneous dataset, is fed into an optimization assessment model pre-trained with extensive data and optimized for parameters. Through complex computations, the model outputs results that accurately reflect the spatiotemporal distribution characteristics of cotton field carbon sinks. The result generation module focuses on conducting in-depth correlation analysis between the spatiotemporal distribution characteristics obtained by the feature analysis module and continuous meteorological data. Statistical methods are used to explore the potential connections between the two and generate a dynamic carbon sink curve. This curve visually depicts the dynamic changes in cotton field carbon sinks over time and at different spatial locations. By integrating multi-source data, this system overcomes the limitations of a single data source, significantly improving the accuracy and comprehensiveness of cotton field carbon sink assessments and accurately reflecting the true state of cotton field carbon sinks under different spatiotemporal conditions. This provides a key basis for agricultural managers to formulate scientific and rational cotton field cultivation and management strategies, helping to optimize agronomic practices, enhance cotton field carbon sequestration capacity, and achieve the coordinated development of agricultural production and ecological and environmental protection, playing an important role in promoting sustainable agricultural development.

[0111] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0112] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method and system for dynamic assessment of cotton field carbon sequestration using multi-source data fusion as described above are implemented.

[0113] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0114] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0115] The above-described embodiments merely represent several implementation methods of the embodiments of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the concept of the embodiments of the present application, and these modifications and improvements fall within the scope of protection of the embodiments of the present application.

Claims

1. A method for dynamic assessment of cotton field carbon sinks based on multi-source data fusion, characterized by: The method comprises: Acquire multi-source data of cotton field regional environment; the multi-source data includes remote sensing image data, meteorological continuity data and soil organic carbon sampling data; The multi-source data are processed using a time window alignment technique to fuse and obtain a multi-source heterogeneous data set; Extracting features based on the multi-source heterogeneous data set and using principal component analysis to determine the contribution of each feature to carbon sink changes, and obtaining a weight table of key influencing factors; Inputting the key influencing factor weight table and the multi-source heterogeneous data set into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks; A correlation analysis is performed on the spatiotemporal distribution characteristics and meteorological continuity data to generate a carbon sink dynamic change curve.

2. The method according to claim 1, characterized in that The multi-source data is processed using a time window alignment technique to fuse the multi-source heterogeneous data set, including: Extracting the observation sequence of the time interval corresponding to the meteorological continuity data in the multi-source data by using the time window alignment technology to generate associated meteorological data; Obtaining spatial coordinate information of the soil organic carbon sampling data in the multi-source data, and generating grid division parameters of the target area according to the spatial resolution; Performing spatial coordinate conversion on the associated meteorological data according to the grid division parameters to obtain spatially aligned meteorological grid data; Extracting corresponding grid cells in the meteorological raster data according to the spatial coordinate information to obtain meteorological element values that match the soil organic carbon sampling points; Normalizing the spectral reflectance data of the remote sensing image data and the meteorological element values to generate a multidimensional feature matrix; The multidimensional feature matrix is spatially indexed and matched according to the spatial coordinates of the soil organic carbon sampling points, and a multi-source heterogeneous data set is obtained using a random forest algorithm; the multi-source heterogeneous data set includes a mapping relationship between remote sensing spectral features, meteorological element features and soil organic carbon content.

3. The method according to claim 1, characterized in that The feature extraction based on the multi-source heterogeneous data set and the principal component analysis method are used to determine the contribution of each feature to the carbon sink change, and a key influencing factor weight table is obtained, including: Preprocessing the multi-source heterogeneous data set to obtain a standardized data set; the multi-source heterogeneous data set includes photosynthesis efficiency characteristics and agronomic measure parameters; Performing feature fusion on the standardized data set, reducing the dimension of the photosynthesis efficiency feature using principal component analysis, and generating a feature contribution matrix; Extracting principal component load values from the characteristic contribution matrix and combining them with the standardized weights of the agronomic practice parameters to construct a weight model of carbon sink change influencing factors; The weight model is combined with real-time environmental monitoring data, and key influencing factors are screened using dynamic thresholds to obtain an updated key influencing factor weight table.

4. The method according to claim 3, characterized in that The step of fusing the features of the standardized data set and reducing the dimension of the photosynthesis efficiency features using principal component analysis to generate a feature contribution matrix includes: Performing feature fusion on the light intensity response parameters and chlorophyll dynamic indicators in the standardized data set to generate a multidimensional feature vector; Inputting the multidimensional feature vector into a principal component analysis model to obtain a principal component vector and a variance ratio parameter; The characteristic contribution weight is calculated according to the variance ratio parameter; the characteristic contribution weight includes the light intensity contribution factor and the chlorophyll contribution factor; A feature contribution matrix is constructed based on the principal component vector and the feature contribution weight; the feature contribution matrix includes principal component load parameters and weight distribution parameters; The principal component load parameters and weight distribution parameters are normalized to generate a standardized feature contribution matrix.

5. The method according to claim 1, wherein The key influencing factor weight table and the multi-source heterogeneous data set are input into the trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks, including: Using a data fusion algorithm to perform spatiotemporal alignment on the data in the multi-source heterogeneous data set to obtain a fused multidimensional feature data set; After preprocessing the multidimensional feature data set, the data is combined with the key influencing factor weight table and weighted aggregation is performed using a feature extraction algorithm to generate a spatiotemporal feature set; Inputting the spatiotemporal feature set into the trained optimization evaluation model for dynamic convolution calculation to obtain the dynamic change trend of carbon sinks; A thermal map of the spatial distribution of carbon sinks in cotton fields is generated according to the dynamic change trend, and feature extraction is performed on the thermal map of the spatial distribution of carbon sinks to obtain spatiotemporal distribution features.

6. The method according to claim 5, characterized in that The spatiotemporal feature set is calculated using the following formula: Among them, F(x,t) represents the spatiotemporal feature set, ω i Represents the weight coefficient of the i-th feature, f i (x, t) represents the value of the i-th feature in the spatiotemporal dimension, α i represents the feature importance coefficient, and n represents the total number of features.

7. The method according to claim 1, characterized in that The correlation analysis of the spatiotemporal distribution characteristics and the meteorological continuity data to generate a carbon sink dynamic change curve includes: Performing correlation analysis on the meteorological continuity data and characteristic parameters of photosynthesis efficiency to obtain a dynamic response matrix of light energy utilization rate; Extracting the irrigation frequency and fertilization amount from the spatiotemporal distribution characteristics and combining them with the dynamic response matrix to calculate the gradient distribution of the dynamic adjustment coefficients; Inputting the gradient distribution into the carbon sink accumulation model and dividing it according to vegetation types to obtain a weighted interval of absorption rate; The dynamic adjustment coefficient is spatially interpolated based on the weight interval to generate a carbon sink dynamic change curve covering the target area.

8. The cotton field carbon sink dynamic assessment system based on multi-source data fusion is characterized by: The system comprises: A data fusion module is used to obtain multi-source data on the cotton field regional environment; the multi-source data includes remote sensing image data, meteorological continuity data, and soil organic carbon sampling data; and is also used to process the multi-source data using time window alignment technology to fuse the multi-source data to obtain a multi-source heterogeneous data set; A feature analysis module is used to extract features based on the multi-source heterogeneous data set and determine the contribution of each feature to carbon sink changes using principal component analysis to obtain a key influencing factor weight table; and is also used to input the key influencing factor weight table and the multi-source heterogeneous data set into a trained optimization evaluation model to obtain the spatiotemporal distribution characteristics of cotton field carbon sinks; The result generation module is used to perform correlation analysis on the spatiotemporal distribution characteristics and meteorological continuity data to generate a carbon sink dynamic change curve.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images