A remote sensing monitoring method for coastal wetlands in the bay

Abstract

The present invention provides a method for remote sensing monitoring of coastal wetlands in the bay, and relates to the fields of environmental science and data science and technology. The method for remote sensing monitoring of coastal wetlands in the bay specifically includes the following steps: S1. Remote sensing data collection; S2. Field data collection; S3. Ecological change analysis; S4. Data integration and processing; S5. Model construction; S6. Results integration. This method integrates multi-source data and interdisciplinary knowledge, and adopts advanced models and machine learning algorithms to provide a comprehensive and in-depth understanding of coastal wetland systems. Its high interpretability and transparency, as well as dynamic adaptation and learning capabilities, ensure high efficiency and accuracy, while promoting scientific research, policy making, collaboration and knowledge sharing with the public. The sustainability of this method and its ability to be flexibly applied to different scales have proposed an effective new approach for environmental monitoring and management.

Description

Technical Field

[0001] The invention relates to the field of environmental science and data science technology, and in particular to a remote sensing monitoring method for bay coastal wetlands. Background Art

[0002] The monitoring of coastal wetlands in the Gulf is based on the fact that coastal wetlands are habitats for a variety of wildlife and plants, especially important stopovers for many migratory birds. Monitoring these ecosystems helps protect their biodiversity from overexploitation and environmental change. Coastal wetlands help protect inland areas from storm surges, wave erosion and flooding. Their vegetation can slow down the impact of wind and water, reducing the risk of natural disasters such as floods. Wetlands have natural filtering capabilities and can remove pollutants from water through physical, chemical and biological processes. Monitoring can assess the health and efficiency of wetlands in purifying water. Coastal wetlands play an important role in the global carbon cycle, especially as one of the main carbon storage reservoirs. Monitoring wetlands can help assess their carbon sequestration potential, thereby playing a role in mitigating climate change. Coastal wetlands play an important role in maintaining regional water cycles and water quality. Monitoring activities can provide the necessary information to ensure the sustainable management of water resources. Coastal wetlands are essential for fisheries, tourism and other economic activities that rely on healthy ecosystems. Monitoring their health helps ensure the long-term sustainability of these economic benefits. Coastal wetlands provide an important platform for learning and studying ecosystem dynamics, interactions between species and responses to environmental changes. Monitoring provides valuable field data for research and education, and coastal wetlands are considered to be highly sensitive ecosystems to climate change. Monitoring can help assess their response to rising sea levels and other climate-related factors. Therefore, monitoring of the Gulf's coastal wetlands is critical to understanding the health of the ecosystem, ensuring the sustainability of its environmental, social and economic values, and guiding future conservation and management measures.

[0003] In the field of environmental science, especially when studying complex ecological processes such as the mechanism of wetland pollutant degradation, existing technologies mainly use the following methods: Traditional statistical methods: rely on basic statistical analysis techniques, such as linear regression, multivariate regression analysis, etc., to study the relationship between variables. This method is simple and direct, but it may not fully reveal the complex nonlinear relationships and interactions between variables; process-based models: these models are built based on the understanding of the physical, chemical and biological processes of wetland systems. Although these models can provide deep insights into how the system works, they often require a large amount of field data for parameterization and verification, which is not always easy to obtain; machine learning and deep learning methods: in recent years, with the improvement of computing power and the development of big data technology, they have been increasingly used in environmental science research. These methods, especially deep learning, can capture complex nonlinear relationships, but their "black box" characteristics make it more difficult to interpret and understand the model results. In the above-mentioned existing technologies, especially for research using machine learning and deep learning methods, although the models have high prediction accuracy, their internal decision-making processes are difficult to understand, which is a big problem in environmental policy and management decisions that require explanation of the model decision-making process. Especially for data-based models, such as machine learning and deep learning, a large amount of high-quality data is required to train the model, which is often a challenge in the field of environmental science because high-quality environmental data is costly to obtain and the data may have spatial and temporal discontinuities. Traditional statistical methods and machine learning models often rely on the laws of specific data sets, which may lead to a decrease in the prediction performance and generalization ability of the model when facing data from different locations or times. Even the most advanced models find it difficult to fully capture and understand the full complexity of natural systems, such as the multi-scale interactions of wetland systems, long-term dynamic changes, and the impact on biodiversity. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the shortcomings of the existing technology, the present invention provides a remote sensing monitoring method for bay coastal wetlands, which solves the obvious shortcomings of the existing technology in terms of spatial and temporal resolution, multi-source data fusion, automated analysis, rapid response to environmental changes, and cost and technical threshold.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for remote sensing monitoring of coastal wetlands in the bay, specifically comprising the following steps:

[0008] S1. Remote sensing data collection

[0009] Use high-resolution, multi-spectral, multi-temporal remote sensing data, preferably new generation remote sensing satellite data for data collection, and use atmospheric correction, radiation correction and terrain correction for the collected data to improve data accuracy;

[0010] S2. Field data collection

[0011] Conduct field visits and carry out more detailed on-site investigations, including the biological communities, physical and chemical conditions of the wetlands, combine with remote sensing data, conduct data verification and detailed analysis, develop on-site monitoring plans, and collect key indicator data, such as pollutant concentrations and plant distribution;

[0012] S3. Ecological change analysis

[0013] Using time series remote sensing data, we conduct an in-depth analysis of the life cycle of the coastal wetlands in the Bay, identify key change time points and events, and combine laboratory simulation data with field data to deeply explore the specific impact of wetland structural changes on pollutant purification functions;

[0014] S4. Data integration and processing

[0015] Comprehensively analyze remote sensing data with field survey and laboratory simulation data, and use multivariate statistical methods to analyze the correlation between data;

[0016] S5. Model construction

[0017] Based on the collected comprehensive data, mathematical models of wetland degradation mechanisms and purification functions are established and optimized to predict future trends;

[0018] S6. Integration of results

[0019] Summarize research results into review reports, data reports and analytical graphics.

[0020] Preferably, the remote sensing platform used in the S1 remote sensing data collection is Gaofen-1, Gaofen-2 or Landsat series, Sentinel series.

[0021] Preferably, the data collected by remote sensing in the S1 remote sensing data collection include but are not limited to surface temperature, vegetation index, surface cover, land and ocean color, wetland and water body range, soil moisture, altitude and terrain data, atmospheric composition, greenhouse gas concentration, and ice and snow cover.

[0022] Preferably, the data collected in the S2 field data collection includes, but is not limited to, soil samples, water quality samples, air samples, biological samples, topographic and geomorphic data, meteorological data, lighting and radiation data, acoustic data, and human activity data.

[0023] Preferably, the comprehensive data analysis of the data integration and processing in step S4 includes data format unification, time series alignment, and spatial alignment for integrating multi-source data.

[0024] Preferably, the multivariate statistical methods in the data integration and processing in step S4 include correlation analysis, principal component analysis, regression analysis, and cluster analysis.

[0025] Preferably, the S4 data integration and processing also includes exploratory data analysis for understanding the data set and laying the foundation for further analysis and modeling.

[0026] (III) Beneficial effects

[0027] The present invention provides a method for remote sensing monitoring of coastal wetlands in the bay. It has the following beneficial effects:

[0028] The present invention provides a method for remote sensing monitoring of coastal wetlands in the bay. By utilizing multi-source data and combining it with advanced data fusion technology, the method can provide a more comprehensive understanding and monitoring of the status of coastal wetlands. This comprehensive data utilization strategy helps to make up for the limitations that may exist in a single data source. The method adopts different types of models and algorithms, as well as the integration of other mathematical and statistical methods, thereby improving the prediction accuracy and the depth of environmental variable analysis. The application of these model sets helps to more comprehensively capture and explain the complexity of coastal wetland systems. By integrating highly explanatory machine learning technology and model visualization tools, the method can provide accurate predictions while also clearly showing the decision-making process and logic to users. This method can improve the transparency and trust of scientific research and policy decisions. By utilizing the adaptability of machine learning and artificial intelligence technology, the method can continuously learn and recalibrate from new data to reflect environmental changes and new scientific understandings, which ensures that the method can maintain high efficiency and high accuracy over time. The method integrates The technologies and knowledge from multiple disciplines such as ecology, climate science, geographic information science, and data science have promoted a deeper understanding of coastal wetland systems through interdisciplinary collaboration, and proposed more comprehensive and effective management strategies. Combining sophisticated ground monitoring data and extensive remote sensing data, this method can simultaneously respond to detailed research needs of specific locations and overall trend analysis of wide areas. The flexible application of this scale enables the method to be effective in different research and management contexts. Through open data platforms and tools, this method promotes collaboration and knowledge sharing among researchers, policymakers, local managers, and the public. This increased participation helps to generate monitoring and management solutions that are closer to actual needs and applications. Therefore, by combining multi-source data, interdisciplinary collaboration, and the integration of advanced models and algorithms, this method not only has significant advantages in monitoring accuracy and efficiency, but also achieves improvements in interpretability, adaptability, and stakeholder participation, providing a comprehensive and effective new approach for scientific research and sustainable management of coastal wetlands. DETAILED DESCRIPTION

[0029] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0030] Example:

[0031] The embodiment of the present invention provides a method for remote sensing monitoring of coastal wetlands in a bay, which specifically includes the following steps:

[0032] S1. Remote sensing data collection

[0033] Platform selection: According to scientific research needs, select appropriate remote sensing satellites or aviation platforms. Considering the monitoring of the Bay Wetland, priority is given to remote sensing platforms that can provide high resolution and multi-spectral capabilities, such as Gaofen-1, Gaofen-2 or foreign earth observation satellites such as Landsat series and Sentinel series.

[0034] Data requirements: Clarify the types of remote sensing data required (such as visible light, infrared, thermal infrared), time series (the need for multi-temporal monitoring), and resolution (including spatial, spectral, and temporal resolution).

[0035] Data Acquisition

[0036] Satellite data acquisition: Download appropriate remote sensing data sets according to needs through major satellite data centers or platforms, such as the Center for Resource Satellite Applications (CRESDA) and the United States Geological Survey (USGS).

[0037] Data requests and customization: If publicly available data do not fully meet your needs (e.g., you need data for a specific period or a special level of processing), consider submitting a data customization request to the service provider.

[0038] There are many types of data collected by remote sensing technology. Depending on the remote sensing instrument and the target used, the following types of data can be collected, including but not limited to:

[0039] Surface temperature: Temperature data of the ground or surface objects can be obtained through thermal infrared remote sensing.

[0040] Vegetation indices: such as the Normalized Difference Vegetation Index (NDVI), which is used to assess the health and density of vegetation on the Earth's surface.

[0041] Surface cover: including vegetation, water bodies, urban buildings, and bare soil, obtained through classification of multispectral and hyperspectral remote sensing data.

[0042] Land and ocean color: reflects the color information of water bodies and land surfaces, and can be used to monitor water quality and vegetation types.

[0043] Wetland and water body extent: Synthetic aperture radar (SAR) radar remote sensing technology can obtain the extent and dynamic changes of wetlands and water bodies even under cloud cover conditions.

[0044] Soil moisture: Assess the moisture conditions on the soil surface or deep layer through radar remote sensing or microwave remote sensing technology.

[0045] Elevation and topographic data: Topographic relief and elevation data are collected through optical remote sensing and radar altimetry techniques.

[0046] Atmospheric composition: Detect gas components and pollutants in the atmosphere through spectral remote sensing.

[0047] Greenhouse gas concentrations: especially carbon dioxide and methane, are extremely important for climate change research.

[0048] Snow and ice cover: Use remote sensing techniques to monitor changes in the extent of glaciers, ice caps, and seasonal snow cover.

[0049] Data preprocessing

[0050] Geometric correction: Ensure the geometric accuracy of data, including correcting the position of ground features so that they correspond accurately to the geographic coordinate system (GCS).

[0051] Atmospheric correction: Eliminate the impact of atmospheric conditions (such as clouds and fog) on ​​the reflectivity of remote sensing data and extract real surface information.

[0052] Radiation correction: Convert the digital measurement value (DN value) of remote sensing data into the physical quantity of ground reflectivity to obtain more realistic surface characteristic information.

[0053] Data analysis

[0054] Feature extraction: Utilize multispectral and / or hyperspectral data to extract relevant wetland characteristics (such as vegetation type, coverage, and water quality parameters) through classification and index calculation methods.

[0055] Time series analysis: Use multi-temporal data to analyze wetland trends over time, including ecological changes and pollution conditions.

[0056] S2. Field data collection

[0057] Preparation for site investigation

[0058] Clarify the survey objectives: Based on the general situation of the wetland, determine the main objectives of the survey, such as wetland vegetation distribution, species diversity, and ecological environment status.

[0059] Choose the right time for the survey: Choose a suitable season and weather for the field survey. According to the plant growth cycle, choose the peak growth period for the survey to obtain more accurate data.

[0060] Survey tool preparation: Prepare necessary survey tools, such as GPS positioning equipment, measuring tape, soil sampler, notebook or electronic recording device.

[0061] Execution of on-site investigation

[0062] Site positioning and recording: Arrive at the survey site and use GPS positioning equipment to record the specific latitude and longitude of the sample plot.

[0063] Sample plot setting and measurement: According to research needs, set up standard sample plots (such as 1×1m) in the target area, and measure and record the basic information of the sample plot, such as size, slope, and altitude.

[0064] Habitat and plant data collection:

[0065] Record habitat types, such as wetlands, soil types.

[0066] Conduct a comprehensive survey of the plants in the sample plot, and record the plant layers (herb layer, shrub layer), Chinese name and scientific name of the species, growth period, quantity, height, and coverage information.

[0067] Pay attention to threatened factors and record any factors that may have a negative impact on wetland ecology.

[0068] Collection of other ecological information: Relevant microbial community information, water quality parameters, and soil chemical properties data can also be collected to conduct a more comprehensive ecological analysis.

[0069] Data collation and analysis

[0070] Data collation: Organize and archive the data recorded on site, including the input of electronic data and the scanning backup of paper records.

[0071] Data analysis: Depending on the purpose of the survey, use appropriate statistical or ecological analysis methods to process the data, such as species diversity index calculation and ecological niche analysis.

[0072] S3. Ecological change analysis

[0073] Remote sensing data, field surveys and historical monitoring data were used to study the changes in the ecological environment of the Jiaozhou Bay coastal wetlands.

[0074] Based on the changes in the main influencing factors in the natural environment and coastal social environment, the driving factors of wetland ecological changes are screened.

[0075] Research on the relationship between wetland structure and purification function:

[0076] Laboratory simulation is used to analyze the distribution, composition and quantity of plants, substrates and microorganisms in wetland systems, and to explore the relationship between wetland structure and pollutant purification effects.

[0077] By studying the impact of different environmental conditions on wetland structure and purification function, the key physical, chemical and biological processes and their controlling factors in wetland purification of pollutants are identified.

[0078] Work plan for the relationship between wetland purification and pollutant input flux:

[0079] Analyze the impact of human activities on wetland ecosystems and functions based on the evolution laws of coastal wetlands.

[0080] Quantify the distribution, migration, transformation and purification mechanisms of different forms of pollutants in wetland systems.

[0081] By integrating these research contents, we can conduct an in-depth analysis of the relationship between wetland ecological changes and purification functions. For example, the research results can show that:

[0082] How the specific structure of wetland plants, soil, and microbial assemblages affects their ability to purify specific pollutants.

[0083] How different socio-economic activities, land use patterns and physical changes drive ecological changes in wetlands.

[0084] How the diversity and complexity of wetland microbial communities promote biodegradation of multiple pollutants.

[0085] S4. Data integration and processing

[0086] Integrate data from multiple sources

[0087] Data format unification: Different data sources often have different formats, including but not limited to CSV files, shape files (GIS), grid data. Data format unification means converting all data into a common format to facilitate subsequent processing. This may involve the use of data conversion tools or scripts.

[0088] Time series alignment: Remote sensing data, field survey data, and historical monitoring data may be recorded at different time points. To integrate these data reasonably, they need to be time series aligned to ensure that the time points match or correspond within an acceptable error range.

[0089] Spatial alignment: In addition to time series, you also need to ensure that your data is spatially aligned. This means that all datasets need to be calibrated to the same geographic coordinate system (GCS) and projection system in order to perform spatial analysis.

[0090] Exploratory Data Analysis (EDA)

[0091] Visual data analysis: Use charts, scatter plots, and heat map visualization tools to explore data distribution, trends, and relationships between variables.

[0092] Statistical Analysis: Descriptive statistics were performed to obtain indicators of central tendency and dispersion of the data. Correlation analysis (such as Pearson or Spearman correlation coefficients) was used to identify linear or nonlinear relationships between variables.

[0093] Advanced Data Processing

[0094] Outlier detection: Use statistical methods to identify and process or remove outliers in data to prevent them from adversely affecting the analysis results.

[0095] Missing Value Handling: Determine the most appropriate strategy to handle missing data, such as interpolation methods, removing records with missing data, or estimating missing values ​​using machine learning algorithms.

[0096] Data smoothing: Applying techniques, such as moving average or Gaussian smoothing, to reduce random variation or noise in the data and make trends more apparent.

[0097] Multivariate statistical analysis

[0098] Correlation analysis: Pearson correlation coefficient and Spearman rank correlation coefficient statistical methods were used to evaluate the correlation between variables.

[0099] Principal Component Analysis (PCA): This is a data dimensionality reduction technique that can be used to reduce the complexity of a dataset while retaining the most important information.

[0100] Regression analysis: Building a model to predict the relationship between one or more independent variables (explanatory variables) and a dependent variable (response variable).

[0101] Cluster analysis: Dividing a data set into groups or “clusters” based on the data’s characteristics, which helps identify patterns or classifications in the data.

[0102] S5. Model construction

[0103] Feature Engineering

[0104] Variable selection: Based on the results of the EDA, variables with a high correlation with the wetland pollutant removal mechanism are selected as features of the model. This may involve removing highly collinear features to reduce model complexity.

[0105] Feature creation: Creating new features based on existing data, such as discretizing certain continuous variables or calculating new indicators from existing variables. This can help the model capture more complex patterns.

[0106] Feature transformation: Apply normalization and standardization techniques to adjust the scale of features to make them more suitable for model input.

[0107] Select Model

[0108] Model type selection: Choose the appropriate model type based on the research objectives and data characteristics. For example, if the goal is to predict the purification effect of Jiaozhou Bay wetlands on a certain pollutant, a regression model may be selected; if the goal is to classify the purification efficiency level of wetlands, a classification model may be selected.

[0109] Consider machine learning applications: Depending on the complexity of the problem and the amount of data, consider whether to use machine learning models such as random forests, support vector machines, or deep learning models, which may be better able to capture complex patterns in the data.

[0110] Model training and validation

[0111] Split data: Split the dataset into training and testing sets to ensure that model training and evaluation are performed on independent data.

[0112] Hyperparameter tuning: Use cross-validation methods to tune the model's hyperparameters and find the best parameter combination to improve the model's accuracy and generalization ability.

[0113] Model evaluation: The performance of the model is evaluated by calculating the accuracy, recall, and F1 score indicators on the test set. For regression problems, the mean square error (MSE) and root mean square error (RMSE) indicators can be used.

[0114] S6. Integration of results

[0115] Summarize research results into review reports, data reports and analytical graphics.

[0116] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for remote sensing monitoring of coastal wetlands in a bay, characterized in that: The specific steps include: S1. Remote sensing data collection First, we select appropriate remote sensing satellites, then clarify the types of remote sensing data, time series and resolution requirements, obtain data through satellite data centers, and submit data customization requests if necessary if publicly available data cannot fully meet the needs. The collected data are then preprocessed through geometric correction, atmospheric correction and radiation correction. Finally, wetland features are extracted through feature extraction, and the changing trends of wetlands over time are analyzed through time series analysis. S2. Field data collection First, the survey objectives are clearly defined, including wetland vegetation distribution and species diversity, and suitable seasons and weather are selected for field surveys. The sample locations are recorded using GPS positioning, standard sample plots are set up and basic information is measured, including habitat types, plant layers, species names, quantities, heights, and coverage. Microbial community information, water quality parameters, and soil chemical properties data are also collected. Finally, the data are sorted and archived, and the data are processed through species diversity index calculation or niche analysis methods. S3. Ecological change analysis Combine remote sensing data, field surveys and historical monitoring data to study the ecological and environmental changes in the bay's coastal wetlands, screen driving factors, analyze the distribution and composition of plants, substrates and microorganisms in wetland systems through laboratory simulation, explore the relationship between wetland structure and pollutant purification effect, identify the key physical, chemical and biological processes of wetland purification of pollutants, analyze the impact of human activities on wetland ecosystems, quantify the distribution, migration and transformation and purification mechanisms of pollutants, integrate research content, and deeply analyze the relationship between wetland ecological changes and purification functions, reveal the impact of the specific structure of wetland plants, soils, and microorganisms on pollutant purification capacity, the driving effect of socio-economic activities and natural changes on wetland ecology, and the role of microbial community diversity in promoting pollutant biodegradation; S4. Data integration and processing First, the multi-source data is unified into a format, including CSV, GIS or grid data, and converted using tools or scripts to align time series and spatial coordinates to ensure that the data matches in time and space. Exploratory data analysis is performed using charts and scatter plot visualization tools, and descriptive statistics and correlation analysis are performed. Advanced data is used to detect outliers, handle missing values ​​and smooth data. Multivariate statistical analysis is then performed. Multivariate statistical analysis uses correlation analysis to evaluate variable relationships, principal component analysis for data dimensionality reduction, regression analysis for predictive model construction, and cluster analysis for pattern recognition to comprehensively process and analyze data. S5. Model construction First, according to the results of EDA, select variables with high correlation with the wetland pollutant degradation mechanism as the features of the model, create new features based on existing data, or use normalization and standardization techniques to adjust the scale of the features. Then, according to the research objectives and the characteristics of the data, select the model type, including regression models and classification models, and consider using machine learning models to capture complex patterns in the data. Split the data set into training sets and test sets. Finally, evaluate the performance of the model by calculating the accuracy, recall rate, and F1 score indicators on the test set. For regression problems, the mean square error and root mean square error indicators can be used. S6. Integration of results Summarize research results into review reports, data reports and analytical graphics.

2. A method for remote sensing monitoring of coastal wetlands in a bay according to claim 1, characterized in that: The remote sensing satellite in step S1 is Gaofen-1, Gaofen-2 or Landsat series, Sentinel series.

3. The method for remote sensing monitoring of coastal wetlands in a bay according to claim 1, characterized in that: The data collected by the remote sensing satellite in step S1 include surface temperature, vegetation index, surface cover, land and ocean color, wetland and water body range, soil moisture, altitude and terrain data, atmospheric composition, greenhouse gas concentration, and ice and snow cover.