A method and system for estimating soil organic carbon content in wetlands with different restoration years

Abstract

The present invention provides a method for estimating the soil organic carbon content of wetlands with different restoration years, which relates to the field of wetland carbon sink accounting technology. The method includes: setting up multiple sampling points within the target study area and collecting soil samples; preprocessing satellite image data and determining a first input feature, and determining a second input feature based on the geographic coordinates of the sampling points, and inputting the above features into a regression prediction model to obtain a calculation result; generating a soil organic matter content data distribution based on the calculation result; using space instead of time, dividing sample areas with different restoration years based on land use history or restoration measures records, and within each sample area, statistically calculating the soil organic matter content and total amount based on the soil organic matter content data distribution to determine the soil organic carbon content of different restoration years. The present invention utilizes remote sensing technology and machine learning algorithms to achieve rapid and non-destructive estimation of wetland soil organic carbon content, reducing the workload and cost of field sampling.

Description

Technical Field

[0001] The present invention relates to the technical field of wetland carbon sink accounting, and in particular to a method and system for estimating the organic carbon content of wetland soil at different restoration years. Background Art

[0002] Wetlands are vast reservoirs of organic carbon in terrestrial ecosystems and a crucial component of Earth's carbon cycle. Wetlands cover only 5% to 8% of the total land area, yet they are one of the world's largest soil carbon reservoirs, storing approximately 455 to 700 petagrams (Pg) of organic carbon (1 Pg = 1 Gt = 1 billion tons). This represents 20% to 30% of the total surface carbon reserves in terrestrial ecosystems. Their carbon storage per unit area ranks first among all terrestrial ecosystems, three times that of forest ecosystems.

[0003] The accelerated pace of global warming and human development and utilization have led to the shrinkage of wetland areas, the deterioration of wetland ecosystems, and changes in their carbon cycle. With the introduction of my country's "dual carbon" goals, a series of major ecological restoration projects have been launched to maintain the stability of wetland carbon sinks and restore the carbon sequestration function of degraded wetlands. Monitoring and accounting for carbon sinks in key wetlands and assessing the impact of major ecological restoration projects on the carbon sequestration function of wetland ecosystems have become urgent issues. However, there is currently a lack of monitoring of carbon sink dynamics in key wetland restoration projects, and understanding of the synergistic effects of wetland ecological restoration projects and carbon sequestration and sink enhancement is incomplete. There is an urgent need to monitor and account for carbon sinks in key wetland restoration areas.

[0004] Due to the distinct regionality and complexity of wetland distribution, encompassing diverse types and dynamic variations, current wetland carbon accounting models and methods still face significant challenges in achieving accurate results. Numerous researchers have estimated wetland carbon storage and sequestration across diverse study areas in my country. However, these estimates are subject to significant uncertainty due to differences in wetland type, geographic conditions, and estimation methods. The main reasons for this uncertainty are: First, a lack of long-term data. The long time span of wetland restoration makes it difficult to continuously monitor organic carbon accumulation rates over different years, resulting in a lack of long-term trend data and hindering accurate analysis of the relationship between restoration years and organic carbon storage. Second, spatial heterogeneity impacts wetland ecosystems. Hydrological conditions, vegetation types, and soil composition vary significantly across wetland ecosystems, leading to diverse carbon accumulation characteristics across different regions. This makes establishing a unified carbon storage estimation method challenging across a wide area. Third, sample acquisition is challenging. Wetland environments are complex, and collecting deep soil samples, in particular, is limited by technical and equipment limitations, making accurate organic carbon content data difficult to obtain. Variation in organic carbon content at different depths also hinders accurate estimates. Therefore, how to accurately and efficiently measure the soil organic carbon content in wetland restoration areas is a hot topic and difficulty in current research. Summary of the Invention

[0005] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a method and system for estimating the soil organic carbon content of wetlands with different restoration years. By integrating remote sensing data, geospatial information and soil sampling data, and conducting machine learning, accurate estimation of the soil organic carbon content of wetlands with different restoration years can be achieved, solving the problems of traditional methods in wetland ecosystem carbon storage assessment, such as lack of data, time-consuming and labor-intensive, and limited accuracy.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for estimating soil organic carbon content in wetlands with different restoration years includes:

[0008] Setting up a plurality of sampling points in the target study area and collecting soil samples at the sampling points;

[0009] Preprocess the satellite image data of the target study area to obtain preprocessed remote sensing image data;

[0010] Determining a first input feature based on the preprocessed remote sensing image data and determining a second input feature based on the geographic coordinates of the sampling points; the first input feature includes: an index band, a radar band index, a spectral band index, and a texture feature image; the second input feature includes a feature value of each sampling point;

[0011] Inputting the first input feature and the second input feature into a trained regression prediction model of soil organic matter content to obtain a calculation result;

[0012] generating soil organic matter content data distribution according to the calculation results;

[0013] Using the method of space instead of time, sample areas with different restoration years were divided according to the land use history or restoration measures records of the target study area. Within each sample area, the soil organic matter content and total amount were statistically calculated based on the soil organic matter content data distribution to determine the soil organic carbon content in different restoration years.

[0014] Preferably, the depth of the soil sample is 0 to 30 cm.

[0015] Preferably, the satellite image data includes: Sentinel-1 and Sentinel-2 data.

[0016] Preferably, the method of preprocessing the satellite image data of the target study area includes: cloud removal and noise removal.

[0017] Preferably, determining the first input feature based on the preprocessed remote sensing image data includes:

[0018] Determining a gray level co-occurrence matrix based on the pre-processed remote sensing image data;

[0019] Performing band extraction based on the pre-processed remote sensing image data to obtain each of the index bands;

[0020] Extracting a band index based on the preprocessed remote sensing image data to obtain the radar band index and the spectral band index;

[0021] Perform feature extraction according to the gray-level co-occurrence matrix to obtain texture features;

[0022] The texture feature image is generated according to the texture features.

[0023] Preferably, the index bands include: normalized vegetation index, enhanced vegetation index and building index.

[0024] Preferably, the regression prediction model of soil organic matter content is obtained by training a random forest regression model.

[0025] Preferably, the evaluation indicators of the random forest regression model include: determination coefficient and root mean square error.

[0026] The present invention also provides a system for estimating soil organic carbon content in wetlands with different restoration years, comprising:

[0027] A sampling module is used to set up multiple sampling points in the target study area and collect soil samples at the sampling points;

[0028] Image preprocessing module, used to preprocess satellite image data of the target study area to obtain preprocessed remote sensing image data;

[0029] A feature determination module is configured to determine a first input feature based on the preprocessed remote sensing image data and a second input feature based on the geographic coordinates of the sampling points; the first input feature includes an index band, a radar band index, a spectral band index, and a texture feature image; and the second input feature includes a feature value of each sampling point.

[0030] a machine learning module, configured to input the first input feature and the second input feature into a trained regression prediction model of soil organic matter content to obtain a calculation result;

[0031] A data distribution generating module, configured to generate soil organic matter content data distribution according to the calculation results;

[0032] The restoration estimation module is used to use the method of using space instead of time to divide the sample areas into different restoration years according to the land use history or restoration measure records of the target study area. In each sample area, the soil organic matter content and total amount are statistically analyzed based on the soil organic matter content data distribution to determine the soil organic carbon content in different restoration years.

[0033] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0034] The present invention provides a method for estimating the soil organic carbon content of wetlands at different restoration years, comprising: setting up multiple sampling points within a target study area and collecting soil samples at the sampling points; preprocessing satellite image data of the target study area to obtain preprocessed remote sensing image data; determining a first input feature based on the preprocessed remote sensing image data, and determining a second input feature based on the geographic coordinates of the sampling points; the first input feature includes an index band, a radar band index, a spectral band index, and a texture feature image; the second input feature includes characteristic values ​​of each sampling point; inputting the first and second input features into a trained regression prediction model for soil organic matter content to obtain a calculation result; generating a soil organic matter content data distribution based on the calculation result; using a spatial substitution method to replace time, dividing the target study area into sample plots with different restoration years based on the land use history or restoration measures records; and within each sample plot, calculating the soil organic matter content and total amount based on the soil organic matter content data distribution to determine the soil organic carbon content at different restoration years. The present invention utilizes remote sensing technology and machine learning algorithms to achieve rapid and non-destructive estimation of wetland soil organic carbon content, reducing the workload and cost of field sampling. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0036] Figure 1 A flow chart of a method provided by an embodiment of the present invention;

[0037] Figure 2 A schematic diagram of the SOC random forest regression model modeling process provided by an embodiment of the present invention;

[0038] Figure 3 The organic carbon content distribution diagram provided by the embodiment of the present invention;

[0039] Figure 4Schematic diagram of 0-30cm organic carbon content in restored plots of different years provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 efforts are within the scope of protection of the present invention.

[0041] The purpose of this invention is to provide a method for estimating the organic carbon content of wetland soils with different restoration years. By using remote sensing technology and machine learning algorithms, a rapid and non-destructive estimation of the organic carbon content of wetland soils is achieved, reducing the workload and cost of field sampling.

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] Figure 1 A flow chart of the method provided in the embodiment of the present invention is shown in FIG. Figure 1 As shown, the present invention provides a method for estimating the organic carbon content of wetland soil with different restoration years, comprising:

[0044] Step 100: setting up multiple sampling points in the target study area and collecting soil samples at the sampling points;

[0045] Step 200: pre-processing the satellite image data of the target study area to obtain pre-processed remote sensing image data;

[0046] Step 300: Determine a first input feature based on the preprocessed remote sensing image data, and determine a second input feature based on the geographic coordinates of the sampling points; the first input feature includes: an index band, a radar band index, a spectral band index, and a texture feature image; the second input feature includes a feature value of each sampling point;

[0047] Step 400: Input the first input feature and the second input feature into the trained regression prediction model of soil organic matter content to obtain a calculation result;

[0048] Step 500: Generate soil organic matter content data distribution according to the calculation results;

[0049] Step 600: Using the method of replacing time with space, sample plots with different restoration years are divided according to the land use history or restoration measures records of the target study area. Within each sample plot, the soil organic matter content and total amount are statistically calculated based on the distribution of soil organic matter content data to determine the soil organic carbon content in different restoration years.

[0050] Specifically, such as Figure 2 As shown, the present invention integrates remote sensing data, geospatial information, and soil sampling data to construct a random forest regression model for machine learning. This model can accurately estimate the organic carbon content in wetland soils with different restoration years, solving the problems of traditional methods in wetland ecosystem carbon storage assessment, such as lack of data, time-consuming and labor-intensive methods, and limited accuracy. The specific implementation steps are as follows:

[0051] 1. Field sampling and testing: Collect soil samples with a thickness of 0-30 cm in the target study area, with one sampling point required for every 16.5 square kilometers.

[0052] Within the target study area, this example uses a fixed-volume (V) soil auger to collect soil samples primarily for soil organic carbon determination. For each soil sampling plot, the 0-10 cm, 10-20 cm, and 20-30 cm intervals are used. Using a soil auger with an inner diameter ≥ 5 cm, sample at least 200 g of mixed sample per layer. After removing non-soil matter, the samples are packaged in ziplock bags, labeled, and brought back to the laboratory. Air-dried soil samples are retained to preserve naturally occurring soil layers. The required number of sampling points is one per 16.5 square kilometers, and the longitude and latitude of the sampling points are recorded.

[0053] 2. Satellite Data Selection and Processing: Sentinel-1 and Sentinel-2 data, including optical and radar imagery, are used. Preprocessing steps include cloud removal and noise reduction to clean the raw image data. Various index images are generated, such as NDVI (Normalized Difference Vegetation Index), EVI (Enhanced Vegetation Index), and IBI (Infrastructure Index).

[0054] This example obtains Sentinel-1 and Sentinel-2 satellite remote sensing data from relevant data sources, such as the official website of the European Space Agency (ESA). The data ensures coverage of the study area and selects an appropriate time period to obtain representative surface information. The acquired data is then corrected for geometry, atmosphere, radiometry, and terrain to remove atmospheric influences on surface reflectivity and improve data accuracy. Furthermore, cloud mask images are generated using the cloud detection bands in the Sentinel-2 data, removing pixels obscured by clouds to ensure the accuracy of subsequent analysis.

[0055] 3. Extract texture feature images and determine the characteristic values ​​of sampling points: Based on the preprocessed remote sensing image data, the gray-level co-occurrence matrix (GLCM) method is used to extract various spectral indices, radar indices and texture characteristics of the wetland surface from the biophysical parameters. At the same time, the characteristic values ​​of each sampling point are obtained according to the geographic coordinates of the soil sampling points.

[0056] This embodiment selects specific bands in the Sentinel-2 data (such as the red edge band, the near infrared band, etc.), calculates the gray level co-occurrence matrix, and extracts texture features such as contrast, energy, homogeneity, and correlation from the gray level co-occurrence matrix. Based on the texture features, an SOC texture feature image is generated. The extraction of other index bands mainly includes calculating the normalized difference vegetation index (NDVI), the enhanced vegetation index (EVI), and the infrared band index (IBI). These indices can reflect information such as vegetation growth status, soil moisture, and surface cover, which are helpful for the prediction of SOC. Finally, radar band indexes such as radar backscatter coefficients are extracted from the Sentinel-1 data, and the spectral bands in the Sentinel-2 data are combined to construct a multi-source data set. At the same time, the characteristic values ​​of each sampling point are obtained according to the geographic coordinates of the soil sampling point.

[0057] 4. Model construction and training: Use the random forest regression model to train and verify the measured soil organic carbon content to determine the model accuracy.

[0058] This embodiment uses SOC texture feature images, index bands such as NDVI, EVI, IBI, and spectral and radar band indexes as input features, and uses the random forest algorithm to establish a regression prediction model for SOC content. By adjusting the parameters of the random forest (such as the number of trees, maximum depth, etc.), the model performance is optimized. Subsequently, an independent validation data set is used to verify the model, and based on the validation results, the model parameters are adjusted to improve the prediction accuracy of the model. Finally, methods such as cross-validation can be used to further evaluate the stability and generalization ability of the model.

[0059] 5. Using space instead of time to obtain soil organic carbon content in different years: Based on the verified random forest regression model, the spatiotemporal distribution map of wetland soil organic carbon was generated, and the soil organic carbon content in different years was determined by using the method of space instead of time.

[0060] This example applies an optimized random forest model to a dataset from the entire study area. Based on the model's calculations, the SOC content distribution is generated and visualized to demonstrate the spatial distribution of SOC content. Using space as a proxy for time, the study area is divided into plots with different restoration years based on land use history or restoration measures. Within each plot, the SOC content and total amount are calculated to determine the soil organic carbon content in different restoration years.

[0061] As an optional implementation, this example evaluated the carbon sequestration and sink benefits of a wetland restoration project in the Hongze Lake Wetland National Nature Reserve in Sihong, Jiangsu Province (hereinafter referred to as the Hongze Lake Reserve). To track changes in carbon storage over the years within the restoration area, a random forest regression model was used to estimate soil organic carbon content in wetland soils at different restoration years. The study was conducted using the following methods:

[0062] 1. Field sampling and testing: In the target study area, soil samples were collected using a fixed volume (V) soil auger, mainly for the determination of soil organic carbon, etc. In each soil sampling area, soil samples were collected at 0-10cm, 10-20cm, and 20-30cm. A soil auger with an inner diameter of 5cm or more was used to collect more than 200g of mixed samples for each layer. After removing non-soil matter, the samples were packaged in self-sealing bags, labeled, and brought back to the laboratory. The naturally occurring layer soil samples were air-dried and retained. A total of 80 sampling points in the entire area were used to determine the soil organic carbon content.

[0063] 2. Satellite impact data selection and processing: Using Sentinel-1 and Sentinel-2 satellite data, various biophysical parameters including the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI), as well as texture feature images extracted by the Gray Level Co-occurrence Matrix (GLCM), were extracted as input features for the random forest regression model.

[0064] 3. Extract texture feature images and determine sampling point feature values:

[0065] (1) Input data: Sentinel-1 and Sentinel-2 data, including optical and radar images, are used. Preprocessing steps include cloud removal and noise removal to clean up the raw image data. Various types of index images are generated, such as NDVI (Normalized Difference Vegetation Index), EVI (Enhanced Vegetation Index), and IBI (Building Index).

[0066] (2) Extracting eigenvalues ​​according to step (1) and using the gray-level co-occurrence matrix (GLCM) method to extract texture feature images from biophysical parameters, while obtaining the eigenvalues ​​of each sampling point according to the geographic coordinates of the soil sampling point.

[0067] 4. Model training: Based on field soil sampling and stratification, the organic carbon content of the soil layer 0-0.3m was used as the target value, and the random forest regression model was used for training. Through the calibration and verification process, the relevant evaluation indicators and the determination coefficient (R 2 ) reached 0.75, the root mean square error (RMSE) was 5.5, and the model simulation effect was good;

[0068] 5. Using space instead of time to obtain soil organic carbon content in different years: Based on the random forest model, the soil organic carbon content in the 0-30cm soil layer of Hongze Lake Nature Reserve was estimated ( Figure 3 ), the results showed that the organic carbon content was as high as 40.97g / kg and as low as 9.45g / kg. More than 80% of the area had soil organic carbon content between 16g / kg and 21g / kg. The spatial vector data of the restoration area was used to submerge the soil organic carbon content result grid to obtain the average organic carbon content of the soil in different restoration years ( Figure 4 The average organic carbon content in plots with different restoration years in the Hongze Lake Nature Reserve first increased and then decreased with the restoration period. The highest organic carbon content was found in the restored plots in 2017 (26.46 g / kg), followed by the 2016 plot (24.37 g / kg), and the lowest was found in the unrestored plots (8.99 g / kg).

[0069] Corresponding to the above method, such as Figure 2 As shown, the present invention also provides a system for estimating the organic carbon content of wetland soils with different restoration years, comprising:

[0070] A sampling module is used to set up multiple sampling points in the target study area and collect soil samples at the sampling points;

[0071] Image preprocessing module, used to preprocess satellite image data of the target study area to obtain preprocessed remote sensing image data;

[0072] A feature determination module is configured to determine a first input feature based on the preprocessed remote sensing image data and a second input feature based on the geographic coordinates of the sampling points; the first input feature includes an index band, a radar band index, a spectral band index, and a texture feature image; and the second input feature includes a feature value of each sampling point.

[0073] a machine learning module, configured to input the first input feature and the second input feature into a trained regression prediction model of soil organic matter content to obtain a calculation result;

[0074] A data distribution generating module, configured to generate soil organic matter content data distribution according to the calculation results;

[0075] The restoration estimation module is used to use the method of using space instead of time to divide the sample areas into different restoration years according to the land use history or restoration measure records of the target study area. In each sample area, the soil organic matter content and total amount are statistically analyzed based on the soil organic matter content data distribution to determine the soil organic carbon content in different restoration years.

[0076] The beneficial effects of the present invention are as follows:

[0077] (1) The present invention integrates Sentinel-1 and Sentinel-2 satellite data, and effectively utilizes high-resolution optical and radar data through preprocessing and cloud masking, thereby enhancing the information precision, richness, and accuracy of soil organic matter content (SOC) prediction.

[0078] (2) The present invention constructs an efficient random forest regression model that can effectively handle a large number of nonlinear relationships and complex feature interactions, reduce the risk of overfitting, and improve the prediction accuracy and generalization ability of SOC content, thereby achieving accurate estimation of wetland soil organic carbon content.

[0079] (3) This invention is not only applicable to the estimation of soil organic carbon content in wetlands with different restoration years, but can also be extended to the assessment of carbon reserves in other types of ecosystems.

[0080] (4) The present invention utilizes remote sensing technology and machine learning algorithms to achieve rapid and non-destructive estimation of wetland soil organic carbon content, reducing the workload and cost of field sampling.

[0081] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0082] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for estimating soil organic carbon content in wetlands with different restoration years, characterized in that: include: Setting up a plurality of sampling points in the target study area and collecting soil samples at the sampling points; Preprocess the satellite image data of the target study area to obtain preprocessed remote sensing image data; Determining a first input feature based on the preprocessed remote sensing image data, and determining a second input feature based on the geographic coordinates of the sampling points; The first input feature includes: index band, radar band index, spectral band index and texture feature image; the second input feature includes the feature value of each sampling point; Inputting the first input feature and the second input feature into a trained regression prediction model of soil organic matter content to obtain a calculation result; generating soil organic matter content data distribution according to the calculation results; Using the method of space replacing time, based on the land use history or restoration measures records of the target study area, sample plots with different restoration years were divided. Within each sample plot, the soil organic matter content and total amount were statistically analyzed based on the soil organic matter content data distribution to determine the soil organic carbon content in different restoration years; Determining a first input feature based on the preprocessed remote sensing image data includes: Determining a gray level co-occurrence matrix based on the pre-processed remote sensing image data; Performing band extraction based on the pre-processed remote sensing image data to obtain each of the index bands; Extracting a band index based on the preprocessed remote sensing image data to obtain the radar band index and the spectral band index; Perform feature extraction according to the gray-level co-occurrence matrix to obtain texture features; The texture feature image is generated according to the texture features.

2. The method for estimating soil organic carbon content in wetlands with different restoration years according to claim 1, characterized in that: The depth of the soil sample is 0 to 30 cm.

3. The method for estimating soil organic carbon content in wetlands with different restoration years according to claim 1, characterized in that: The satellite image data includes: Sentinel-1 and Sentinel-2 data.

4. The method for estimating soil organic carbon content in wetlands with different restoration years according to claim 1, characterized in that: The preprocessing methods for satellite image data of the target study area include: cloud removal and noise removal.

5. The method for estimating soil organic carbon content in wetlands with different restoration years according to claim 1, characterized in that: The index bands include: normalized vegetation index, enhanced vegetation index and building index.

6. The method for estimating soil organic carbon content in wetlands with different restoration years according to claim 1, characterized in that: The regression prediction model of soil organic matter content is obtained by training a random forest regression model.

7. The method for estimating soil organic carbon content in wetlands with different restoration years according to claim 6, characterized in that: The evaluation indicators of the random forest regression model include: determination coefficient and root mean square error.

8. A system for estimating soil organic carbon content in wetlands with different restoration years, characterized by: include: A sampling module is used to set up multiple sampling points in the target study area and collect soil samples at the sampling points; Image preprocessing module, used to preprocess satellite image data of the target study area to obtain preprocessed remote sensing image data; a feature determination module, configured to determine a first input feature based on the preprocessed remote sensing image data, and to determine a second input feature based on the geographic coordinates of the sampling points; The first input feature includes: index band, radar band index, spectral band index and texture feature image; the second input feature includes the feature value of each sampling point; a machine learning module, configured to input the first input feature and the second input feature into a trained regression prediction model of soil organic matter content to obtain a calculation result; A data distribution generating module, configured to generate soil organic matter content data distribution according to the calculation results; A restoration estimation module is used to use a spatial substitution method to divide the target study area into sample plots with different restoration years based on the land use history or restoration measures records. Within each sample plot, the soil organic matter content and total amount are calculated based on the soil organic matter content data distribution to determine the soil organic carbon content in different restoration years; Determining a first input feature based on the preprocessed remote sensing image data includes: Determining a gray level co-occurrence matrix based on the pre-processed remote sensing image data; Performing band extraction based on the pre-processed remote sensing image data to obtain each of the index bands; Extracting a band index based on the preprocessed remote sensing image data to obtain the radar band index and the spectral band index; Perform feature extraction according to the gray-level co-occurrence matrix to obtain texture features; The texture feature image is generated according to the texture features.

Images