Saline-alkali soil improvement data analysis method and system based on remote sensing data

Abstract

The invention provides a saline-alkali soil improvement data analysis method and system based on remote sensing data, and relates to the technical field of land scientificity, and the method comprises the steps: collecting ground monitoring data and remote sensing data, and generating a comprehensive feature vector and a spectrum feature value; then, outputting a saline-alkali soil improvement grade by using a convolutional neural network model; a correction coefficient is generated through the soil salinization index, the spectral index and the pH value, an upgrading and downgrading mechanism is set in combination with a correction threshold interval, and the improvement grade is dynamically adjusted to make up for the model error and improve the evaluation accuracy. According to the method, the saline-alkali soil quantitative analysis model based on the convolutional neural network is constructed, and the ground monitoring data and the remote sensing spectrum characteristic data are combined, so that the improvement grade of the saline-alkali soil is scientifically evaluated.

Description

Technical Field

[0001] The present invention relates to the field of land science and technology, and in particular to a saline-alkali land improvement data analysis method and system based on remote sensing data. Background Art

[0002] The global saline-alkali land area exceeds 1 billion hectares, mainly distributed in arid and semi-arid areas. Soil salinization has become one of the important factors restricting agricultural development. my country's saline-alkali land is mainly distributed in the Huanghuaihai Plain, the Northeast Plain, the Northwest Arid Region, etc., with a total area of ​​about 150 million hectares. Therefore, saline-alkali land improvement is an important issue for land ecological restoration and agricultural sustainable development. Especially in my country and around the world, saline-alkali land is widely distributed and affects food security, ecosystem stability and social and economic development. In recent years, with the development of information technology tools such as remote sensing technology, artificial intelligence, and big data technology, the monitoring, evaluation and optimization methods of saline-alkali land improvement have been significantly improved.

[0003] Insufficient existing technology:

[0004] Among the currently known existing technologies, traditional saline-alkali land improvement mainly relies on ground surveys and soil sampling, which is costly, inefficient, and lacks spatial continuity. It is often limited to spatial coverage and data timeliness, and cannot comprehensively and dynamically monitor the changes in saline-alkali land. In addition, the analysis accuracy is low, and there is a lack of multi-source data integration capabilities. It is difficult to adapt to large-scale and long-term improvement needs, and it is costly and has poor scalability.

[0005] Traditional methods mainly rely on a single data source, and it is difficult for traditional methods to deal with the complex nonlinear relationship between soil salinization and multiple factors such as precipitation, irrigation, and crop planting, and it is impossible to provide a specific standard for judging the grade of improved saline-alkali land.

[0006] Therefore, it is necessary to provide a saline-alkali land improvement data analysis method and system based on remote sensing data to solve the above problems.

[0007] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0008] The object of the present invention is to provide a saline-alkali land improvement data analysis method and system based on remote sensing data to solve the problems raised in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] A method for analyzing saline-alkali land improvement data based on remote sensing data, the specific steps comprising:

[0011] Step 1: Collect improved ground monitoring data of saline-alkali land, including soil salinity, surface reflectivity, vegetation coverage, soil moisture content and pH value;

[0012] Step 2: Obtain remote sensing data of the improved saline-alkali land, extract spectral characteristic indexes in the remote sensing data, including salt index, normalized difference salinity index and soil-adjusted vegetation index, and use the three indexes to comprehensively generate remote sensing spectral characteristic values ​​of saline-alkali land;

[0013] Step 3: Combine soil salt content, surface reflectivity, vegetation coverage, and soil moisture content to generate soil salinization index and saline-alkali land spectral index. Use the generated soil salinization index and saline-alkali land spectral index and pH value to comprehensively generate the characteristic vector of saline-alkali land.

[0014] Step 4: Construct a saline-alkali land quantitative analysis model, using a convolutional neural network model structure, including an input layer, a hidden layer, and an output layer, wherein the input layer is used to receive the characteristic vector and remote sensing spectrum characteristic value of the saline-alkali land, the hidden layer is used to perform data processing on the characteristic vector, and the output layer is used to output the saline-alkali land improvement grade index, which includes three improvement grade indicators: excellent, medium, and general;

[0015] Step 5: Obtain the characteristic vector and remote sensing spectral characteristic value of the saline-alkali land to be analyzed, input them into the trained and optimized saline-alkali land quantitative analysis model, output them to obtain the improvement grade index of the saline-alkali land to be evaluated, and obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate the grade correction coefficient;

[0016] Step 6: Establish an upgrade and downgrade mechanism to compare the grade correction coefficient of the saline-alkali land to be evaluated with the preset correction threshold range, and revise the improvement grade index according to the comparison result to determine the final index of the saline-alkali land to be evaluated.

[0017] Furthermore, the remote sensing spectral characteristic values ​​of saline-alkali land are comprehensively generated based on the following formula:

[0018] The salinity index is calculated based on the reflectivity of the red light band and the reflectivity of the blue light band in the remote sensing data. The formula is:

[0019]

[0020] Among them, SI represents the salinity index, B4 and B2 are the reflectance of the red light band and the blue light band in the remote sensing data respectively;

[0021] According to the reflectivity of the red light band and the reflectivity of the blue light band in the remote sensing data, the normalized difference salinity index is calculated based on the following formula:

[0022]

[0023] Among them, NDSI stands for Normalized Difference Salinity Index;

[0024] The soil vegetation index is calculated based on the red light band reflectance, near infrared band reflectance and soil adjustment factor in remote sensing data. The formula is:

[0025]

[0026] Among them, SAVI represents the soil vegetation index, B5 is the reflectivity of the near-infrared band in the remote sensing data, and L is the soil adjustment factor;

[0027] The three indices are used to comprehensively generate the remote sensing spectral characteristic values ​​of saline-alkali land, and the formula is as follows:

[0028] TZ=ω1*SI+ω2*NDSI+ω3*SAVI

[0029] Among them, TZ represents the remote sensing spectral characteristic value of saline-alkali land, ω1, ω2, and ω3 represent the proportional weights of the corresponding indexes, and ω1>ω2>ω3>0.

[0030] Furthermore, the soil salinization index and saline-alkali land spectral index are generated based on the following method:

[0031] The soil salt content is obtained by measuring the conductivity of the soil solution with a conductivity meter. The corresponding soil salt content is obtained according to the measured conductivity. The soil moisture content is directly measured by a soil moisture sensor. The soil salinization index is generated by combining the soil salt content and the soil moisture content. The formula is:

[0032]

[0033] Among them, SSI represents soil salinization index, YF represents soil salt content, and HS represents soil water content;

[0034] The surface reflectance is obtained by performing radiation correction and atmospheric correction on remote sensing data. The digital values ​​collected by satellite sensors are converted into radiation brightness through radiation calibration, and then the atmospheric top reflectance is calculated through normalization of solar irradiance. The atmospheric effect is deducted by the atmospheric correction method to finally obtain the surface reflectance. The vegetation coverage is obtained by extracting the vegetation coefficient from the remote sensing data after radiation correction and atmospheric correction, and then combining the pixel binary model to decompose the pixels into two components: complete vegetation coverage and complete bare soil. The vegetation coverage is obtained according to the range difference, and the saline-alkali land spectral index is generated by combining the surface reflectance and vegetation coverage. The formula is as follows:

[0035]

[0036] Among them, GP represents the saline-alkali land spectral index, FS is the surface reflectance, and FG is the vegetation coverage.

[0037] Furthermore, a quantitative analysis model for saline-alkali land is constructed based on the following method:

[0038] The generated soil salinization index and saline-alkali land spectral index are combined with the improved saline-alkali land pH value to generate the saline-alkali land feature vector. The combined feature vector and the saline-alkali land remote sensing spectral feature value are used as the input of the model. The saline-alkali land improvement grade index is used as the label to train the saline-alkali land quantitative analysis model.

[0039] The structure of the saline-alkali land quantitative analysis model adopts a neural network convolution structure, including an input layer, a hidden layer and an output layer. The input layer is responsible for receiving the processed feature data and feature vectors; the hidden layer is used to extract local and high-level features in the input feature data. By applying multiple convolution kernels, the hidden layer can recognize complex feature patterns. By using the ReLu activation function, nonlinear relationships are introduced so that the model can fit complex feature relationships; an independent neuron is set in the output layer, which is responsible for converting the feature data extracted by the hidden layer into the final prediction result, that is, outputting the saline-alkali land improvement grade index.

[0040] Furthermore, the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed are obtained and integrated to generate the grade correction coefficient, and the formula based on it is:

[0041]

[0042] Among them, X z Indicates the level correction index, PH best is the optimal pH value of saline-alkali land, PH is the pH value of the saline-alkali land to be analyzed, GP now 、SSI now They are respectively the soil salinization index and the saline-alkali land spectral index of the saline-alkali land to be analyzed.

[0043] Furthermore, a promotion and demotion mechanism is established to compare the improvement grade index of the saline-alkali land to be evaluated with the preset correction threshold interval to determine whether to promote or demote the improvement grade index of the saline-alkali land to be evaluated. The method is as follows:

[0044] Establish improvement grade indicators including excellent, medium and general improvement grade indicators and correction threshold interval [X min , X max ], the level correction factor X z and the modified threshold interval [X min , X max ]Compare, if X min ≤Xz ≤X max , then the improvement grade index of the saline-alkali land to be evaluated output by the analysis model is directly used as the final index; if X z >X max , then the improvement grade index of the saline-alkali land to be evaluated output by the analysis model is taken as the final index after being raised one level; if X z <X min , the improvement grade index of the saline-alkali land to be evaluated output by the analysis model will be reduced by one level as the final index; when the improvement grade index of the saline-alkali land output by the model happens to be of excellent grade, it is limited that no upgrading operation is required; when the improvement grade index of the saline-alkali land output by the model happens to be of general grade, it is limited that no downgrading operation is required.

[0045] The present invention further provides a saline-alkali land improvement data analysis system based on remote sensing data, the analysis system is used to execute the above-mentioned saline-alkali land improvement data analysis method based on remote sensing data, comprising:

[0046] A ground monitoring data acquisition module, which is used to collect improved saline-alkali land ground monitoring data, including soil salt content, surface reflectivity, vegetation coverage, soil moisture content and pH value;

[0047] A saline-alkali land remote sensing spectral characteristic value generation module, which is used to obtain remote sensing data of the improved saline-alkali land ground, extract spectral characteristic indexes in the remote sensing data, including saline-alkali index, normalized difference saline-alkali index and soil-adjusted vegetation index, and use the three indexes to comprehensively generate saline-alkali land remote sensing spectral characteristic values;

[0048] An index generation module, wherein the index generation module is used to generate a soil salinization index and a saline-alkali land spectral index by combining soil salt content, surface reflectivity, vegetation coverage, and soil moisture content, and to generate a characteristic vector of the saline-alkali land by combining the generated soil salinization index and saline-alkali land spectral index with a pH value;

[0049] A model building module, the model building module is used to build a saline-alkali land quantitative analysis model, adopts a convolutional neural network model structure, and includes an input layer, a hidden layer, and an output layer, wherein the input layer is used to receive the characteristic vector and remote sensing spectrum characteristic value of the saline-alkali land, the hidden layer is used to perform data processing on the characteristic vector, and the output layer is used to output the saline-alkali land improvement grade index, the improvement grade index includes three improvement grade indexes of excellent, medium, and general;

[0050] A saline-alkali land grade assessment module, which is used to obtain the characteristic vector and remote sensing spectral characteristic value of the saline-alkali land to be analyzed, input them into the trained and optimized saline-alkali land quantitative analysis model, output them to obtain the improvement grade index of the saline-alkali land to be evaluated, and obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate a grade correction coefficient;

[0051] The grade correction and upgrading module is used to establish an upgrading and upgrading mechanism, compare the grade correction coefficient of the saline-alkali land to be evaluated with a preset correction threshold range, and correct the improvement grade index according to the comparison result to determine the final index of the saline-alkali land to be evaluated.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] By combining ground monitoring data with remote sensing spectral data, the present invention can comprehensively and dynamically evaluate the improvement effect of saline-alkali land, overcome the shortcomings of single data source and lack of integration in traditional methods, and adopt multi-dimensional parameters such as soil salinization index, spectral index and comprehensive characteristic vector to ensure the evaluation results. The generation module of remote sensing spectral characteristic values ​​and real-time dynamic collection of ground monitoring data can effectively realize large-scale monitoring, avoid the limitations of traditional ground sampling methods that are limited to a small range, and realize dynamic monitoring of saline-alkali land change trends through superposition analysis of multi-phase data, providing a decision-making basis for the optimization of long-term improvement measures;

[0054] A convolutional neural network is used to construct a quantitative analysis model for saline-alkali land. Through deep learning technology to process multi-source heterogeneous data, the complex nonlinear relationships in the data can be efficiently extracted, which greatly improves the efficiency of data processing and result analysis. The improvement grade index output by the model directly quantifies the improvement effect, which is convenient for further zoning management. The ground data such as soil salt content, vegetation coverage, soil moisture content, pH value, etc. are combined with the salinity index, normalized difference salinity index and soil adjusted vegetation index generated by remote sensing, breaking through the dependence of traditional technology on a single data source. The data fusion method provides more comprehensive information support for saline-alkali land assessment, making the analysis results have spatial continuity and ground representativeness of saline-alkali land.

[0055] The present invention constructs a comprehensive, intelligent and dynamic saline-alkali land improvement effect evaluation system through multi-source data fusion, remote sensing spectral technology, artificial intelligence model and dynamic correction mechanism, and realizes the accuracy, flexibility and efficiency of improvement data evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the overall method flow of the present invention.

[0057] Figure 2It is a schematic diagram of the system module flow of the present invention. DETAILED DESCRIPTION

[0058] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0059] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0060] Example:

[0061] See also Figure 1 , a saline-alkali land improvement data analysis method based on remote sensing data, the specific steps include:

[0062] Step 1: Collect improved ground monitoring data of saline-alkali land, including soil salinity, surface reflectivity, vegetation coverage, soil moisture content and pH value;

[0063] Step 2: Obtain remote sensing data of the improved saline-alkali land, extract spectral characteristic indexes in the remote sensing data, including salt index, normalized difference salinity index and soil-adjusted vegetation index, and use the three indexes to comprehensively generate remote sensing spectral characteristic values ​​of saline-alkali land;

[0064] Step 3: Combine soil salt content, surface reflectivity, vegetation coverage, and soil moisture content to generate soil salinization index and saline-alkali land spectral index. Use the generated soil salinization index and saline-alkali land spectral index and pH value to comprehensively generate the characteristic vector of saline-alkali land.

[0065] Step 4: Construct a saline-alkali land quantitative analysis model, using a convolutional neural network model structure, including an input layer, a hidden layer, and an output layer, wherein the input layer is used to receive the characteristic vector and remote sensing spectrum characteristic value of the saline-alkali land, the hidden layer is used to perform data processing on the characteristic vector, and the output layer is used to output the saline-alkali land improvement grade index, which includes three improvement grade indicators: excellent, medium, and general;

[0066] Step 5: Obtain the characteristic vector and remote sensing spectral characteristic value of the saline-alkali land to be analyzed, input them into the trained and optimized saline-alkali land quantitative analysis model, output them to obtain the improvement grade index of the saline-alkali land to be evaluated, and obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate the grade correction coefficient;

[0067] Step 6: Establish an upgrade and downgrade mechanism to compare the grade correction coefficient of the saline-alkali land to be evaluated with the preset correction threshold range, and revise the improvement grade index according to the comparison result to determine the final index of the saline-alkali land to be evaluated.

[0068] It should be noted that by combining the multiple key features of remote sensing monitoring of saline-alkali land, the advantages of different bands and vegetation indices are organically integrated, providing an accurate and scientific basis for the evaluation of the improvement effect of saline-alkali land. By introducing the saline-alkali index, normalized difference saline-alkali index and soil vegetation index, the characteristics of saline-alkali land are characterized from multiple dimensions such as soil salinity, salinity distribution differences and vegetation coverage status, making up for the shortcomings of single indicator monitoring. The setting of weights when comprehensively generating remote sensing spectral characteristic values ​​ensures that the contribution of the three indexes to the overall evaluation is in line with their importance ranking, highlighting the leading role of key indicators, and further improving the scientificity and practicality of the evaluation results.

[0069] Therefore, it is necessary to comprehensively generate the remote sensing spectral characteristic values ​​of saline-alkali land, based on the following formula:

[0070] The salinity index is calculated based on the reflectivity of the red light band and the reflectivity of the blue light band in the remote sensing data. The formula is:

[0071]

[0072] Among them, SI represents the salinity index, B4 and B2 are the reflectance of the red light band and the reflectance of the blue light band in the remote sensing data respectively; in the above formula, the red light band and the blue light band are highly sensitive to the spectral reflectance characteristics of soil and vegetation. In saline-alkali land, due to the significant influence of salt and alkali on the reflectance characteristics of the surface, the red light band often shows a higher reflectance, while the reflectance of the blue light band is relatively low. In saline-alkali land, due to the accumulation of salt on the surface, the reflectance of the red light band is usually higher than that of the blue light band. Therefore, the higher the degree of salinity, the greater the value of SI. This relationship can better describe the relationship between the degree of soil salinization and the reflectance of the band, which is intuitive and scientific.

[0073] The normalized difference salinity index is calculated based on the red light band reflectance and blue light band reflectance in the remote sensing data. The formula is:

[0074]

[0075] Among them, NDSI stands for Normalized Difference Salinity Index; in the above formula, B4-B2 is used to describe the difference in band reflectance, which can accurately reflect the characteristics of salinization. The introduction of B4+B2 plays a role in normalization and proportional adjustment, avoiding the deviation caused by the overall band reflectance when the difference formula is used directly, such as the salt-alkali ground with high surface brightness;

[0076] The soil vegetation index is calculated based on the red light band reflectance, near infrared band reflectance and soil adjustment factor in remote sensing data. The formula is:

[0077]

[0078] Among them, SAVI represents the soil vegetation index, B5 is the reflectivity of the near-infrared band in the remote sensing data, and L is the soil adjustment factor; in the above formula, (B5-B4)*(1+L) emphasizes the contrast between the high reflectivity of the near-infrared band and the low reflectivity of the red band, highlighting the spectral characteristics of vegetation. The addition of L in the denominator of (B5+B4+L) effectively suppresses the influence of the soil background on the total reflectivity. This normalization method can also reduce the difference in band reflectivity caused by factors such as lighting conditions and imaging angles; the soil adjustment factor can be measured by laboratory analysis of the reflectivity of different soil samples to obtain the reflectivity data of each soil type. Based on these data, the soil adjustment factor suitable for the region can be calculated, and the soil adjustment factor value range is generally [0.1,0.5].

[0079] The three indices are used to comprehensively generate the remote sensing spectral characteristic values ​​of saline-alkali land, and the formula is as follows:

[0080] TZ=ω1*SI+ω2*NDSI+ω3*SAVI

[0081] Among them, TZ represents the remote sensing spectral characteristic value of saline-alkali land, ω1, ω2, and ω3 represent the proportional weights of the corresponding indexes, and ω1>ω2>ω3>0; in the above formula, the reason why the proportional weight is set to ω1>ω2>ω3>0 is that SI is an index specially used to extract surface salt information, which directly reflects the salt content and distribution in the soil, and the spectral characteristics of saline-alkali land are mainly closely related to the surface salt content, and SI plays a leading role in distinguishing saline-alkali land from other land types such as ordinary soil and vegetation-covered saline-alkali land. Therefore, the highest proportional weight ω1 should be given. NDSI eliminates some influencing factors such as lighting conditions and sensor angles to make the extraction of salt characteristics more stable, but its direct sensitivity to salt concentration is not as good as SI. Although NDSI has a significant auxiliary extraction effect on salt information, it cannot completely replace SI, so it is given the second weight ratio ω2; SAVI mainly reflects vegetation coverage and provides vegetation information. In saline-alkali land with vegetation patches or mixed surface features, it helps to distinguish saline-alkali land from low-coverage vegetation or bare soil. However, since vegetation characteristics are relatively unimportant in saline-alkali land extraction, its weight ω3 needs to be set to the minimum.

[0082] It should be noted that soil salinization is a common soil degradation phenomenon worldwide, especially in arid and semi-arid areas. Salinization will lead to a decrease in soil fertility and limited plant growth, thus affecting agricultural production and ecosystem stability. Monitoring and evaluation of soil salinization is a key step in preventing and controlling salinization. The salinization index, as a quantitative indicator of the degree of salinization, can provide an important basis for soil health assessment and agricultural management. The calculation of the saline-alkali land spectral index is because the spectral index can quickly identify the distribution characteristics of large areas of saline-alkali land, providing a basis for land use planning on saline-alkali land. In saline-alkali land with lack of data or complex ground conditions, such as remote deserts and grasslands, the spectral index can make up for the lack of ground monitoring.

[0083] Therefore, it is necessary to generate soil salinization index and saline-alkali land spectral index, based on the following method:

[0084] The soil salt content is obtained by measuring the conductivity of the soil solution with a conductivity meter. The corresponding soil salt content is obtained according to the measured conductivity. The soil moisture content is directly measured by a soil moisture sensor. The soil salinization index is generated by combining the soil salt content and the soil moisture content. The formula is:

[0085]

[0086] Among them, SSI represents soil salinization index, YF represents soil salt content, and HS represents soil water content. In the above formula, when YF increases, SSI will also increase, which means that higher salt content will lead to more severe soil salinization, which may cause greater inhibition on crop growth. When HS increases, SSI will decrease, which means that higher water content will dilute the salt concentration in the soil, thereby reducing the risk of salinization. Therefore, the smaller the SSI value, the better, indicating that the improvement effect of saline-alkali land is good.

[0087] The surface reflectance is obtained by performing radiation correction and atmospheric correction on remote sensing data. The digital values ​​collected by satellite sensors are converted into radiation brightness through radiation calibration, and then the atmospheric top reflectance is calculated through normalization of solar irradiance. The atmospheric effect is deducted by the atmospheric correction method to finally obtain the surface reflectance. The vegetation coverage is obtained by extracting the vegetation coefficient from the remote sensing data after radiation correction and atmospheric correction, and then combining the pixel binary model to decompose the pixels into two components: complete vegetation coverage and complete bare soil. The vegetation coverage is obtained according to the range difference, and the saline-alkali land spectral index is generated by combining the surface reflectance and vegetation coverage. The formula is as follows:

[0088]

[0089] Among them, GP represents the spectral index of saline-alkali land, FS is the surface reflectance, and FG is the vegetation coverage. In the above formula, as FS increases, GP also increases. Soil with a higher degree of salinization often shows a higher reflectivity, because salt crystals or salt crusts will significantly enhance the reflection of light on the surface, indicating that the saline-alkali land has a higher degree of salinization. When FG increases, GP will decrease. Due to the increase in vegetation coverage, it means that vegetation gradually covers the exposed saline-alkali land on the surface. Vegetation coverage can block the salt crystals or salt crusts on the soil surface, thereby weakening the spectral reflectance characteristics of the saline-alkali land. Therefore, the smaller the GP value, the better. The smaller the GP value, the higher the vegetation coverage of the saline-alkali land and the lower the degree of salinization.

[0090] It should be noted that the importance of the saline-alkali land quantitative analysis model lies in its ability to construct a comprehensive feature vector by combining multi-source feature data, such as salinization index, spectral index and pH value after soil improvement, and accurately evaluate the improvement level of saline-alkali land through the deep learning ability of neural networks. The core of the design of this model is that the input layer extracts multi-dimensional data, and the hidden layer extracts local and high-level features through convolution kernels and introduces nonlinear relationships, thereby effectively capturing complex feature patterns. The final output layer converts data into usable saline-alkali land improvement level indicators. This process can provide strong support for the scientific management of land resources and ecological governance.

[0091] Therefore, it is necessary to construct a quantitative analysis model for saline-alkali land, based on the following method:

[0092] The structure of the saline-alkali land quantitative analysis model adopts a neural network convolution structure, including an input layer, a hidden layer and an output layer. The input layer is responsible for receiving the processed saline-alkali land remote sensing spectral feature values ​​and feature vectors; the hidden layer is used to extract local and high-level features in the input feature data. By applying multiple convolution kernels, the hidden layer can identify complex feature patterns. By using the ReLu activation function, nonlinear relationships are introduced so that the model can fit complex feature relationships; an independent neuron is set in the output layer, which is responsible for converting the feature data extracted by the hidden layer into the final prediction result, that is, outputting the saline-alkali land improvement grade index;

[0093] The relevant characteristic data of previously improved saline-alkali land are obtained, including the characteristic values ​​and characteristic vectors of the remote sensing spectrum of the saline-alkali land, and the grade of the improved saline-alkali land is evaluated by experts and divided into three grades: excellent, medium and general. The characteristic values ​​and characteristic vectors of the remote sensing spectrum of the saline-alkali land are used as the input of the model, and the saline-alkali land improvement grade index is used as the label to train the model. During the training process, the mean square error function is selected as the loss function, and the loss function value is calculated according to the output result and the true label. The gradient is calculated through the back propagation algorithm, and the weights and biases of the neural network are updated. The above operations are repeated until the model reaches the predetermined number of training rounds.

[0094] It should be noted that the importance of calculating the grade correction index lies in that it provides a scientific basis for dynamically adjusting the improvement grade of saline-alkali land by comprehensively considering key factors such as soil salinization index, spectral index and pH value. Specifically, pH value is an important soil property in the process of saline-alkali land improvement, which directly affects soil fertility and plant growth. The soil salinization index and spectral index reflect the degree of salinization and overall condition of the soil. Therefore, the combination of these data can more accurately evaluate the improvement potential and current status of saline-alkali land.

[0095] Therefore, it is necessary to obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate the grade correction coefficient based on the formula:

[0096]

[0097] Among them, X z Indicates the level correction index, PH best is the optimal pH value of saline-alkali land, PH is the pH value of the saline-alkali land to be analyzed, GP now 、SSI now are the soil salinization index and saline-alkali land spectral index of the saline-alkali land to be analyzed respectively; in the above formula, |PH best-The larger the pH value, the smaller the soil improvement degree is and the further the soil is from the optimal improvement condition. This reflects that the further the current soil pH is from the suitable range, the higher the difficulty of improvement. now The larger the X z The smaller the SSI now A larger value of X means that the soil salinization phenomenon is more serious and the improvement potential is lower. When the salinization index is close to zero, it means that the soil salt level is ideal. z The inhibitory effect of GP now The larger the X z The smaller, the GP now Larger values ​​of mean that the soil is more exposed and less covered by plants, reflecting a lower degree of soil improvement.

[0098] It should be noted that the establishment of an upgrade and downgrade mechanism and the dynamic adjustment of saline-alkali land improvement indicators based on the correction threshold interval can effectively compensate for the possible errors of a single analysis model and make the evaluation results closer to the actual situation. This mechanism ensures the accuracy and efficiency of resource allocation through scientific classification of improvement potential and difficulty, and provides a dynamic and flexible decision-making basis for land improvement, which helps to maximize the value of land use and optimize the input-output ratio, thereby promoting sustainable agricultural development and ecological environmental protection.

[0099] Therefore, it is necessary to establish an upgrade and downgrade mechanism to compare the improvement grade index of the saline-alkali land to be evaluated with the preset correction threshold interval to determine whether to upgrade or downgrade the improvement grade index of the saline-alkali land to be evaluated. The method is based on:

[0100] Establish improvement grade indicators including excellent, medium and general improvement grade indicators and correction threshold interval [X min , X max ], the level correction factor X z and the modified threshold interval [X min , X max ]Compare, if X min ≤X z ≤X max , then the improvement grade index of the saline-alkali land to be evaluated output by the analysis model is directly used as the final index; if X z >X max , then the improvement grade index of the saline-alkali land to be evaluated output by the analysis model is taken as the final index after being raised one level; if X z <X min , the improvement grade index of the saline-alkali land to be evaluated output by the analysis model will be reduced by one level as the final index; when the improvement grade index of the saline-alkali land output by the model happens to be of excellent grade, it is limited that no upgrading operation is required; when the improvement grade index of the saline-alkali land output by the model happens to be of general grade, it is limited that no downgrading operation is required.

[0101] The present invention also provides a saline-alkali land improvement data analysis system based on remote sensing data. The analysis system is used for the above-mentioned saline-alkali land improvement data analysis method based on remote sensing data, comprising:

[0102] A ground monitoring data acquisition module, which is used to collect improved saline-alkali land ground monitoring data, including soil salt content, surface reflectivity, vegetation coverage, soil moisture content and pH value;

[0103] A saline-alkali land remote sensing spectral characteristic value generation module, which is used to obtain remote sensing data of the improved saline-alkali land ground, extract spectral characteristic indexes in the remote sensing data, including saline-alkali index, normalized difference saline-alkali index and soil-adjusted vegetation index, and use the three indexes to comprehensively generate saline-alkali land remote sensing spectral characteristic values;

[0104] An index generation module, wherein the index generation module is used to generate a soil salinization index and a saline-alkali land spectral index by combining soil salt content, surface reflectivity, vegetation coverage, and soil moisture content, and to generate a characteristic vector of the saline-alkali land by combining the generated soil salinization index and saline-alkali land spectral index with a pH value;

[0105] A model building module, the model building module is used to build a saline-alkali land quantitative analysis model, adopts a convolutional neural network model structure, and includes an input layer, a hidden layer, and an output layer, wherein the input layer is used to receive the characteristic vector and remote sensing spectrum characteristic value of the saline-alkali land, the hidden layer is used to perform data processing on the characteristic vector, and the output layer is used to output the saline-alkali land improvement grade index, the improvement grade index includes three improvement grade indexes of excellent, medium, and general;

[0106] A saline-alkali land grade assessment module, which is used to obtain the characteristic vector and remote sensing spectral characteristic value of the saline-alkali land to be analyzed, input them into the trained and optimized saline-alkali land quantitative analysis model, output them to obtain the improvement grade index of the saline-alkali land to be evaluated, and obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate a grade correction coefficient;

[0107] The grade correction and upgrading module is used to establish an upgrading and upgrading mechanism, compare the grade correction coefficient of the saline-alkali land to be evaluated with a preset correction threshold range, and correct the improvement grade index according to the comparison result to determine the final index of the saline-alkali land to be evaluated.

[0108] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0109] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for analyzing saline-alkali land improvement data based on remote sensing data, characterized in that: The specific steps include: Step 1: Collect improved ground monitoring data of saline-alkali land, including soil salinity, surface reflectivity, vegetation coverage, soil moisture content and pH value; Step 2: Obtain remote sensing data of the improved saline-alkali land, extract spectral characteristic indexes in the remote sensing data, including salt index, normalized difference salinity index and soil-adjusted vegetation index, and use the three indexes to comprehensively generate remote sensing spectral characteristic values ​​of saline-alkali land; Step 3: Combine soil salt content, surface reflectivity, vegetation coverage, and soil moisture content to generate soil salinization index and saline-alkali land spectral index. Use the generated soil salinization index and saline-alkali land spectral index and pH value to comprehensively generate the characteristic vector of saline-alkali land. Step 4: Construct a saline-alkali land quantitative analysis model, using a convolutional neural network model structure, including an input layer, a hidden layer, and an output layer, wherein the input layer is used to receive the characteristic vector and remote sensing spectrum characteristic value of the saline-alkali land, the hidden layer is used to perform data processing on the characteristic vector, and the output layer is used to output the saline-alkali land improvement grade index, which includes three improvement grade indicators: excellent, medium, and general; Step 5: Obtain the characteristic vector and remote sensing spectral characteristic value of the saline-alkali land to be analyzed, input them into the trained and optimized saline-alkali land quantitative analysis model, output them to obtain the improvement grade index of the saline-alkali land to be evaluated, and obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate the grade correction coefficient; Step 6: Establish an upgrade and downgrade mechanism to compare the grade correction coefficient of the saline-alkali land to be evaluated with the preset correction threshold range, and revise the improvement grade index according to the comparison result to determine the final index of the saline-alkali land to be evaluated.

2. The method for analyzing saline-alkali land improvement data based on remote sensing data according to claim 1, characterized in that: The formula for comprehensively generating the remote sensing spectral characteristic values ​​of saline-alkali land is as follows: The salinity index is calculated based on the reflectivity of the red light band and the reflectivity of the blue light band in the remote sensing data. The formula is: Among them, SI represents the salinity index, B4 and B2 are the reflectance of the red light band and the blue light band in the remote sensing data respectively; According to the reflectivity of the red light band and the reflectivity of the blue light band in the remote sensing data, the normalized difference salinity index is calculated based on the following formula: Among them, NDSI stands for Normalized Difference Salinity Index; The soil vegetation index is calculated based on the red light band reflectance, near infrared band reflectance and soil adjustment factor in remote sensing data. The formula is: Among them, SAVI represents the soil vegetation index, B5 is the reflectivity of the near-infrared band in the remote sensing data, and L is the soil adjustment factor; The three indices are used to comprehensively generate the remote sensing spectral characteristic values ​​of saline-alkali land, and the formula is as follows: TZ=ω1*SI+ω2*NDSI+ω3*SAVI Among them, TZ represents the remote sensing spectral characteristic value of saline-alkali land, ω1, ω2, and ω3 represent the proportional weights of the corresponding indexes, and ω1>ω2>ω3>0.

3. The method for analyzing saline-alkali land improvement data based on remote sensing data according to claim 1, characterized in that: The method for generating soil salinization index and saline-alkali land spectral index is based on: The soil salt content is obtained by measuring the conductivity of the soil solution with a conductivity meter. The corresponding soil salt content is obtained according to the measured conductivity. The soil moisture content is directly measured by a soil moisture sensor. The soil salinization index is generated by combining the soil salt content and the soil moisture content. The formula is: Among them, SSI represents soil salinization index, YF represents soil salt content, and HS represents soil water content; The surface reflectance is obtained by performing radiation correction and atmospheric correction on remote sensing data. The digital values ​​collected by satellite sensors are converted into radiation brightness through radiation calibration, and then the atmospheric top reflectance is calculated through normalization of solar irradiance. The atmospheric effect is deducted by the atmospheric correction method to finally obtain the surface reflectance. The vegetation coverage is obtained by extracting the vegetation coefficient from the remote sensing data after radiation correction and atmospheric correction, and then combining the pixel binary model to decompose the pixels into two components: complete vegetation coverage and complete bare soil. The vegetation coverage is obtained according to the range difference, and the saline-alkali land spectral index is generated by combining the surface reflectance and vegetation coverage. The formula is as follows: Among them, GP represents the saline-alkali land spectral index, FS is the surface reflectance, and FG is the vegetation coverage.

4. The method for analyzing saline-alkali land improvement data based on remote sensing data according to claim 1, characterized in that: The quantitative analysis model of saline-alkali land is constructed based on the following methods: The generated soil salinization index and saline-alkali land spectral index are combined with the improved saline-alkali land pH value to generate the saline-alkali land feature vector. The combined feature vector and the saline-alkali land remote sensing spectral feature value are used as the input of the model. The saline-alkali land improvement grade index is used as the label to train the saline-alkali land quantitative analysis model. The structure of the saline-alkali land quantitative analysis model adopts a neural network convolution structure, including an input layer, a hidden layer and an output layer. The input layer is responsible for receiving the processed feature vector and the remote sensing spectral feature values ​​of the saline-alkali land; the hidden layer is used to extract local and high-level features in the input feature data. By applying multiple convolution kernels, the hidden layer can recognize complex feature patterns. By using the ReLu activation function, nonlinear relationships are introduced so that the model can fit complex feature relationships; an independent neuron is set in the output layer, which is responsible for converting the feature data extracted by the hidden layer into the final prediction result, that is, outputting the saline-alkali land improvement grade index.

5. The method for analyzing saline-alkali land improvement data based on remote sensing data according to claim 1, characterized in that: The soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed are obtained and combined to generate the grade correction coefficient. The formula is as follows: Among them, X z Indicates the level correction index, PH best is the optimal pH value of saline-alkali land, PH is the pH value of the saline-alkali land to be analyzed, GP now 、SSI now They are respectively the soil salinization index and the saline-alkali land spectral index of the saline-alkali land to be analyzed.

6. The method for analyzing saline-alkali land improvement data based on remote sensing data according to claim 1, characterized in that: Establish an upgrade and downgrade mechanism to compare the improvement grade index of the saline-alkali land to be evaluated with the preset correction threshold interval to determine whether to upgrade or downgrade the improvement grade index of the saline-alkali land to be evaluated. The method is based on: Establish improvement grade indicators including excellent, medium and general improvement grade indicators and correction threshold interval [X min ,X max ], the level correction factor X z and the modified threshold interval [X min ,X max ]Compare, if X min ≤X z ≤X max , then the improvement grade index of the saline-alkali land to be evaluated output by the analysis model is directly used as the final index; if X z >X max , then the improvement grade index of the saline-alkali land to be evaluated output by the analysis model is taken as the final index after being raised one level; if X z <X min , the improvement grade index of the saline-alkali land to be evaluated output by the analysis model will be reduced by one level as the final index; when the improvement grade index of the saline-alkali land output by the model happens to be of excellent grade, it is limited that no upgrading operation is required; when the improvement grade index of the saline-alkali land output by the model happens to be of general grade, it is limited that no downgrading operation is required.

7. A saline-alkali land improvement data analysis system based on remote sensing data, characterized in that: The analysis system is used to execute the saline-alkali land improvement data analysis method based on remote sensing data according to any one of claims 1 to 6, comprising: A ground monitoring data acquisition module, which is used to collect improved saline-alkali land ground monitoring data, including soil salt content, surface reflectivity, vegetation coverage, soil moisture content and pH value; A saline-alkali land remote sensing spectral characteristic value generation module, which is used to obtain remote sensing data of the improved saline-alkali land ground, extract spectral characteristic indexes in the remote sensing data, including saline-alkali index, normalized difference saline-alkali index and soil-adjusted vegetation index, and use the three indexes to comprehensively generate saline-alkali land remote sensing spectral characteristic values; An index generation module, wherein the index generation module is used to generate a soil salinization index and a saline-alkali land spectral index by combining soil salt content, surface reflectivity, vegetation coverage, and soil moisture content, and to generate a characteristic vector of the saline-alkali land by combining the generated soil salinization index and saline-alkali land spectral index with a pH value; A model building module, the model building module is used to build a saline-alkali land quantitative analysis model, adopts a convolutional neural network model structure, and includes an input layer, a hidden layer, and an output layer, wherein the input layer is used to receive the characteristic vector and remote sensing spectrum characteristic value of the saline-alkali land, the hidden layer is used to perform data processing on the characteristic vector, and the output layer is used to output the saline-alkali land improvement grade index, the improvement grade index includes three improvement grade indexes of excellent, medium, and general; A saline-alkali land grade assessment module, which is used to obtain the characteristic vector and remote sensing spectral characteristic value of the saline-alkali land to be analyzed, input them into the trained and optimized saline-alkali land quantitative analysis model, output them to obtain the improvement grade index of the saline-alkali land to be evaluated, and obtain the soil salinization index, saline-alkali land spectral index and pH value of the saline-alkali land to be analyzed, and comprehensively generate a grade correction coefficient; The grade correction and upgrading module is used to establish an upgrading and upgrading mechanism, compare the grade correction coefficient of the saline-alkali land to be evaluated with a preset correction threshold range, and correct the improvement grade index according to the comparison result to determine the final index of the saline-alkali land to be evaluated.

Images