A method and apparatus for determining the deployment strategy of a soil moisture sensor network

Abstract

This invention discloses a method and apparatus for determining the deployment strategy of a soil moisture sensor network. It utilizes soil moisture data retrieved from microwave remote sensing data as reference data, and then obtains feature data for each pixel based on digital elevation data of the area to be deployed. Clustering is performed on the feature data of each pixel to obtain the location of representative sample points. A Gaussian process regression model is then used for modeling. The optimal deployment location and number of sensors are determined by combining the location of the representative sample points with the retrieved soil moisture data. The number and location of sensors selected using this method have high accuracy.

Description

Technical Field

[0001] This invention relates to the field of soil moisture sensor network deployment technology, and in particular to a method and apparatus for determining soil moisture sensor network deployment strategies. Background Technology

[0002] Soil moisture content is a key variable affecting surface and atmospheric water cycles and energy balance. Fields such as ecological drought, water resource planning, soil management, underground pollutant transport, and precision agriculture all rely on the spatial distribution information of soil moisture (i.e., soil moisture content). In recent years, advancements in Internet of Things (IoT) technology have driven the application of Wireless Sensor Networks (WSNs) in intelligent soil sensing. However, soil moisture exhibits high spatial heterogeneity, with its distribution significantly influenced by watershed-scale vegetation cover, soil texture differentiation, topography, and physicochemical properties. Therefore, it is crucial to rationally select the deployment locations of soil characteristic sensors to representatively observe and predict watershed-scale soil moisture, ensuring that the collected data accurately reflects the soil moisture characteristics of the entire region.

[0003] To optimize the deployment of soil moisture sensor networks, existing methods typically rely on collecting data from sample points within a specific time period and employing methods such as Affinity Propagation (AP) clustering to consider the similarities and differences in soil moisture data. By calculating the similarity between soil moisture data from different geographical locations, similar sample points are grouped into one category to reduce data redundancy and lower sensor investment costs. However, such methods depend on deploying a large number of collection nodes within the study area to acquire soil moisture data, which is not only labor-intensive but also has a long data collection cycle. Furthermore, while unsupervised classification methods (such as K-means clustering and hierarchical clustering) can identify soil areas with different characteristics, they cannot determine the optimal number of categories, and therefore, the optimal number of sensors to deploy.

[0004] Currently, microwave remote sensing technology calculates soil moisture using optical and thermal data. The most widely used model is the Thermal-Optical Trapezoid Model (TOTRAM), based on land surface temperature and vegetation indices. This model utilizes the trapezoidal distribution relationship between land surface temperature and vegetation indices to identify different soil moisture conditions, thereby enabling soil moisture estimation and retrieval. Microwave remote sensing technology has been successfully applied to estimate surface soil moisture and can use the soil moisture distribution information provided by microwave remote sensing data to verify the simulation effect of the proposed sensor deployment. However, existing methods for optimizing soil moisture sensor locations have the following drawbacks, failing to meet the requirements for efficient, high-quality, and representative deployment of field sensors:

[0005] (1) While clustering based on the similarity of soil moisture data collected from sample points (such as AP clustering) can reduce redundant sensor data and improve data utilization efficiency, it cannot fully consider complex geographical and soil properties, resulting in inaccurate sensor deployment in some areas. At the same time, the cost of acquiring soil moisture data is high, and the labor and time costs are also high, especially in remote areas at the watershed scale where soil moisture data is relatively scarce, and sensor points may not be able to provide enough samples for accurate cluster analysis.

[0006] (2) Unsupervised classification based on watershed-scale characteristics (such as slope, aspect, climate, solar radiation, soil texture, and vegetation cover type) cannot determine the appropriate number of clusters. If the number of clusters is not chosen properly, it may lead to over-clustering or under-clustering, affecting the validity of the analysis results, resulting in redundancy in the deployment of sites and reducing economic practicality. In addition, clustering results usually do not have clear labels to verify their correctness. Without a reference dataset, the classification results may only be based on the inherent patterns of the data and lack ground truth for verification. Summary of the Invention

[0007] To address the aforementioned technical problems, embodiments of the present invention provide a method and apparatus for determining a soil moisture sensor network deployment strategy, thereby resolving the technical problem in the prior art where the location and number of sensors within a watershed are not sufficiently accurate due to the inability to fully consider complex geographical and soil properties.

[0008] A first aspect of this invention provides a method for determining a soil moisture sensor network deployment strategy, the method comprising:

[0009] Remote sensing data of the area to be deployed is extracted to obtain the shortwave infrared transform reflectance and normalized vegetation index of the area to be deployed.

[0010] A pixel distribution map is constructed based on shortwave infrared transform reflectance and normalized vegetation index. Dry edges and wet edges are obtained based on the pixel distribution map. Soil moisture is inverted based on dry edges and wet edges to obtain the soil moisture content of the area to be deployed.

[0011] Collect digital elevation data of the area to be deployed, obtain feature data of each pixel based on digital elevation data, soil texture and vegetation cover type data, determine the initial number of cluster centers, obtain the initial number, and obtain the likelihood function under given model parameters based on the feature data of each pixel. The feature data includes latitude, longitude, slope, aspect, altitude, potential relative radiation, soil texture and vegetation cover type.

[0012] Based on the feature data of each pixel, the prior distribution under the given model parameters is obtained. The prior distribution is multiplied by the likelihood function to obtain the posterior distribution under the given model parameters. The posterior distribution is estimated using numerical methods to obtain the estimated value. The position of each cluster center is determined based on the estimated value. Based on the position of each cluster center, the nearest neighbor search is used to obtain the pixels whose distance to each cluster center is less than a preset value. These pixels are used as representative samples. The given model parameters include the mixing coefficient, mean, and covariance matrix.

[0013] A Gaussian process regression model is used to model the covariance model. The feature data of each representative sample point are input into the covariance model for prediction to obtain the predicted soil moisture content of the representative sample points. The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. If the RMSE is still greater than a preset threshold, the process moves to the step of determining the initial number of cluster centers. The initial number of cluster centers is increased to obtain an updated initial number. Based on the updated initial number, the positions of each updated cluster center and the new RMSE are calculated until the new RMSE is less than the preset threshold. The corresponding updated initial number is then determined as the final number of cluster centers. The final number of cluster centers is the number of sensors to be deployed, and the positions of the updated cluster centers are the locations of the sensors to be deployed.

[0014] In one possible implementation of the first aspect, dry and wet edges are obtained based on the pixel distribution map, and soil moisture is inverted based on the dry and wet edges to obtain the soil moisture content of the area to be deployed, including:

[0015] Based on shortwave infrared transform reflectance and normalized vegetation index, the pixel distribution relationship map is obtained through linear regression to obtain dry and wet edges;

[0016] Based on dry and wet edges, soil moisture is inverted using a thermo-optical trapezoidal model to obtain the soil moisture content of the area to be deployed.

[0017] In one possible implementation of the first aspect, the formula for calculating the slope is:

[0018]

[0019]

[0020]

[0021] In the formula, For pixels in Rate of change of elevation on the axis For pixels in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis;

[0022] The formula for calculating slope aspect is:

[0023]

[0024] In the formula, For pixels in Rate of change of elevation on the axis For pixels in Rate of change of elevation on the axis;

[0025] The formula for calculating potential relative radiation is:

[0026]

[0027] In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. The angle between the slope and the solar altitude angle, It is the angle between the slope direction and the solar azimuth.

[0028] In one possible implementation of the first aspect, the likelihood function under given model parameters is obtained based on the feature data of each pixel, including:

[0029] Based on the feature data of each pixel, the feature space range of each pixel is represented by a multivariate normal distribution, resulting in the multivariate normal distribution probability density function. The expression for the multivariate normal distribution probability density function is as follows:

[0030]

[0031] In the formula, For the expected value, Let covariance matrix be the variance matrix. For the feature data of each pixel;

[0032] The probability density of the Gaussian mixture model for each pixel is obtained based on the multivariate normal distribution probability density function of each pixel. The expression for the probability density of the Gaussian mixture model is as follows:

[0033]

[0034] In the formula, For the expected value, Let covariance matrix be the variance matrix. It is the mixing coefficient of the m-th Gaussian component. For the feature data of each pixel, Let be the mean of the m-th Gaussian component. Let m be the covariance matrix of the m-th Gaussian component. Let be the probability density function of the Gaussian mixture model for each pixel;

[0035] Based on the probability density of the Gaussian mixture model for each pixel, the likelihood function is obtained for the given model parameters, where the likelihood function is:

[0036]

[0037] In the formula, This indicates that the first... The probability of a pixel It is the first A multivariate normal distribution with Gaussian components It is the first The mixing coefficient of the Gaussian components, For the first The mean of the Gaussian components, For the first The covariance matrix of Gaussian components, The number of pixels. is the number of classes in the Gaussian mixture model.

[0038] In one possible implementation of the first aspect, the root mean square error of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated, including:

[0039] The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained through soil moisture inversion is calculated. The formula for calculating the RMSE is as follows:

[0040]

[0041] In the formula, To represent the number of sample points, To represent the predicted soil moisture content of the sample points, Soil moisture content obtained from soil moisture inversion.

[0042] To address the same technical problem, a second aspect of the present invention provides a soil moisture sensor network deployment strategy determination apparatus, the apparatus comprising:

[0043] The extraction module is used to extract remote sensing data of the area to be deployed, and obtain the shortwave infrared transform reflectance and normalized vegetation index of the area to be deployed.

[0044] The inversion module is used to construct a pixel distribution relationship map based on shortwave infrared transform reflectance and normalized vegetation index, obtain dry and wet edges based on the pixel distribution relationship map, and perform soil moisture inversion based on dry and wet edges to obtain the soil moisture content of the area to be deployed.

[0045] The first calculation module is used to collect digital elevation data of the area to be deployed, obtain feature data of each pixel based on digital elevation data and soil texture and vegetation cover type data, determine the initial number of cluster centers, obtain the initial number, and obtain the likelihood function under given model parameters based on the feature data of each pixel. The feature data includes latitude, longitude, slope, aspect, altitude, potential relative radiation, soil texture and vegetation cover type.

[0046] The second calculation module is used to obtain the prior distribution under given model parameters based on the feature data of each pixel, multiply the prior distribution and the likelihood function to obtain the posterior distribution under given model parameters, estimate the posterior distribution using numerical methods to obtain the estimated value, determine the position of each cluster center based on the estimated value, and use nearest neighbor search to obtain the pixels whose distance to each cluster center is less than a preset value based on the position of each cluster center, and use the pixels as representative samples. The given model parameters include mixing coefficients, mean and covariance matrix.

[0047] The third calculation module is used to model using a Gaussian process regression model to obtain a covariance model. The feature data of each representative sample point are input into the covariance model for prediction to obtain the predicted soil moisture content of the representative sample points. The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. If the RMSE is still greater than a preset threshold, the module proceeds to the step of determining the initial number of cluster centers. The initial number of cluster centers is increased to obtain an updated initial number. Based on the updated initial number, the positions of each updated cluster center and the new RMSE are calculated until the new RMSE is less than the preset threshold. Then, the corresponding updated initial number is determined as the final number of cluster centers. The final number of cluster centers is determined as the number of sensors to be deployed, and the positions of the updated cluster centers are the positions of the sensors to be deployed.

[0048] In one possible implementation of the second aspect, the inversion module includes a linear regression calculation unit and a soil moisture inversion unit, wherein,

[0049] The linear regression calculation unit is used to obtain the dry and wet edges by calculating the pixel distribution relationship map based on shortwave infrared transform reflectance and normalized vegetation index.

[0050] The soil moisture inversion unit is used to invert soil moisture based on dry and wet edges using a thermo-optical trapezoidal model to obtain the soil moisture content of the area to be deployed.

[0051] In one possible implementation of the second aspect, the formula for calculating the slope is:

[0052]

[0053]

[0054]

[0055] In the formula, For pixels in Rate of change of elevation on the axis For pixels in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis;

[0056] The formula for calculating slope aspect is:

[0057]

[0058] In the formula, For pixels in Rate of change of elevation on the axis For pixels in Rate of change of elevation on the axis;

[0059] The formula for calculating potential relative radiation is:

[0060]

[0061] In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. It is the angle between the slope and the solar altitude angle. It is the angle between the slope direction and the solar azimuth.

[0062] A third aspect of the present invention provides a computer device, comprising:

[0063] Memory, used to store computer programs;

[0064] A processor is used to implement the steps of the method for determining a soil moisture sensor network deployment strategy as described in the first aspect when executing the computer program.

[0065] A fourth aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for determining a soil moisture sensor network deployment strategy as described in the first aspect.

[0066] The technical solution of this invention has the following advantages:

[0067] The method for determining the deployment strategy of a soil moisture sensor network provided in this invention uses soil moisture data retrieved from microwave remote sensing data as reference data. Then, based on the digital elevation data of the area to be deployed, the feature data of each pixel is obtained. Clustering is performed based on the feature data of each pixel to obtain the location of representative sample points. A Gaussian process regression model is then used for modeling. The location of the obtained representative sample points and the retrieved soil moisture data are combined for evaluation and screening to determine the optimal deployment location and number of sensors. The number and location of sensors selected by the above method have high accuracy. Attached Figure Description

[0068] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0069] Figure 1 This is a flowchart of the method for determining the deployment strategy of a soil moisture sensor network in an embodiment of the present invention;

[0070] Figure 2 This is a pixel distribution diagram of the method for determining the soil moisture sensor network deployment strategy in this embodiment of the invention.

[0071] Figure 3 This is a flowchart illustrating the network deployment strategy determination method for the soil moisture sensor network deployment strategy in this embodiment of the invention.

[0072] Figure 4 This is a structural block diagram of the soil moisture sensor network deployment strategy determination device in an embodiment of the present invention. Detailed Implementation

[0073] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0074] In the description of this invention, it should be noted that the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0075] The method for determining the deployment strategy of a soil moisture sensor network provided in this embodiment of the invention, such as... Figure 1 As shown, Figure 1 The flowchart for determining the deployment strategy of a soil moisture sensor network includes steps S101 to S103, and the specific steps are as follows:

[0076] S101: Extract remote sensing data of the area to be deployed to obtain the shortwave infrared transform reflectance and normalized vegetation index of the area to be deployed.

[0077] In this embodiment, the shortwave infrared transform reflectance (STR) and normalized difference vegetation index (NDVI) of the area to be deployed are extracted from the band reflectance data of remote sensing satellites (such as Landsat and Sentinel). The calculation formulas are as follows:

[0078]

[0079]

[0080] In the formula, , , These represent the reflectance of remote sensing data in the shortwave infrared, red, and near-infrared bands, respectively. 1 is the shortwave infrared transform reflectance, and NDVI is the normalized vegetation index.

[0081] S102: A pixel distribution map is constructed based on shortwave infrared transform reflectance and normalized vegetation index. Dry edges and wet edges are obtained based on the pixel distribution map. Soil moisture is inverted based on dry edges and wet edges to obtain the soil moisture content of the area to be deployed.

[0082] In this embodiment, the shortwave infrared transform reflectance is plotted. and Normalized Difference Vegetation Index Relationship diagram, - The spatial distribution forms a trapezoid, such as Figure 2 As shown, where, Points with high values ​​are wet edges. Points with low values ​​are dry edges.

[0083] According to the Kubelka-Munk two-way radiation model, we can obtain:

[0084]

[0085] In the formula, , These are the light absorption coefficient and scattering coefficient of the layer, respectively. The transformed reflectivity, Given shortwave infrared reflectance, and pixels with the same NDVI value representing a fixed vegetation cover ratio (FVC), then:

[0086]

[0087] In the formula, , These correspond to the light absorption coefficient and scattering coefficient of bare soil and water body, respectively, where θ is the soil moisture content in the root zone. It refers to the surface soil moisture content, therefore:

[0088]

[0089] In the shortwave infrared (SWIR) wavelength range (S water If ≈ 0, then:

[0090]

[0091] In the formula, The transformed reflectivity, The scattering coefficient of bare soil;

[0092] Then, the wet and dry edges corresponding to the pixel distribution map are calculated using linear regression. Based on the wet and dry edges, the soil moisture (Soil Moisture) is determined using the Optram thermo-optical trapezoidal model. The soil moisture content of the area to be deployed is obtained by inversion.

[0093] In one embodiment, dry and wet edges are obtained based on a pixel distribution map. Soil moisture inversion is performed based on the dry and wet edges to obtain the soil moisture content of the area to be deployed, including:

[0094] Based on shortwave infrared transform reflectance and normalized vegetation index, the pixel distribution relationship map is obtained through linear regression to obtain dry and wet edges;

[0095] Based on dry and wet edges, soil moisture is inverted using a thermo-optical trapezoidal model to obtain the soil moisture content of the area to be deployed.

[0096] In this embodiment, in different Moisture content at the value If both STR and λ have a linear relationship, then the wet edge corresponding to the pixel distribution map can be calculated using linear regression. and dry edge The calculation formulas are as follows:

[0097]

[0098]

[0099] In the formula, , The intercept of the equation for the linear regression of dry and wet edges is given. , Let be the slope of the equation for the linear regression of dry and wet edges.

[0100] Based on soil surface moisture content and short-wave infrared convertible reflectance ( The linear relationship between soil moisture content and soil moisture content It can be calculated using the following equation:

[0101]

[0102] In the formula, This represents the true shortwave infrared transform reflectance under a certain NDVI condition.

[0103] In summary, the thermo-optical trapezoidal model (OPTRAM) was used to invert soil moisture ( The formula for ) is:

[0104]

[0105] In the formula, , The intercept of the equation for the linear regression of dry and wet edges is given. , Let be the slope of the equation for the linear regression of dry and wet edges.

[0106] S103: Collect digital elevation data of the area to be deployed, obtain feature data of each pixel based on digital elevation data, soil texture and vegetation cover type data, determine the initial number of cluster centers, obtain the initial number, and obtain the likelihood function under given model parameters based on the feature data of each pixel. The feature data includes latitude, longitude, slope, aspect, altitude, potential relative radiation, soil texture and vegetation cover type.

[0107] In this embodiment, high-resolution 30m raster data at the watershed scale is collected, mainly including digital elevation data (DEM). Further, based on the DEM, soil texture, and vegetation cover data, eight watershed features of soil pixels are calculated: latitude, longitude, slope, aspect, elevation, and potential relative radiation. Simultaneously, soil texture and vegetation cover data are also collected. A Gaussian mixture model (GMM) is used to cluster the eight features of each soil pixel. First, the number of cluster centers is set. The range of the feature space observed by each sensor is represented by a multivariate normal distribution. The feature space of each sensor is (…). , ).

[0108] In one embodiment, the formula for calculating the slope is:

[0109]

[0110]

[0111]

[0112] In the formula, For pixels in Rate of change of elevation on the axis For pixels in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis;

[0113] The formula for calculating slope aspect is:

[0114]

[0115] In the formula, For pixels in Rate of change of elevation on the axis For pixels in Rate of change of elevation on the axis;

[0116] The formula for calculating potential relative radiation is:

[0117]

[0118] In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. It is the angle between the slope and the solar altitude angle. It is the angle between the slope direction and the solar azimuth.

[0119] In this embodiment, the slope S and aspect can be calculated based on high-resolution digital elevation data, and the formulas are as follows:

[0120]

[0121]

[0122]

[0123] In the formula, For pixels in Rate of change of elevation on the axis For pixels in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis;

[0124] The formula for calculating slope aspect is:

[0125]

[0126] In the formula, For pixels in Rate of change of elevation on the axis For pixels in Rate of change of elevation on the axis;

[0127] The formula for calculating potential relative radiation is:

[0128]

[0129] In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. It is the angle between the slope and the solar altitude angle. It is the angle between the slope direction and the solar azimuth.

[0130] In one embodiment, based on the feature data of each pixel, the likelihood function under given model parameters is obtained, including:

[0131] Based on the feature data of each pixel, the feature space range of each pixel is represented by a multivariate normal distribution, resulting in the multivariate normal distribution probability density function. The expression for the multivariate normal distribution probability density function is as follows:

[0132]

[0133] In the formula, For the expected value, Let covariance matrix be the variance matrix. For the feature data of each pixel;

[0134] The probability density of the Gaussian mixture model for each pixel is obtained based on the multivariate normal distribution probability density function of each pixel. The expression for the probability density of the Gaussian mixture model is as follows:

[0135]

[0136] In the formula, For the expected value, Let covariance matrix be the variance matrix. Let be the mixing coefficient of the m-th Gaussian component. For the feature data of each pixel, Let be the mean of the m-th Gaussian component. Let m be the covariance matrix of the m-th Gaussian component. Let be the probability density function of the Gaussian mixture model for each pixel;

[0137] Based on the probability density of the Gaussian mixture model for each pixel, the likelihood function is obtained for the given model parameters, where the likelihood function is:

[0138]

[0139] In the formula, To determine the first given model parameters The probability of a pixel For the first A multivariate normal distribution with Gaussian components For the first The mixing coefficient of the Gaussian components, For the first The mean of the Gaussian components, For the first The covariance matrix of Gaussian components, The number of pixels. is the number of classes in the Gaussian mixture model.

[0140] In this embodiment, the number of cluster centers is a hyperparameter set in the model, representing the number of watershed feature clusters and the number of sensors initially selected. Each sensor observes the feature space ( , The range of ) is represented by a multivariate normal distribution, and the expression for the probability density function of the multivariate normal distribution is:

[0141]

[0142] In the formula, For the expected value, Let covariance matrix be the variance matrix. Feature data for each pixel:

[0143] After obtaining the multivariate normal probability density function for each pixel, we obtain the probability density of the Gaussian mixture model for each pixel. The expression for the probability density of the Gaussian mixture model is:

[0144]

[0145] In the formula, For the expected value, Let covariance matrix be the variance matrix. Let be the mixing coefficient of the m-th Gaussian component. For the feature data of each pixel, Let be the mean of the m-th Gaussian component. Let m be the covariance matrix of the m-th Gaussian component. Let be the probability density function of the Gaussian mixture model for each pixel;

[0146] Then, based on the likelihood function, the probability of the occurrence of data from known sample points is calculated as follows:

[0147]

[0148] In the formula, To determine the first given model parameters The probability of a pixel For the first A multivariate normal distribution with Gaussian components For the first The mixing coefficient of the Gaussian components, For the first The mean of the Gaussian components, For the first The covariance matrix of Gaussian components, The number of pixels. is the number of classes in the Gaussian mixture model.

[0149] S104: Based on the feature data of each pixel, obtain the prior distribution under the given model parameters. Multiply the prior distribution and the likelihood function to obtain the posterior distribution under the given model parameters. Estimate the posterior distribution using numerical methods to obtain the estimated value. Determine the position of each cluster center based on the estimated value. Based on the position of each cluster center, use nearest neighbor search to obtain the pixels whose distance to each cluster center is less than a preset value. Use these pixels as representative samples. The given model parameters include the mixing coefficient, mean, and covariance matrix.

[0150] In this embodiment, given model parameters (such as...) are obtained based on the feature data of each pixel. , Given the data to be clustered, according to Bayes' theorem, the posterior distribution is the product of the likelihood function and the prior distribution, thus yielding the posterior distribution of the model parameters, i.e.:

[0151]

[0152] In the formula, For the expected value, Let covariance matrix be the variance matrix. These are the feature data for each pixel.

[0153] During the calculation process, numerical methods are used to estimate the posterior distribution and determine the cluster centers.

[0154] For each cluster center Using nearest neighbor search, find the point closest to the center among all data points. As a representative sample point, the calculation formula is:

[0155]

[0156]

[0157]

[0158] in, yes The distance to the query point q is calculated using the Euclidean distance metric, where D represents an 8-dimensional vector in this case. These are representative sample points.

[0159] S105: A Gaussian process regression model is used to model the covariance model. The feature data of each representative sample point are input into the covariance model for prediction to obtain the predicted soil moisture content of the representative sample points. The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. If the RMSE is still greater than the preset threshold, the process moves to the step of determining the initial number of cluster centers. The initial number of cluster centers is increased to obtain the updated initial number. Based on the updated initial number, the positions of each updated cluster center and the new RMSE are calculated until the new RMSE is less than the preset threshold. The corresponding updated initial number is then determined as the final number of cluster centers. The final number of cluster centers is determined as the number of sensors to be deployed, and the positions of the updated cluster centers are the positions of the sensors to be deployed.

[0160] In this embodiment, based on the soil moisture data of initially selected representative sampling points, the distribution of soil moisture (i.e., soil water content) across the entire watershed is estimated using a Gaussian process (GP). Specifically, a covariance-based model is established using the Gaussian process, employing known input-output relationships and training data from the sampling point locations. Predictions can obtain soil moisture and uncertainties at new locations. Watershed characteristic variables such as slope, aspect, elevation, and potential relative radiation (PRDR) can be used. , The dependent variable used in the estimation is:

[0161]

[0162]

[0163] In the formula, Soil moisture predicted by the covariance model, Gaussian process By a mean function and a covariance function Sure, The variance of all representative samples, The length scale for all representative sample points.

[0164] The error in the Gaussian process model is estimated by the root mean square error (RMSE) between the predicted soil moisture and the inverted soil moisture data of the area to be deployed. If the minimum RMSE is not found, the number of sensors is increased sequentially, and the process of S103-S104 is repeated to calculate the RMSE after increasing the number of representative sample points. The number of representative sample points corresponding to the minimum RMSE is the most reasonable number of sensors. At this time, the distribution location is determined by the latitude and longitude coordinates of the representative sample points through the nearest neighbor search, which is the best deployment location.

[0165] In one embodiment, calculating the root mean square error of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion includes:

[0166] The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained through soil moisture inversion is calculated. The formula for calculating the RMSE is as follows:

[0167]

[0168] In the formula, To represent the number of sample points, To represent the predicted soil moisture content of the sample points, Soil moisture content obtained from soil moisture inversion.

[0169] In this embodiment, the root mean square error of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is estimated. The formula for calculating the root mean square error is as follows:

[0170]

[0171] In the formula, To represent the number of sample points, To represent the predicted soil moisture content of the sample points, Soil moisture content obtained from soil moisture inversion.

[0172] The proposed method for determining the deployment strategy of a soil moisture sensor network utilizes soil moisture retrieved using an OPTRAM model based on remote sensing information. This eliminates the need for pre-deploying experimental stations in the watershed, saving significant manpower costs, and also avoids the need to measure soil moisture data over a period of time, thus saving time. Furthermore, it selects representative soil moisture deployment stations at the watershed scale by clustering based on multi-dimensional watershed-scale characteristic data. The method also incorporates soil moisture data retrieved using microwave remote sensing to verify the effectiveness of the machine learning model (Gaussian process regression), thereby evaluating the optimal number of sensors.

[0173] By employing a supervised machine learning model (Gaussian process regression) to simulate and verify representative samples using an unsupervised clustering algorithm (Gaussian mixture model), the accuracy of sensor location and quantity was improved.

[0174] The soil moisture sensor network deployment strategy determination device provided in this embodiment of the invention, such as... Figure 4 As shown, Figure 4 A block diagram is provided to determine the deployment strategy for a soil moisture sensor network, including:

[0175] The extraction module 401 is used to extract remote sensing data of the area to be deployed, and obtain the shortwave infrared transform reflectance and normalized vegetation index of the area to be deployed.

[0176] The inversion module 402 is used to construct a pixel distribution relationship map based on shortwave infrared transform reflectance and normalized vegetation index, obtain dry and wet edges based on the pixel distribution relationship map, and perform soil moisture inversion based on dry and wet edges to obtain the soil moisture content of the area to be deployed.

[0177] The first calculation module 403 is used to collect digital elevation data of the area to be deployed, obtain feature data of each pixel based on digital elevation data, soil texture and vegetation cover type data, determine the initial number of cluster centers, obtain the initial number, and obtain the likelihood function under given model parameters based on the feature data of each pixel. The feature data includes latitude, longitude, slope, aspect, altitude, potential relative radiation, soil texture and vegetation cover type.

[0178] The second calculation module 404 is used to obtain the prior distribution under given model parameters based on the feature data of each pixel, multiply the prior distribution and the likelihood function to obtain the posterior distribution under given model parameters, estimate the posterior distribution using numerical methods to obtain the estimated value, determine the position of each cluster center based on the estimated value, and use nearest neighbor search to obtain the pixels whose distance to each cluster center is less than a preset value based on the position of each cluster center, and use the pixels as representative sample points. The given model parameters include mixing coefficients, mean and covariance matrix.

[0179] The third calculation module 405 is used to model using a Gaussian process regression model to obtain a covariance model. The characteristic data of each representative sample point are input into the covariance model for prediction to obtain the predicted soil moisture content of the representative sample points. The root mean square error of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. If the root mean square error is still greater than a preset threshold, the process proceeds to the step of determining the initial number of cluster centers. The initial number of cluster centers is increased to obtain an updated initial number. Based on the updated initial number, the positions of each updated cluster center and the new root mean square error are calculated until the new root mean square error is less than the preset threshold. Then, the corresponding updated initial number is determined as the final number of cluster centers. The final number of cluster centers is determined as the number of sensors to be deployed, and the positions of the updated cluster centers are the positions of the sensors to be deployed.

[0180] In one embodiment, the inversion module 402 includes a linear regression calculation unit and a soil moisture inversion unit, wherein,

[0181] The linear regression calculation unit is used to obtain the dry and wet edges by calculating the pixel distribution relationship map based on shortwave infrared transform reflectance and normalized vegetation index.

[0182] The soil moisture inversion unit is used to invert soil moisture based on dry and wet edges using a thermo-optical trapezoidal model to obtain the soil moisture content of the area to be deployed.

[0183] In one embodiment, the formula for calculating the slope is:

[0184]

[0185]

[0186]

[0187] In the formula, For pixels in Rate of change of elevation on the axis For pixels in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis;

[0188] The formula for calculating slope aspect is:

[0189]

[0190] In the formula, For pixels in Rate of change of elevation on the axis For pixels in Rate of change of elevation on the axis;

[0191] The formula for calculating potential relative radiation is:

[0192]

[0193] In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. It is the angle between the slope and the solar altitude angle. It is the angle between the slope direction and the solar azimuth.

[0194] The specific implementation of the device for determining the deployment strategy of a soil moisture sensor network is basically the same as the specific embodiment of the method for determining the deployment strategy of a soil moisture sensor network described above, and will not be repeated here.

[0195] In one embodiment of this application, a computer device is provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above steps. The implementation principle and technical effects of the computer device provided in this embodiment are similar to those of the above method embodiments, and will not be repeated here.

[0196] In one embodiment of this application, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it performs the above steps; the implementation principle and technical effects of the computer-readable storage medium provided in this embodiment are similar to those of the above method embodiments, and will not be repeated here.

[0197] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0198] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for determining the deployment strategy of a soil moisture sensor network, characterized in that, include: Remote sensing data of the area to be deployed is extracted to obtain the shortwave infrared transform reflectance and normalized vegetation index of the area to be deployed. A pixel distribution map is constructed based on the shortwave infrared transform reflectance and the normalized vegetation index. Dry edges and wet edges are obtained based on the pixel distribution map. Soil moisture inversion is performed based on the dry edges and wet edges to obtain the soil moisture content of the area to be deployed. Collect digital elevation data of the area to be deployed, obtain feature data of each pixel based on the digital elevation data, soil texture and vegetation cover type data, determine the initial number of cluster centers, obtain the initial number, and obtain the likelihood function under given model parameters based on the feature data of each pixel. The feature data includes latitude, longitude, slope, aspect, altitude, potential relative radiation, soil texture and vegetation cover type. Based on the feature data of each pixel, a prior distribution under given model parameters is obtained. The prior distribution is multiplied by the likelihood function to obtain a posterior distribution under given model parameters. The posterior distribution is estimated using numerical methods to obtain an estimated value. The position of each cluster center is determined based on the estimated value. Based on the position of each cluster center, a nearest neighbor search is used to obtain pixels whose distance to each cluster center is less than a preset value. These pixels are used as representative samples. The given model parameters include mixing coefficients, mean, and covariance matrix. A Gaussian process regression model is used to model the covariance model. The feature data of each representative sample point are input into the covariance model for prediction to obtain the predicted soil moisture content of the representative sample point. The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. If the RMSE is still greater than a preset threshold, the process proceeds to the step of determining the initial number of cluster centers. The initial number of cluster centers is increased to obtain an updated initial number. Based on the updated initial number, the positions of each updated cluster center and the new RMSE are calculated until the new RMSE is less than the preset threshold. The corresponding updated initial number is then determined as the final number of cluster centers. The final number of cluster centers is determined as the number of sensors to be deployed, and the positions of the updated cluster centers are the locations of the sensors to be deployed.

2. The method for determining the deployment strategy of a soil moisture sensor network as described in claim 1, characterized in that, The process of obtaining dry and wet edges based on the pixel distribution map, and performing soil moisture inversion based on the dry and wet edges to obtain the soil moisture content of the area to be deployed includes: Based on shortwave infrared transform reflectance and normalized vegetation index, the pixel distribution relationship map is obtained by linear regression to obtain dry and wet edges. Based on the dry and wet edges, the soil moisture is inverted using a thermo-optical trapezoidal model to obtain the soil moisture content of the area to be deployed.

3. The method for determining the deployment strategy of a soil moisture sensor network as described in claim 1, characterized in that, The formula for calculating the slope is: In the formula, For the pixel point in Rate of change of elevation on the axis For the pixel point in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis; The formula for calculating the slope aspect is: In the formula, For the pixel point in Rate of change of elevation on the axis For the pixel point in Rate of change of elevation on the axis; The formula for calculating the potential relative radiation is as follows: In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. It is the angle between the slope and the solar altitude angle. It is the angle between the slope direction and the solar azimuth.

4. The method for determining the deployment strategy of a soil moisture sensor network as described in claim 1, characterized in that, The process of obtaining the likelihood function based on the feature data of each pixel under given model parameters includes: Based on the feature data of each pixel, the feature space range of each pixel is represented by a multivariate normal distribution, resulting in a multivariate normal distribution probability density function. The expression for the multivariate normal distribution probability density function is as follows: In the formula, As the expected value, Let covariance matrix be the variance matrix. For each of the aforementioned pixels; The probability density of the Gaussian mixture model for each pixel is obtained based on the multivariate normal distribution probability density function of each pixel, wherein the expression for the probability density of the Gaussian mixture model is: In the formula, As the expected value, Let covariance matrix be the variance matrix. Let be the mixing coefficient of the m-th Gaussian component. For the feature data of each of the aforementioned pixels, Let be the mean of the m-th Gaussian component. Let m be the covariance matrix of the m-th Gaussian component. Let be the probability density function of the Gaussian mixture model for each pixel; Based on the probability density of the Gaussian mixture model for each pixel, the likelihood function under given model parameters is obtained, wherein the likelihood function is: In the formula, To determine the first given model parameters The probability of a pixel For the first A multivariate normal distribution with Gaussian components For the first The mixing coefficient of the Gaussian components, For the first The mean of the Gaussian components, For the first The covariance matrix of Gaussian components, The number of pixels. is the number of classes in the Gaussian mixture model.

5. The method for determining the deployment strategy of a soil moisture sensor network as described in claim 1, characterized in that, The calculation of the root mean square error of the soil moisture content of the representative sample points and the soil moisture content distribution data includes: The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. The formula for calculating the RMSE is as follows: In the formula, To represent the number of sample points, To represent the predicted soil moisture content of the sample points, Soil moisture content obtained from soil moisture inversion.

6. A device for determining the deployment strategy of a soil moisture sensor network, characterized in that, include: The extraction module is used to extract remote sensing data of the area to be deployed, and obtain the shortwave infrared transform reflectance and normalized vegetation index of the area to be deployed. The inversion module is used to construct a pixel distribution relationship map based on the shortwave infrared transform reflectance and the normalized vegetation index, obtain dry edges and wet edges based on the pixel distribution relationship map, and perform soil moisture inversion based on the dry edges and wet edges to obtain the soil moisture content of the area to be deployed. The first calculation module is used to collect digital elevation data of the area to be deployed, obtain feature data of each pixel based on the digital elevation data, soil texture and vegetation cover type, determine the initial number of cluster centers, obtain the initial number, and obtain the likelihood function under given model parameters based on the feature data of each pixel. The feature data includes latitude, longitude, slope, aspect, altitude, potential relative radiation, soil texture and vegetation cover type. The second calculation module is used to obtain the prior distribution under given model parameters based on the feature data of each pixel, multiply the prior distribution and the likelihood function to obtain the posterior distribution under given model parameters, estimate the posterior distribution using numerical methods to obtain an estimated value, determine the position of each cluster center based on the estimated value, and use nearest neighbor search to obtain pixels whose distance to each cluster center is less than a preset value based on the position of each cluster center, and use the pixels as representative samples. The given model parameters include mixing coefficients, mean, and covariance matrix. The third calculation module is used to model using a Gaussian process regression model to obtain a covariance model. The characteristic data of each representative sample point are input into the covariance model for prediction to obtain the predicted soil moisture content of the representative sample point. The root mean square error (RMSE) of the predicted soil moisture content and the soil moisture content obtained by soil moisture inversion is calculated. If the RMSE is still greater than a preset threshold, the module proceeds to the step of determining the initial number of cluster centers. The initial number of cluster centers is increased to obtain an updated initial number. Based on the updated initial number, the positions of each updated cluster center and the new RMSE are calculated until the new RMSE is less than the preset threshold. The corresponding updated initial number is then determined as the final number of cluster centers. The final number of cluster centers is determined as the number of sensors to be deployed, and the positions of the updated cluster centers are the locations of the sensors to be deployed.

7. The soil moisture sensor network deployment strategy determination device as described in claim 6, characterized in that, The inversion module includes a linear regression calculation unit and a soil moisture inversion unit, wherein, The linear regression calculation unit is used to obtain the dry and wet edges by linear regression calculation based on shortwave infrared transform reflectance and normalized vegetation index. The soil moisture inversion unit is used to invert soil moisture based on the dry and wet edges using a thermo-optical trapezoidal model to obtain the soil moisture content of the area to be deployed.

8. The soil moisture sensor network deployment strategy determination device as described in claim 6, characterized in that, The formula for calculating slope is: In the formula, For the pixel point in Rate of change of elevation on the axis For the pixel point in The rate of change of elevation on the axis, where Z(x,y) is the elevation value, representing the elevation at pixel (x,y). , The digital elevation data are respectively in On the axis and Vertical resolution on the axis; The formula for calculating the slope aspect is: In the formula, For the pixel point in Rate of change of elevation on the axis For the pixel point in Rate of change of elevation on the axis; The formula for calculating the potential relative radiation is as follows: In the formula, The solar constant, Solar altitude angle, representing the angle between the sun and the horizon. It is the angle between the slope and the solar altitude angle. It is the angle between the slope direction and the solar azimuth.

9. A computer device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the method for determining a soil moisture sensor network deployment strategy as described in any one of claims 1 to 5.

10. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for determining the deployment strategy of a soil moisture sensor network as described in any one of claims 1 to 5.

Images